FIRST STEPS TOWARDS SANSKRIT

First Steps Towards Sanskrit: Language, Linguistics and Culture is an

accessible first introduction to this ancient Indian language.
Complete beginners are introduced to the language from scratch. Key
terms are explained clearly and there is an extensive glossary to assist the
reader who is unfamiliar with the terminology of language learning. By the
end of the book, learners will have grasped the basics of the language and be
prepared to engage readily in an introductory college or university course
or through private study. The addition of cultural, linguistic and historical
notes will appeal to learners with diverse interests, ranging from religious
studies and philosophy to yoga and comparative or historical linguistics.
The book includes references to classical and modern European
languages. Parallels are also drawn with Indic languages where these are
relevant, particularly as concerns the writing system. No knowledge of any
language other than English is, however, presupposed. This book is ideal
for both self-­study and in-­class use as a primer or core text for pre-­sessional
courses.

Anil K. Biltoo is an Honorary Fellow of the Chartered Institute of Linguists

and has taught Sanskrit for many years at the University of London. He is
currently teaching at King’s College London.
FIRST STEPS TOWARDS
SANSKRIT
Language, Linguistics
and Culture

Anil K. Biltoo
First published 2022
by Routledge
2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN
and by Routledge
605 Third Avenue, New York, NY 10158
Routledge is an imprint of the Taylor & Francis Group,
an informa business
© 2022 Anil K. Biltoo
The right of Anil K. Biltoo to be identified as author of this work
has been asserted in accordance with sections 77 and 78 of the
Copyright, Designs and Patents Act 1988.
All rights reserved. No part of this book may be reprinted
or reproduced or utilised in any form or by any electronic,
mechanical, or other means, now known or hereafter invented,
including photocopying and recording, or in any information
storage or retrieval system, without permission in writing from the
publishers.
Trademark notice: Product or corporate names may be trademarks
or registered trademarks, and are used only for identification and
explanation without intent to infringe.
British Library Cataloguing-­in-­Publication Data
A catalogue record for this book is available from the British
Library
Library of Congress Cataloging-­in-­Publication Data
A catalog record for this book has been requested
ISBN: 978-­0-­367-­34386-­6 (hbk)
ISBN: 978-­0-­367-­34385-­9 (pbk)
ISBN: 978-­0-­429-­32543-­4 (ebk)
DOI: 10.4324/9780429325434
Typeset in Times New Roman
by Apex CoVantage, LLC
CONTENTS

Acknowledgementsix

Introduction1
Introductory remarks 1
How to use this book 4

1 The Sanskrit language 6

1.1 Linguistic background 6
1.2 The ‘discovery’ of Sanskrit 9
1.3 Cognates 11
1.4 The origin of Sanskrit 13
1.5 Indus Valley connections 15
1.6 Myths and realities 17

2 The International Alphabet of Sanskrit Transliteration 20

2.1 Preliminaries 20
2.2 Analysis of a word in the IAST 21
2.3 Additional remarks 23

3 The sound system 26

3.1 Vowels and diphthongs 26
3.1.1 Preliminaries 26
vi Contents

3.1.2 Pronunciation 27
3.1.3 Overview 30
3.2 Consonants 31
3.2.1 Preliminaries 31
3.2.2 Pronunciation 33
3.2.2.1 Velars 33
3.2.2.2 Palatals 36
3.2.2.3 Cerebrals 37
3.2.2.4 Dentals 38
3.2.2.5 Labials 40
3.2.3 Overview 40
3.3 Semivowels 41
3.3.1 Preliminaries 41
3.3.2 Pronunciation 41
3.3.3 Overview 42
3.4 Sibilants 42
3.4.1 Preliminaries 42
3.4.2 Pronunciation 43
3.4.3 Overview 43
3.5 Additional sounds 44
3.5.1 Anusvāra 44
3.5.2 Visarga 46
3.5.3 The inherent vowel 48
3.5.4 Virāma 49
3.6 Exercise: true or false? 50

4 The writing system 52

4.1 Origin of the Devanāgarī 52
4.2 The Devanāgarī investigated 53
4.2.1 Preliminaries 53
4.2.2 Vowels 55
4.2.3 Consonants 59
4.2.4 Semivowels and sibilants 62
4.3 Conjuncts and conjunct formation 64
4.3.1 Preliminaries 64
4.3.2 Removal of the vertical stroke 65
4.3.3 Stacking 66
4.3.4 Conjuncts involving r 67
4.3.5 Unique forms 69
 Contents vii

4.3.6 Multiple conjuncts 71

4.4 Variant forms 74
4.4.1 Orthographic variants 74
4.4.2 Font variation 75
4.4.3 Sample text in various fonts 76
4.4.4 Other writing systems 79
4.5 Writing the Devanāgarī 81
4.6 Exercises 83
4.6.1 Identify the consonant 83
4.6.2 Transliteration 85

5 The Sanskrit word 87

5.1 Preliminaries 87
5.2 Analysis of words 88
5.3 Approaching Sanskrit grammar 98

6 The Sanskrit noun: Declension 100

6.1 Preliminaries 100
6.2 Cases and case functions 102
6.3 Sample paradigms: short -­a stems 109
6.4 More paradigms: long -­ā and long -­ī stems 110
6.5 The demonstrative pronoun tat 112
6.6 Personal pronouns 116
6.7 Exercises 120
6.7.1 Questions on declension 120
6.7.2 Sentences for translation 121

7 The Sanskrit verb: Conjugation 124

7.1 Preliminaries 124
7.2 Root, stem and vowel mutation 126
7.3 Thematic verbs 128
7.4 The imperfect 131
7.5 The infinitive 133
7.6 The gerund 134
7.7 Exercise: sentences for translation 136

8 Sound changes: Sandhi 140

8.1 Preliminaries 140
8.2 Visarga sandhi 141
viii Contents

8.3 Consonant sandhi 145

8.4 Vowel sandhi 147
8.5 Cerebralization 148
8.6 Writing conventions 151
8.7 Exercise: reading a mantra 154

9 From sentence to text 156

9.1 Preliminaries 156
9.2 Reading Sanskrit 157
9.3 Selected readings 159
9.3.1 Reading One 159
9.3.2 Reading Two 163
9.3.3 Reading Three 165
9.3.4 Reading Four 167

10 Texts for the study of Sanskrit 170

Glossary176
Index and list of verb roots 190
ACKNOWLEDGEMENTS

The present work is dedicated to the many hundreds of Sanskrit students

and to the University of London staff whom I had the pleasure to teach
whilst at the School of Oriental and African Studies. Their patience and
dedication to the study of Sanskrit went from beyond mere curiosity to a
search to find meaning in a challenging but beguiling language. My former
students are too many to name. Thanks are due, however, to Celeste Cronin,
whose humour and diligence were a source of encouragement, not only in
the teaching of the language but in the writing of this book.
Thanks are due, also, to my elder sister Bi and her husband David for the
gift of food and ‘fur therapy’ (a Great Dane and a French Bulldog) when
such therapy was most needed. Lockdown in the time of a pandemic would
have been less endurable had it not been for both food and fur. I am indebted
to Rosie McEwan, Editorial Assistant with Taylor & Francis, for her under-
standing when it proved necessary to extend the submission date for the
manuscript, following a year that would not be wished on anyone and where
not even food and fur could revive the Sanskrit muse.
Finally, I would like to acknowledge the reader, for having demonstrated
such extraordinarily good taste in getting hold of an introductory text on
Sanskrit. It is heartening to see that there is still interest in Sanskrit and
that the modern world has not condemned it to obsolescence. Teaching stu-
dents for more than a decade has informed me as to what an introduction
x Acknowledgements

to Sanskrit should look like, what it ought to contain and what might tact-
fully be omitted for the sake of clarity. Any mistakes are mine alone and
I strongly welcome all comments and suggestions.
Anil K. Biltoo
[email protected]
Hampton Court
April 2021
INTRODUCTION

Introductory remarks
Sanskrit is a captivating language. It has long been the treasure of lin-
guists for whom it represents a linguistic fossil, like a fly trapped in amber,
reflecting a form of speech which existed before the Common Era and has
remained unchanged and unchangeable. For an untold number of others,
Hindu, Buddhist and Jain, it lies at the core of their religious tradition, since
the literatures of these religions find expression in Sanskrit or in one of its
sister languages (the Prakrits) of which the best known is Pāḷi, itself the
language of a large and important body of texts.
There are many other people, neither linguists nor Hindu, Buddhist or
Jain, for whom a knowledge of Sanskrit is something worth acquiring.
Students of ancient languages and cultures are drawn to Sanskrit, many
of them with an ability to read Greek or Latin (or indeed both). There are
also the students of comparative literature for whom Sanskrit offers endless
opportunities to explore the human condition in any number of texts rang-
ing from remote antiquity (starting with the Ṛgveda, c. 1500–1200 bce) to
the later Upaniśads and the twin epics of India, the Mahābhārata and the
Rāmāyaṇa.
For practitioners of yoga, Sanskrit represents the language in which their
philosophical practice is explored. It is the language which gives voice to
the principles of yoga – the thing which literately brings the mind and body
together (from the verbal root yuj: to unite, conjoin). The various postures

DOI: 10.4324/9780429325434-1
2 Introduction

of yoga, the āsanas, all have a Sanskrit name, highly illustrative in terms
of how they are formed. The suptavīrāsana posture, for example, is easily
comprehensible with only a small amount of Sanskrit at one’s disposal. It is
the posture in which one lies like a sleeping warrior: supta (sleeping), vīra
(warrior), āsana (posture).
Even if one is not a linguist, or a Hindu, Buddhist or Jain and has neither
an interest in classical studies and comparative literature or an inclination
to practise yoga, Sanskrit still has something to offer such an individual.
Sanskrit requires concentration and a disciplined mind. It is methodical,
rigorous, complex. To study Sanskrit is to immerse oneself in a mass of
verbal roots, an enormous working vocabulary and a writing system much
more principled in its design than the Phoenician alphabet from which the
Greek, Cyrillic and Roman alphabets evolved. For all who come to the
study of Sanskrit, for whatever reason, a rich and rewarding experience
awaits. Svāgataṃ (Welcome).
The present work is intended as an initial guide to Sanskrit, aimed at
those who wish to explore the language, its linguistic affiliations, its origin,
writing system and structure. The language is presented in terms of its cul-
tural significance as well as its structure. For those contemplating further
study, the present work is ideal. It does not in any way presume to replace
textbooks currently available, of which there are a fair few. Suggestions
as to further reading (Chapter 10) will mention a number of these along
with publications that are good to consult on various topics. It is important
for the learner to be prepared to read widely around the language. San-
skrit, divorced from any context and studied simply as a language system,
may have yielded its treasures to philologists in the past, whose chief con-
cern was to the investigate the language much as a laboratory researcher
might do with the contents of a Petri dish. For the contemporary learner,
this approach is unlikely to be the favoured one. Information about Sanskrit
literature and ancient Indian history is abundant, and the Internet has made
much of this information accessible.
The present work aims to introduce the reader to Sanskrit in gentle, man-
ageable steps which discuss the context in which the study of the language
is relevant (Chapter 1). It includes an exploration of a transliteration system
which allows the reader, from the outset, to tackle the sounds of Sanskrit
without first having to master the writing system associated with it (Chap-
ter 2). The sound system is then introduced in sufficient detail to allow
the reader to gain confidence that the sounds being uttered are accurate
(Chapter 3). Thereafter, the writing system – the Devanāgarī – is intro-
duced, allowing the reader to connect the sounds of Sanskrit with their
 Introduction 3

visual representation (Chapter 4). In the subsequent chapters, a knowledge

of both the pronunciation and writing system of Sanskrit are used to ease the
learner into the reading of Sanskrit words (Chapter 5) and, subsequently, to
investigate the structure of the language (Chapters 6–7) and sound changes
(Chapter 8). In Chapter 9, the reader is presented with verses in Sanskrit.
These are explained in detail, permitting the reader to see how Sanskrit
expresses ideas.
Use is made of a transliteration system throughout, so the reader can keep
checking the transliteration against the Devanāgarī as the eye slowly and
naturally adjusts to the latter. Where the transliteration is used, it appears in
bold typeface. If a word is bolded in the main body of the text (section or
table headings aside), it represents Sanskrit, hence the fact that the words
Upaniśads and āsanas were indicated in bold apart from the final -­s, which
does not represent the Sanskrit plural. A Sanskrit term, once used, is not
repeatedly bolded, since that would unnecessarily distract the eye. By con-
trast, vocabulary items and examples from Sanskrit are bolded throughout
(apart from the index). Certain words are so frequently encountered in texts
related to Sanskrit that their everyday spelling in English is used. The word
‘Sanskrit’ is the obvious example; another is ‘Prakrit’. It seems undesirable
to refer to these as Saṃskr̥ta and Prākr̥ta in the body of the text. The same
practice is applied to the modern languages of India.
Words and phrases from all languages other than Sanskrit, when used to
illustrate examples, are in italics, as are translations or the titles of publica-
tions. Italics also indicate individual sounds (e.g. sh-­sound) where English
spelling is used to identify them. The alternative to this would necessar-
ily have involved the reader learning the International Phonetic Alphabet –
which is hardly to be expected of those coming from disciplines other than
linguistics. The Devanāgarī appears throughout in a larger font, for the sake
of legibility, addressing a frustration frequently voiced by learners of San-
skrit: ‘I can’t see the letters clearly. They’re too small’. Underlining is used
to emphasize or highlight a word or part of a word. Finally, a preceding
asterisk indicates unattested or non-­licit forms. This is a convention in lin-
guistics and a useful one to employ.
No knowledge of any language other than English is presupposed. Whilst
there are frequent examples of words from other languages, as and when
they serve to illustrate particular points, it is not assumed that the reader
will have a speaking or reading ability in Latin (or Greek, or Lithuanian, or
any other language). Pronunciation tips refer to ‘English’, pure and simple,
in preference to the more customary ‘Received Pronunciation’ (RP), occa-
sionally still defined as ‘BBC English’. This is because terms ‘RP/BBC
4 Introduction

English’ tend to bring to mind a particular pronunciation which has moved

away from everyday norms and represents, certainly for speakers of British
English, an unnatural and archaic pronunciation.
The present work does not rely on recordings for the simple reason that
there is no fixed standard for Sanskrit pronunciation. In any event, those
who promote Sanskrit as a spoken language very frequently bring in fea-
tures from their mother tongues. Sanskrit classifies its sound system in a
way which is rigorous enough to allow the reader to grasp the pronuncia-
tion of any given phoneme. The reader ought not to feel that the ability to
mimic spoken Sanskrit is necessary. That is absolutely not the case. The
voice which the reader possesses is more than sufficient.

How to use this book

The present work is intended to be read in the order in which the chapters
appear. The reader who has some understanding of Sanskrit or can make
headway with the Devanāgarī is urged not to skip through chapters. These
follow in a logical sequence. Many of the points made are followed up in
later chapters. There is a clean division as regards the aspect of the language
under investigation. First comes the sound system, then the writing system,
then the formal grammar. This is a natural way to proceed, given that the
utterance precedes writing, both in terms of historical language develop-
ment and in the way in which people acquire their first language. The reader
should allow for a pause for consolidation between chapters, ensuring that
the information has been absorbed and retained before reading further.
A good place at which to take a longer pause to consolidate fully is
the end of Chapter 5. This represents the halfway point in the First Steps
Towards Sanskrit. It is the first of two steps. On reaching that point, the
reader is strongly recommended to comb through Chapter 1 to Chapter 5,
picking up any points that were missed in the first reading. The aim of the
author is to make the path towards a knowledge of Sanskrit as effortless as
possible. The style is not overly dry and there is no assumption made that
the reader would have to remedy any lacunae in understanding by going to
other sources. That said, the chapters beyond Chapter 2 are information-
ally rich; and new information merits a rereading. The second step (Chap-
ter 6 through to Chapter 10) requires the reader to have taken the first step
securely.
For the reader who has the time to explore Sanskrit resources on the
Internet, these are an excellent source of information on such things as
font variation within the Devanāgarī (This is something which the reader
 Introduction 5

is urged to investigate, after completing Chapter 4 or Chapter 5). It is not

difficult to find examples of mantras or verses in Sanskrit, and accustoming
the eye to fonts that are different to the one used in the present work will
stand the learner in good stead. As illustrated in 4.4.2 and 4.4.3, certain
fonts are easy to read; others, less so. Importantly, the Internet also allows
the reader to hear a range of different voices reciting Sanskrit and to see the
extent to which the cursory transliteration of Sanskrit is to be avoided. Just
as all roads lead to Rāma, excusing the dreadful pun, every exploration of
Sanskrit leads to greater familiarity with it.
1
THE SANSKRIT LANGUAGE

1.1 Linguistic background

Sanskrit is an Indo-­European language. This means that it belongs to the
same family of languages as do, for example, English, French, Spanish, Ital-
ian, German, Russian, Persian and the languages of northern India. The list
of languages classified as Indo-­European is long, for if German is Indo-­
European, so too are Dutch, Danish, Norwegian and Swedish – and all the
other languages that are also categorized as ‘Germanic’. From the outset,
there is a need for clarity as regards various terms. When linguists talk of a
‘language family’, they refer to the concept that languages are related. The
relationship is predicated around the concept that related languages have a
common ancestor. So far, so good; but a language can have several ances-
tors: some are recent and are well documented, others go back to a point in
remote history and have left behind no written records. If one takes Italian as
a case study, the following propositions can be made and are all equally true:

‘Italian is an Indo-­European language’.

‘Italian is descended from Latin’.
‘Italian is a Romance (or Italic) language’.

It does not, however, follow that English is an Indo-­European language

and must, as a result, also be descended from Latin. Here, the concept of a
language branch is important.

DOI: 10.4324/9780429325434-2
 The Sanskrit language 7

When one investigates the relationship between languages within the

Indo-­European family, one rapidly encounters the term ‘branch’. It is often
the cause of confusion, in that many sources conflate family with branch
whereas these concepts are quite distinct. Modern Italian is descended from
Latin, as has been asserted. Both Italian and Latin, as distinct languages,
belong to the Romance branch – not the Romance family. The relationship
between these two languages can easily be shown, since there is ample his-
torical evidence attesting to the changes that took place from early Latin,
through to Latin in its classical form (the form taught in schools), followed
by the Vulgar Latin of later centuries, when Rome as an imperial power
had collapsed. Thereafter came the formation of the early modern dialects
of the Italian peninsula which gave rise to the modern dialects, many of
which still exist (Ligurian, Venetian, Sardinian, Sicilian, etc.). All the dia-
lects identifiable with Latin, including Latin itself, are Romance languages
(if defined according to a branch) and Indo-­European (if defined according
to the language family).
The Indo-­European language family can be divided into one of a dozen
branches, of which Romance is only one. Going roughly from west to east
(and excluding the Americas), these are as follows:

Branch Representative languages (not exhaustive)

Germanic Icelandic, Norwegian, German, English

Celtic Irish, Scots Gaelic, Welsh, Cornish, Breton
Romance (Italic) Portuguese, Spanish, French, Italian, Romanian
Slavonic (Slavic) Slovenian, Czech, Polish, Russian, Bulgarian
Baltic (Balto-­Slavic) Lithuanian, Latvian
Illyrian Albanian
Greek Greek
Thracian Thracian (Extinct)
Anatolian Hittite, Luwian, Palaic, Lydian (All extinct)
Armenian Armenian
Indo-­Iranian Persian, Pashto, Panjabi, Hindi, Bengali, Sanskrit
Tocharian Tocharian (Extinct)

The precise number of branches within Indo-­European is not subject to

universal agreement, since there are languages of which little is known and
on which archaeology has yet to shed light (Often, the evidence for these
languages takes the form of words and scraps of data preserved in other
languages). Within the branches themselves, there is also disagreement as
to distinguishing between language and dialect. It lies outside of the scope
8 The Sanskrit language

of the present work to investigate the arguments surrounding the issue of

language versus dialect. Suffice it to say that the issue is often far from
clear, even with present-­day linguistic varieties. Despite the fact, for exam-
ple, that there is a high degree of intelligibility between Norwegian, Danish
and Swedish, these languages are not referred to as dialects. The issue, ulti-
mately, is a political one. Norwegian, Danish and Swedish are recognized
as languages of three separate nation states, irrespective of the fact that they
show greater linguistic similarities with each other than is the case with
many of the so-­called ‘Hindi dialects’. The last word on this matter is best
provided by Max Weinreich, a linguist who devoted his life to the study
of the Yiddish language. According to Weinreich, “a language is a dialect
with an army and a navy”.
Sanskrit is an Indo-­European language belonging to the Indo-­Iranian
branch. As the table on the previous page indicated, the Indo-­Iranian branch
(which may further be split into two sub-­branches: Iranian and Indic/Aryan)
includes not only languages of South Asia but those spoken on the Iranian
plateau and in Afghanistan. This will be important to remember when inves-
tigating the origin of Sanskrit (1.4). This branch contains a great number
of modern languages which, taken together, are spoken by over a thousand
million people, from eastern Turkey (Kurdish) to the China/Myanmar bor-
der with India (Assamese). No other branch of Indo-­European has as many
speakers except for Germanic, courtesy of English in its role as the global
language.
The Indo-­Iranian branch has left important literary records from before
the Common Era, with Avestan, Old Persian, Middle Persian, Bactrian and
Sogdian on the Iranian side, and the Prakrits on the Indic (or Aryan) side.
The oldest representatives of this branch are Avestan and Vedic Sanskrit,
which served as the liturgical languages of two religions: Zoroastrianism
and Hinduism, respectively. Whilst knowledge of Avestan has been kept
alive by priests in a dwindling community of Zoroastrians in Iran and
amongst the Parsee community, largely resident in the Indian state of Guja-
rat (and also in diaspora), Vedic Sanskrit is chanted daily by many hundreds
of millions of Hindus.
One frequently hears and reads the assertion that the modern languages
of India are all descended from Sanskrit, but this is simply not the case.
Sanskrit was reserved for cultural activities and was not the language of
everyday speech. The latter function was served by the Prakrits (language
forms literally fashioned by nature). Three major Prakrits can be identi-
fied, descended from an Old Indic language which, for want of any other
contender, is stated as Vedic Sanskrit. Apabhraṃśa refers to a transitional
 The Sanskrit language 9

phase between the Prakrits and the older forms of the modern languages.
Sanskrit had no linear descendants. She was the spinster sister of the
Prakrits, with her mind set on higher things.

Vedic Sanskrit
↓ ↓
↓

↓
Sanskrit Mahārāṣṭrī Prakrit Śaurasenī Prakrit Māgadhī Prakrit
↓ ↓ ↓ ↓
No descendants Apabhraṃśa Apabhraṃśa Apabhraṃśa
↓ ↓ ↓
Marathi Hindi, Gujarati Oriya, Bengali

1.2 The ‘discovery’ of Sanskrit

Stating that Sanskrit is an Indo-­European language leads to the obvious
question as to the evidence supporting such a statement. Often, languages
are related if speakers of one can understand speakers of another without
having formally acquired each other’s language. It was noted that Norwe-
gian, Danish and Swedish exhibited just this type of mutual intelligibility.
So too, speakers of Spanish and Italian claim to be able to understand each
other without too much difficulty. The same is, however, not often heard
as between Spanish and French, or Italian and French. In India, something
similar occurs. Speakers of Urdu and Hindi have little difficulty understand-
ing each other (as long as the language register is of the everyday variety as
opposed to technical). That seems analogous to the Scandinavian situation.
As for speakers of Panjabi and Hindi, or Gujarati and Hindi, there can often
be a high degree of intelligibility, although this is not guaranteed (This is
more indicative of the situation as between speakers of Spanish and Italian).
Situations such as these would tend to indicate that speakers are able to
understand a great deal of the core vocabulary and grammar of certain other
languages. This is a good starting point in an investigation of Sanskrit’s cre-
dentials as regards membership of the Indo-­European language family; but
first, it is necessary to look at how Sanskrit was ‘discovered’ by Europeans,
since this goes to the heart of the matter.
Although not the first European to identify similarities between Sanskrit
and some of the languages of Europe, the accolade for first having posited
a substantive linguistic connection goes to Sir William Jones, a judge serv-
ing in Calcutta in the 1780s. Jones’s interest in the languages and cultures
of India was such that he established the Asiatick Society – ultimately to
become the Royal Asiatic Society of Bengal – in 1784. The aims of the
10 The Sanskrit language

Society were to explore a broad range of subjects relating to Asia, but it is

clear, from the number of manuscripts it collected, that the Society placed
great importance on the exploration of the languages and literatures of
India. In 1786, Jones gave a talk to the Society which has been hailed by
linguists as signifying the starting point of the systematic study of language
relationships which led, in the first instance, to comparative linguistics and
the evolution of Indo-­European studies. It is not out of the question to sug-
gest, therefore, that Sanskrit is responsible for the creation of linguistics as
a recognized discipline.
The talk that Jones gave in 1786 contains what is possibly the longest
sentence in the English language and certainly one of the most momentous
in terms of the impact it had on the study of Sanskrit. The talk in question
is worth reading in its entirety, no less for its detail than for the fact that
Jones displayed an extraordinary insight in having hypothesized connec-
tions between language that were subsequently proved to be correct. Jones
asserted that Sanskrit showed too many similarities with Greek and Latin for
such similarities to be coincidental. From there, he suggested that not only
did Sanskrit, Greek and Latin appear to be related but, so too, “the Gothic
and the Celtic [. . .] and the old Persian” (The term ‘Gothic’ would eventu-
ally be replaced by ‘Teutonic’ then ‘Germanic’ – the latter corresponding to
current usage – whilst the term ‘Celtic’ has endured). As for Jones’s instinct
regarding “the old Persian” being related to Sanskrit, that too was proved
correct. Strangely, however, the connection was only established once the
cuneiform writing system used for Old Persian monumental inscriptions
had been deciphered. That did not take place until the following century.
Sanskrit had a significant impact on the intellectual thought of Europe
in the period following the Age of Enlightenment, which culminated in the
French Revolution. The term ‘Romanticism’ is often applied to this period,
although one may suggest that there was nothing particularly romantic
about the rise of the nation state and a period of expansionism which saw
the British and French vying for hegemony within Europe as well as further
afield. European scholarship as regards Sanskrit was initially dominated
by the British, who had access to source material from India. Along with
Sir William Jones, mention should be made of Charles Wilkins (who had
helped Jones set up the Asiatick Society) and who was himself a scholar
of Sanskrit. Wilkins produced a translation of the Bhagavadgītā in 1785,
with Jones’s translation of a work by the foremost dramatist in Sanskrit
(Kālidāsa) following shortly thereafter: Sacontalá; Or, The Fatal Ring: An
Indian Drama by Cálidás (1789). Then followed the Sanskrit Grammar of
Henry Thomas Colebrooke (1805). Sanskrit captured the imagination of
 The Sanskrit language 11

poets such as Johann Wolfgang von Goethe, but it was in Sanskrit philology
that German scholarship excelled. Franz Bopp published his Detailed Sys-
tem of the Sanskrit Language in 1827 and the groundbreaking Comparative
Grammar of the Sanskrit, Zend, Greek, Latin, Lithuanian, Gothic, German,
and S(c)lavonic Languages (1833–1852). The latter has remained a key
resource in comparative linguistics and Indo-­European studies.

1.3 Cognates
The moment that speech communities are in contact, languages are subject
to borrowing. Often, borrowed items are easy to spot. The majority of Eng-
lish speakers would doubtless identify pizza, kebab, tikka masala, sushi and
coffee as borrowings, even if they might be unclear as to the actual origin
of the words in question (Coffee, for example, might be identified as having
come from the Italian caffè or even the French café but the word appears
to have entered English through the Dutch koffie, borrowed from the Turk-
ish kahve which, in turn, was borrowed from the Arabic qahwa). Borrow-
ings are not always easily identifiable as such to speakers of a language.
One would have to think carefully about such words as vest, jacket, over-
coat, shoes, since they are everyday items with nothing seemingly exotic
or imported about them (Overcoat and shoes are native English words; vest
and jacket are not).
Cognates are different from borrowings in that they refer to words that
exist in one language, are attested in another and are clearly related in
some way. Here, borrowing is not the reason for the existence of similarity.
The reason is that the languages in question both possess the word, having
inherited it from their ancestral form. Sanskrit contains many words that
are cognate with words in other Indo-­European languages – and not just
those spoken in India. Cognates in English and the Indic languages pre-
cede any direct contact or cultural influence between speakers of English
and speakers of Indic languages. To use Sir William Jones’s words, this
indicates that the words in question must have “sprung from some com-
mon source, which, perhaps, no longer exists”. Jones had roots of verbs
and forms of grammar in mind, but the same argument applies to nouns:
English mother is the Sanskrit mātr̥; brother is bhrātr̥; name is nāma;
tooth is danta (Consider the words dentist, dental plan, etc.). Cognates
with English and Sanskrit often differ considerably in pronunciation and,
very often, the Sanskrit cognate is with an English word that entered the
language through borrowing, as is the case with dentist and dental, from
the French dent (meaning tooth).
12 The Sanskrit language

Although there are a number of exceptions, languages do not gener-

ally borrow their counting systems from other languages, simply because
they do not need to. The following table presents the numbers one to ten
from English, Sanskrit, Greek, Latin and Lithuanian to illustrate cognate
forms.

Cognates

English Sanskrit Greek Latin Lithuanian

(Ancient)

one eka hen * ūnum * vienas

two dva düo duo * du
three tri tria * tria * trys
four catur tettara * quattuor keturi
five pañca pente quīnque penki
six ṣaṣ heks sex šeši
seven sapta hepta septem septyni
eight aṣṭa oktō octō aštuoni
nine nava ennea novem devyni
ten daśa deka decem dešimt

Notes
The Sanskrit forms are stem forms, meaning that they do not indicate any
grammatical endings.
The neuter nominative forms have been selected for Greek and Latin, where
an asterisk appears after the word in question.
For Lithuanian, the forms are masculine.
If the following three sounds changes are considered, the data presented
start to take on a remarkable similarity across all four languages:

(i) Where Sanskrit has c (pronounced like the ch in chip), Latin and
Lithuanian have a k-­sound, although Latin has the added feature of an
accompanying w (kw). Greek, by contrast, has t.
(ii) The Sanskrit ś (signifying a sh-­sound), is a k-­sound in both Latin and
Greek, but Lithuanian aligns with Sanskrit in having a sh-­sound.
(iii) The Sanskrit ṣ (another sh-­sound), is k or ks in Greek and Latin, except
for when it is at the start of a word. Lithuanian, again, resembles San-
skrit in having sh.
 The Sanskrit language 13

1.4 The origin of Sanskrit

What was the ‘common source’ to which Sir William Jones had referred, in
the talk he gave to the Asiatick Society? Linguistics as a discipline did not
exist at the time Jones addressed the Society, although philology (the study
of language in historical contexts and from historical sources) was certainly
an object of interest amongst those educated in the classical languages of
Europe. Philology, in Jones’s time, was very much the pastime of people
who were well-­read in Greek and Latin, for there lay the focus: the explora-
tion of the languages of ancient Greece and Rome, and how these had influ-
enced the modern languages of Europe. Language change was – and still
is – at the heart of philology, and that was where Jones placed the emphasis.
If one suggests that languages change over the centuries and that they
diverge in the process, this accords not only with common sense but
acknowledges the data available from, for example, Latin and its descend-
ants, or Old English, Middle English and modern English. What is impor-
tant is that an assertion as to language change is capable of being proved;
that there are data to support such an assertion. With Sanskrit, as Jones was
aware, there was ample language information available – it was all around
him, in Calcutta, in the mouths of Brahmins and the general population, as
they recited scriptures and prayers. What was missing was the information
regarding the earliest phase of the language. Certainly, there was Vedic San-
skrit, and that was undoubtedly archaic, but what of the language preced-
ing Vedic Sanskrit? Here, as Jones concluded, was the unknown territory
in which one might explore ancient connections between Sanskrit and the
classical languages of Europe.
Positing an ancestor for Sanskrit, which would also have had a connec-
tion with the oldest-­known languages of Europe, was a tricky affair. Much
territory lay between India and Europe: the Russian steppes, the Black
Sea, the Caucasus (the Russian Empire’s ‘mountain of languages’), Persia,
Afghanistan and the Hindu Kush. If the ancestor of Sanskrit were indeed the
same remote ancestor of the classical languages of Europe, it stood to rea-
son that it must have had a point of dispersal – a point from which it spread
out, over untold centuries, accounting for the majority of the languages of
Europe and northern India. What followed was a hypothesis which has been
the basis of an extraordinary amount of heated disagreement and one that
has persisted to the present day.
The ‘discovery’ of Sanskrit by European scholars, properly speaking,
was the identification of data which strongly indicated a genetic connec-
tion between Sanskrit and the languages of Europe. It coincided with the
14 The Sanskrit language

expansion of the British Empire, particularly in Africa and Asia, and the
scholars who expressed an interest in Sanskrit, if not British, came from
polities which either had similar expansionist aspirations (France and Rus-
sia) or were in the process of working their way to the creation of a nation
state (Germany). The ‘discovery’ in question was, therefore, timely in that
it served to advance the case that civilization had come early to Europe and
that the Asian territories in which European powers were consolidating their
influence, including India, were the heirs of an ancient linguistic heritage
that was European in origin.
German scholars adduced data from languages for which they had
access to literary sources (Greek and Latin, naturally, but also Gothic and
Old Prussian – both extinct – and Lithuanian, related to Old Prussian and
still fully alive). Since the data from Gothic, Old Prussian and Lithuanian
showed many stark similarities with Sanskrit, it was perhaps natural for
German scholarship to conclude that the roots of the ancestral language lay
in Nordic or German territory, an area roughly demarcated as lying between
southern Sweden and the Baltic coast. The available data permitted such an
analysis. In so doing, such scholarship laid the basis for what would, in due
course, become a poisonous ideology which conflated language with eth-
nic identity, culminating in the myth of Aryan supremacy – a concept fully
exploited and promoted by the Third Reich with murderous consequences.
British scholarship, by contrast, was more equitable, although the conse-
quences of its findings served to promote cultural (if not indeed outrightly
racial) superiority all the same. In the two centuries following Jones’s talk
to the Asiatick Society, scholarship has tended to support the idea that the
ancestor of Sanskrit and the languages of Europe, except for the non-­Indo-­
European Basque, Finnish, Estonian, Hungarian and Maltese, originated
from an ill-­defined area spanning Anatolia, the Russian steppes and central
Asia. There is still no consensus as to precisely where within this vast area
the original ancestral language, known as PIE (Proto-­Indo-­European), had
originated. There has been broad consensus as to one thing, however: that
PIE did not originate in South Asia. For many Indian researchers, such a
thing is unacceptable in that it divorces Sanskrit, the language of Hindu
scriptures deemed primordial, from India.
The quest to find the origin of PIE has become, for many, the battle
for the ownership of Sanskrit – a battle which has intensified since Indian
independence and has attained new heights under the BJP, the ruling party
of India at the time the present work is being written. For many millions of
Indians, Sanskrit has served from time immemorial as the language of reli-
gion and of high culture. It is in every sense the hallmark of a Hindu Indian
 The Sanskrit language 15

identity. The extent to which Sanskrit is of significance to the substantial

Muslim, Sikh, Buddhist, Jain, Parsee and Christian population of India is
something which is difficult to gauge. Sanskrit is not much in evidence in
Urdu, where the vocabulary of the language favours words of Persian or
Arabic provenance in the higher registers. Urdu is nevertheless an Indic
language and, as such, belongs to the same branch of Indo-­European as
Hindi, Gujarati, Panjabi, etc. The other languages of India (and not simply
those which are Indic) are flooded with words from Sanskrit. Sanskrit thus
represents, if not a language associated with religious identity, an important
element in the make-­up of the mother tongue of over a thousand million
Indians. If language is identity, then it follows that speakers of Indian lan-
guages (Indic and non-­Indic alike) have a vested interest in the ownership
of Sanskrit.
From a non-­Indian perspective, Sanskrit represents a language that is
not an intrinsic element in the everyday life of the speaker. If it is impor-
tant, that is because it is the language of a civilization which has excelled
in philosophy and literature and has produced a rich material culture along
with a complex musical tradition. For non-­Indians, the positing of an Indo-­
European homeland outside of South Asia is not something likely to pro-
voke an impassioned response. Whether the speakers of PIE migrated to
India from an area south of the Black Sea, through Iran, during a prehistoric
agricultural revolution (the ‘Anatolian Hypothesis’) or whether the migra-
tion was from the Russian steppes, across what are now the independent
states of Kazakhstan, Turkmenistan and Uzbekistan, then down through
Afghanistan, is not a matter of much significance. From an Indian perspec-
tive, however, the matter is key to cultural identity. The notion of a migra-
tion into India, from an area which has yet to be agreed upon, constitutes the
‘Aryan Migration Theory’ (AMT) on which an enormous amount has been
published, largely by researchers working in archaeology and linguistics
(The concept of an ‘Aryan Invasion Theory’ has been superseded but is still
evoked by Indian authors). The response to this is the ‘Out of India Theory’
(OIT), in which the Indus Valley Civilization plays a key role.

1.5 Indus Valley connections

The Indus Valley Civilization arose on the banks of the Indus and was as inti-
mately tied to the Indus River system and dependent on it as the Egyptians
were on the Nile. The civilization was an ancient one, as became quickly
evident to the first archaeologists to discover sites in the early 1920s. Sub-
sequent archaeological work has hinted at dates around 3500 bce, meaning
16 The Sanskrit language

that the early stages of the civilization (often called Harappan, after one of
the chief sites located at Harappa, in the Indian state of Panjab) predates the
building of the pyramids in Egypt by around a thousand years. The Indus
Valley Civilization was extensive, with sites discovered in places as far
afield as Shortugal, close to the Afghanistan-­Tajikistan border, to Sutkagen
Dor, near the Pakistan-­Iran border. The greater part of the sites are, however,
located within the Pakistani provinces of Panjab, Sindh and Balochistan and,
across the border, in the Indian states of Panjab, Haryana, Rajasthan and
Gujarat (Panjab was divided between Pakistan and India during Partition).
Political factors have not been conducive to the excavation and catalogu-
ing of Indus Valley Civilization sites, with the result that many sites are in a
poor state of preservation. This was a civilization built on the alluvial mud
of the Indus, baked hard into bricks which, once excavated, are left open to
the elements. The artifacts which have come to light are tantalizing. They
hint at a sophisticated, urban culture possessing standard weights and meas-
ures (strongly indicative of the importance of trade) and settlements were
planned with great knowledge concerning the management of water for
irrigation and sanitation purposes. Here was the first civilization on Indian
soil, hence the importance accorded to it by modern India in its attempt to
reclaim and rewrite its ancient history rather than accepting a narrative dat-
ing back to colonial times.
The Indus Valley Civilization is, at one and the same time, ancient and
relevant to modern India. It is within this context that one must assess the
claims made, largely by Indian scholars and researchers, that the Aryans
were not an invading force from outside India nor was there any migra-
tion of Aryans into India speaking an early form of Sanskrit. The assertion,
previously stated as the ‘Out of India Theory’ (OIT), is that the roots of
Sanskrit are Indian and that the Indus Valley Civilization, as old as that of
the Sumerians and Egyptians (and considerably older than both Greece and
Rome), was the civilization responsible for creating Sanskrit.
There are many problems in attempting to equate the Indus Valley Civi-
lization with the culture associated with Sanskrit. The following points
address four such problems:

(i) The language of the Indus Valley Civilization has been preserved in
a large number of inscriptions of the cylinder seal variety. These are
short inscriptions and there is no evidence of another, better known lan-
guage to help in the process of decipherment. Claims as to successful
decipherment, which periodically surface, have not met with scholarly
approval.
 The Sanskrit language 17

(ii) The OIT relies overwhelmingly on evidence from the Ṛgveda, the
oldest-­known text in Sanskrit; but the Ṛgveda was not written in
remote antiquity. It was composed and transmitted orally, as continues
to be the case amongst Brahmin communities. There is no way of gaug-
ing if the original composition has remained unchanged or if there has
been later interpolation, when the Ṛgveda was codified. In any event,
the Ṛgveda is mute on the subject of Aryan origins.
(iii) Much is made by the OIT of the mention in the Ṛgveda of the Sarasvatī
river, on whose shores the Ṛgveda was composed. Yet the identification
and location of the Sarasvatī is problematic. Sources from Avestan, a
sister language of Sanskrit and remarkably close to it in terms of its
structure, mentions the river Haraxvatī (Old Persian Harauvati), and
there is little doubt that this is the Harut, in southwest Afghanistan.
(iv) The Indus Valley Civilization went into terminal decline after around
1800 bce and urban centres such as Harappa were abandoned by about
1300 bce, according to archaeological findings. There is no evidence
of an Aryan Invasion having taken place but ample evidence for shifts
in the Indus River system, with rivers changing course or drying up.
Whilst this lends the OIT a certain appeal, the fact remains that this
would not account for a migration of the Indus Valley population to
central Asia or the Iranian plateau. A sedentary society is more likely to
have relocated, over the course of several centuries, to an adjacent terri-
tory (The Indian Panjab is a prime contender). This means that Sanskrit
would not have travelled out of India but, rather, further into India from
a territory which was, to the west, bordered by speakers of Iranian lan-
guages who may well have been part of this gradual migration.

1.6 Myths and realities

The origin of Sanskrit is obscure, with the prevailing schools of thought
inclining towards either the migration of an Indo-­European-­speaking popu-
lation to South Asia or a hypothesis that the language was that of the Indus
Valley Civilization, from whence Indo-­European speech originated. It is
not surprising, therefore, that many assertions are made regarding Sanskrit.
This leads to the perpetuation of language myths of which four are fre-
quently encountered.

(i) ‘Sanskrit is the oldest language’. This is easily dismissed on the basis
that assessing the age of a language is not possible in the absence of
18 The Sanskrit language

documents attesting to its early stages (There is ample evidence, for

example, for the development of Afrikaans as distinct from Dutch but
none that allows us to speculate as to when ancient Egyptian began
to be spoken in the Nile Valley). As regards written evidence for San-
skrit, this is far later than is the case with a number of other Indo-­
European languages. Both Hittite and Greek are older than Sanskrit, if
the determining criterion is written evidence. Until archaeology sheds
further light on the writing of Sanskrit, the earliest documentation of it
comes from the early centuries of the Common Era. It is not in ques-
tion that Sanskrit is the oldest language of the Indo-­Iranian branch of
Indo-­European, based on its conservative structure. Whilst the claim
that it is the oldest language cannot be supported, it is the oldest-­known
language within its branch, with only Avestan, on the Iranian side, as a
contender with respect to antiquity.
(ii) ‘Sanskrit is a perfect language’. From a purely linguistic perspective,
this is easily disproved. Sanskrit contains verbs which are irregular in
the way they are formed when conjugated and there are many excep-
tions to general rules (The reader will be in a better position to evaluate
the claim after having read the present work). The notion of a ‘perfect’
language is highly subjective, appealing to the emotions, and is a con-
sequence of the prestige accorded to the language in question within a
specific cultural context. Without a doubt, Sanskrit was the subject of
extensive polishing throughout its development, as its very name sug-
gests: it is Saṃskr̥ta (refined; composed). It is only from a sociocul-
tural point of view, however, that a claim relating to its perfection has
any real credence.
(iii) ‘Sanskrit is the mother of all languages’. Whilst the first two myths
contain an element of truth, as discussed, this claim is outrightly false.
Sanskrit may, with some justification, be hailed as an older relative
of the Indic languages, such as Panjabi, Gujarati, Hindi and Bengali
but, even with India, it is not the mother of all the other languages.
The four literary languages of the south (Tamil, Telugu, Kannada and
Malayalam) have been heavily influenced by Sanskrit, yet they are not
descended from it. These languages belong to the Dravidian language
family. ‘Indic’ does not connote ‘Indian’, for the simple reason that
the former is a linguistic term, whilst the latter is geographical/politi-
cal. Even as regards Sanskrit being the mother of the Indic languages,
this obscures the fact that the modern languages are derived from ear-
lier forms of spoken language (the Prakrits) via an intermediate phase
(Apabhraṃśa). Sanskrit is not so much the mother of the modern Indic
 The Sanskrit language 19

languages as a great aunt. Languages such as Arabic and Hebrew are

not related to Sanskrit, since they belong to the Semitic language fam-
ily. There is nothing in the vocabularies of Arabic or Hebrew to indicate
a connection with Sanskrit, nor do the forms of nouns or verbs bear any
resemblance to Sanskrit declension or conjugation.
(iv) ‘Sanskrit comes from India’. Notwithstanding the debate regarding the
origin of Sanskrit (1.4. and 1.5), it is not in doubt that Sanskrit was
cultivated on Indian soil. If one focuses on Sanskrit as the language of
Hinduism and of high culture, as opposed to concentrating on the fact
that it descended from an earlier form of Indo-­European speech, then
Sanskrit in its refined form does indeed come from India. An analogy
can be made with English, in that English was formed in the British Isles
(There is ample documentary evidence regarding its development from
Old English through to Middle English and finally to Modern English).
It is irrelevant that the roots of English lie in Continental Europe. The
language in its earliest, distinct form is a product of the British Isles.
By that reasoning, Sanskrit can genuinely be claimed as the product of
Indian society, whatever its roots. The claim that Sanskrit descended
from earlier speech which was indigenous to India is not supported
by the archaeological, historical or linguistic record. Ultimately, such
a claim relies on Vedic Sanskrit having been the primordial language,
with no antecedents. That is more in the nature of philosophy or belief
than provable fact.
2
THE INTERNATIONAL ALPHABET
OF SANSKRIT TRANSLITERATION

2.1 Preliminaries
The International Alphabet of Sanskrit Transliteration (henceforth IAST)
has a venerable pedigree, owing its inception to the collaboration of San-
skrit scholars in the nineteenth century. It aims to provide an accurate way
of transliterating Sanskrit into the Roman alphabet, making Sanskrit more
accessible to those unfamiliar with one or more of the numerous Indian
writing systems which have, for two thousand years, been used to record
Sanskrit and the related Prakrits. It is a stringent system of transliteration,
allowing the reader to work with confidence on Sanskrit, knowing that the
IAST reflects spelling in the Devanāgarī with complete accuracy. Its useful-
ness cannot be overstated. Suffice it to say that, with the IAST, one can start
studying Sanskrit before one has full functional use of the Devanāgarī. The
IAST assists the learner in the process of investigating the language in terms
of its structure and its distinctiveness whilst simultaneously working on the
Devanāgarī. One does not have to acquire the Devanāgarī prior to taking
the first steps into the language. The IAST is the most widely used system
of transliteration in scholarly publications relating not only to Sanskrit but
to the Prakrits.
Whilst non-­Indian scholars of Sanskrit are invariably familiar with the
IAST, it is regrettably the case that it continues to elude a considerable
number of those with an interest in the language. The Internet bristles with

DOI: 10.4324/9780429325434-3
 The IAST 21

websites dedicated to yoga and to other matters pertaining to several mil-

lennia of Indian thought and cultural practice, yet by no means the majority
of such websites – Indian and non-­Indian alike – can ensure that Sanskrit is
accurately represented in the Roman alphabet. Sometimes, the attempts at
transliteration are partial; often, they are merely cursory attempts, undisci-
plined in their nature, to approximate the sounds of Sanskrit. Yet Sanskrit
is precise. Very precise. When a vowel is long, it must be pronounced as
long (and indicated as such). Also, there are two distinct sounds for what
we would in English write as sh and two sets of consonants that, in English,
are transliterated using t, d or n, thereby making no distinction between the
two. The differences between Sanskrit and English (or, indeed, any other
language) must be capable of being reflected by a transliteration system.
Casual transliteration fails to do this.
Using English spelling for Sanskrit is not an option for the serious
learner. In English, for example, a consonant can be used to indicate a sound
which, elsewhere, is usually represented by another letter. Consider the
sound which the letter g has in the word exaggerate. It could be argued that
the letter g has a sound more customarily associated with the letter j and,
additionally, that the first g has the sound of a d. Yet the word is not spelled
*exadjerate (Note, also, the fact that each e has a separate pronunciation,
with the final one being silent). In Sanskrit, this is not possible. Each letter
of the Devanāgarī has its own distinct sound and cannot represent another
sound. Furthermore, there are no silent letters in the Devanāgarī. If some-
thing is spelled, it is pronounced. The IAST is the transliteration system par
excellence for Sanskrit, in that it identifies specific sounds within the lan-
guage and does not leave room for ambiguity as to the corresponding spell-
ing in the Devanāgarī. Non-­IAST transliteration is unsystematic. It does not
assist the learner in the least and should be avoided from the outset of one’s
study of Sanskrit.

2.2 Analysis of a word in the IAST

In the IAST, there are as many letters as are needed to represent, with com-
plete fidelity, the sounds of Sanskrit and the letters of the Devanāgarī writ-
ing system. No fewer, no more. There is nothing arbitrary or redundant in
the IAST. What is visible in the IAST is audible in Sanskrit and capable,
thereby, of being transposed mechanically into the Devanāgarī. The follow-
ing example shows the extent to which the IAST represents the sounds of
22 The IAST

Sanskrit faithfully where other, less principled attempts to represent San-

skrit fall flat:

Non-­IAST transliteration: Krishna

IAST transliteration: Kr̥ṣṇa

There are five letters in the IAST in the spelling of the word in question,
as opposed to the seven employed in the non-­IAST example. First comes
the letter k, then the short vowel r̥ (see 3.1.2) followed by one of two sh-­
sounds in Sanskrit (ṣ), then a nasal (ṇ) and the inherent vowel (a) – which
the Devanāgarī does not represent, if it occurs in the middle or at the end
of a word but which is pronounced all the same (see 3.5.3). Why does the
non-­IAST spelling use two additional letters?
In the IAST, r̥ signifies not a consonant but a sound referred to by
Sanskritists as the ‘vocalic r’. That is to say, it is a vowel. The non-­IAST
transliteration does not reflect this fact and substitutes the ‘vocalic r’ with
what, in English, is a consonant. Since a vowel is missing in the non-­
IAST, resulting in the impossible cluster krshn-­, English spelling requires
a vowel to be added in order that the word may be pronounced. The vowel
in question is i. This is followed by a sh-­sound which is usually repre-
sented in English by the letters s and h in combination (There are some
exceptions, such as the use of -­ss-­ in the words passion, Russian, etc., or
the single s in names of Celtic origin – Sean, Sinead, etc. – and in words
such as sugar and sure). Yet there are two distinct sh-­sounds in Sanskrit,
differentiated according to the position of the tongue during pronuncia-
tion. Non-­IAST transliteration is unable to distinguish between the two,
so attempting to convert the word into Devanāgarī immediately raises the
question as to which Sanskrit sh is involved. Similarly, there are two n-­
sounds in Sanskrit, again differentiated according to the position of the
tongue. Non-­IAST transliteration is understandable, given that it is com-
mon practice to use the everyday writing system and its conventions when
representing words from other languages. Users of English write the Hun-
garian word gulyás as goulash, since it is the closest indication of the
pronunciation, according to English ears.
The word Kr̥ ṣṇa contains no ambiguity in the IAST. The letter k iden-
tifies one – and only one – sound in Sanskrit, for which the Devanāgarī
has only one letter. The letter r̥ likewise indicates only one sound in
Sanskrit which, being a vowel, neither needs nor permits the use of a
vowel immediately after it (Putting two vowel sounds together in the
same word is prohibited according to the rules of Sanskrit phonology,
 The IAST 23

unless dealing with a diphthong). The ambiguity of the two sh-­sounds

of Sanskrit is resolved by the IAST with the use of a dot under the s to
signify that the sh-­sound in question is produced by touching the tip of
the tongue against the hard palate, situated quite high in the mouth (ṣ).
The other sh-­sound is not formed in this way and does not, as a result,
have the same spelling (ś). The IAST also resolves the ambiguity of the
nasal (ṇ) by indicating, as with ṣ, that the n-­sound in question is pro-
duced by raising the tip of the tongue and bringing it into contact with
the hard palate.
With respect to Krishna/Kr̥ṣṇa, the only sound which the English pro-
nunciation of the word has in common with the Sanskrit is k. The English
ri does not equate with the Sanskrit r̥ any more than the English sh results
in the pronunciation of the Sanskrit ṣ or the English n equates to the San-
skrit ṇ. Neither ṣ nor ṇ are sounds that a native speaker of English is
likely to produce, in that the tip of the tongue is further back in the mouth
than is the case for the English sh and n. The sh-­sound of Sanskrit closer
to the English sh is transliterated in the IAST as ś. This is the sh of Śiva.
As for the Sanskrit n-­sound which most closely resembles the English n,
in words such as night, this appears in the IAST as n (Note the absence
of the dot underneath). Care should be taken, however, in the pronuncia-
tion of n, since there is a requirement with respect to the position of the
tongue in its articulation, as shall be made clear in 3.2. Little mention has
been made of the final sound in Kr̥ṣṇa, but this will be rectified in due
course (3.1).

2.3 Additional remarks

When one sees the letter ā, the line above it (the macron) indicates that this
is a long vowel. Non-­scholarly transliteration often makes use of double let-
ters (e.g. aa) to signify that a vowel is long – when an attempt is being made
to distinguish vowel length at all, which is not always the case. In slipshod,
non-­IAST transliteration, the name Rāma can appear as either Raama or
Rama. Worse still, the name is sometimes spelled Raam or Ram, reflecting
its pronunciation in the modern languages of India (The removal of the final
a when reading or reciting Sanskrit is a persistent mispronunciation associ-
ated with speakers of modern Indian languages).
There are a number of transliteration systems which sidestep the use
of double letters to indicate long vowels by using a capital letter (i.e. A
for ā). An example of one such system is the scholarly Harvard-­Kyoto
Convention, of which more will be said at the end of the current chapter.
24 The IAST

The reason is that the IAST is not keyboard friendly: it contains a macron
above the long vowels, a dot below many letters, a dot above a letter (ṅ),
a tilde (ñ) and a slash (ś). In most textbooks using the IAST, the vocalic r̥
and l̥ are dotted, rather than given a subscript circle. This is a pure software
limitation and one that has the disadvantage of occasionally confusing
learners into assuming that a cerebral sound is being indicated, as is the
case with ṭ, ṭh, ḍ, ḍh and ṇ (Cerebral sounds are described and explained
in 3.2.2.3). Unlike the Harvard-­Kyoto Convention, the IAST does not use
uppercase letters to indicate vowel length. Uppercase letters are distinctly
optional.
Since the IAST is not sensitive to the distinction between uppercase
and lowercase letters, it is possible to use uppercase letters for proper
nouns, the titles of literary works and philosophical terms. It goes without
saying that the use of uppercase letters after a full stop is permissible.
About the full stop, this is the only punctuation mark which is neces-
sary when using the IAST. Punctuation is Sanskrit is cursory: there are
no signs or symbols equating to the comma, the semicolon, the hyphen,
speech marks, the exclamation mark or the question mark. Whether or
not a text employing the IAST is enhanced using any of the foregoing
is a moot point and must be left up to the discretion of the user. The fact
remains that the Devanāgarī writing system recognizes only a single ver-
tical stroke (daṇḍa) and a double vertical stroke (dvadaṇḍa) to indicate
the end of a sentence, and the end of a paragraph, text (or a stanza in
poetry), respectively:

daṇḍa
।
dvadaṇḍa ॥
The IAST predates the invention of the modern keyboard and uses dia-
critical marks whereas the Harvard-­Kyoto Convention is keyboard-­friendly.
As with the IAST, every phoneme of Sanskrit can be represented without
any ambiguity by the Harvard-­Kyoto Convention. When the learner reaches
the point where access to an online Sanskrit dictionary is required, the
Harvard-­Kyoto Convention becomes indispensable. Adjustment from the
IAST to the Harvard-­Kyoto Convention is not difficult. Vowels indicated
with a macron in the IAST are uppercase letters in the Harvard-­Kyoto Con-
vention, as are letters with a dot (or circle) underneath them in the IAST.
 The IAST 25

There are only a handful of additional differences to keep in mind. The

Harvard-­Kyoto Convention is case-­sensitive, as the following examples
demonstrate:

IAST: Kr̥ṣṇa/kr̥ṣṇa
Harvard-­Kyoto Convention: kRSNa
IAST: Śiva/śiva
Harvard-­Kyoto Convention: ziva
IAST: Rāma/rāma
Harvard-­Kyoto Convention: rAma
3
THE SOUND SYSTEM

3.1 Vowels and diphthongs

3.1.1 Preliminaries
There are fourteen vocalic sounds in Sanskrit, arranged as follows:

Primary (or simple) vowels: a ā i ī u ū r̥ r̥̄ l̥ l̥̄

Secondary (or complex) vowels: eo
Diphthongs: ai au

To begin with, it is important to distinguish between the primary (or

simple) and the secondary (or complex) vowels. Sanskrit considers the pri-
mary vowels to be in some sense ‘pure’ or ‘basic’, whereas the secondary
vowels are deemed to be composed of a fusion of primary vowels. The
primary vowels of Sanskrit are equally divided between short and long.
Five of them are short and five of them are long. A short vowel is half the
length of a long vowel in terms of the time it takes to pronounce it. In this
respect, Sanskrit is more scrupulous than English, since the short and long
vowels in English do not always differ from each other in length but also in
quality. If one listens carefully to the pronunciation of pull and pool, or look
and Luke, one hears a difference in the vowel which is not just a question of
length: the lips are more rounded with the long vowels and, although maybe
less perceptible, the tongue is in a different position. This difference does
DOI: 10.4324/9780429325434-4
 The sound system 27

not occur in Sanskrit. The lips have the same shape for both short and long
vowels and the tongue is in the same position.
In Sanskrit, the only distinction between short and long vowels is their
duration. This is not always the case as regards a and ā, as will be noted
when investigating pronunciation. The IAST indicates the difference
between short and long primary vowels in Sanskrit by placing a macron
above the vowel if it is long. This is, however, not the case with e and o
which, being secondary (or complex), are deemed to be long. There is no
short e or o in Sanskrit. Both e and o possess a strong relationship with the
two diphthongs of Sanskrit, ai and au, respectively. The diphthongs are
classified as individual sounds within the vowel system (the vocalic inven-
tory) of Sanskrit.

3.1.2 Pronunciation
When exploring the sound system of Sanskrit, it is appropriate to start with
the vowels, as these are the first phonemes identified in the Indic tradition as
regards the phonemic inventory of the language. In so doing, the running order
of Monier-­Williams’s Sanskrit-­English Dictionary will be adhered to, with the
exception of the secondary vowels and the diphthongs. Monier-­Williams lists
e, ai, o and au, in that order, whereas it seems appropriate to investigate e and
o first, then ai and au, given that the latter two are classified as diphthongs
(i.e. a vocalic phoneme which contains two audibly distinct vowels, such as
we hear in the English pronunciation of the words ice and ounce).
With the first pair of primary vowels, a and ā, a word of caution is
required. Short a is often described as having the sound of the letter u in
English words such as cut, sum and up, but there is often variation in the
pronunciation of a amongst Sanskritists. Sanskrit short a is realized, by
some, as precisely the vowel in English cut, sum, up, etc. For others, it is
closer to the American o in the word copper. There is one piece of informa-
tion which suggests strongly that the latter pronunciation is to be favoured.
Short a, if pronounced like the English u in cup, etc., is not the short equiva-
lent of ā. The tongue is in a different position in the mouth, which ought not
to be the case. There is greater uniformity in the pronunciation of long ā: it
is the English a in farmer, father, calm, etc. The short equivalent is obtained
by reducing the length of time taken to pronounce the a in farmer, etc. That
is like the o in the American pronunciation of Boston. Another way to pro-
duce this sound is to aim for a sound intermediate between English cut and
cot, where it is unclear which of the two words is being pronounced. That is
the short a in Sanskrit. It is not pronounced like the a in cat.
28 The sound system

With i and ī, the situation is more straightforward. One has but to bear in
mind that the tongue must not move and thereby produce a vowel which is
qualitatively different between the short and the long forms. Just as was the
case with a and ā, it is easier to articulate the long vowel and then to shorten
it. Long ī is the vowel sound one hears in the English feel, peak, seat, etc.
The short equivalent has, therefore, the same quality but only half of the
duration. With short u and long ū, one may begin with the short vowel,
attested in the English words book, look, etc., and generate the long vowel
by simply lengthening it.
As regards the simple vowels r̥, r̥̄ , l̥ and l̥̄ , these are a source of confu-
sion for learners. This is no doubt because speakers of most languages
are apt to see the letters r and l as capable only of being consonants,
with the result that the idea they could represent vowels appears outland-
ish. Nothing could be further from the truth, but one has to bear in mind
the following crucial fact about vowels. They do not involve any contact
between articulators within the oral tract (lips, teeth, gum, tongue). In
fact, such contact is expressly prohibited. When that simple fact is borne
in mind, it becomes clear that to pronounce the vowel r̥ as if it were ri or
ru is to misunderstand the nature of a vowel. Not only is r – without the
circle beneath it – a different phoneme in Sanskrit, but both *r̥i and *r̥u
represent a diphthong (i.e. a sound made up of two audibly distinct vowel
sounds) whereas Sanskrit only recognizes the diphthongs ai and au. The
vowel r̥ is not composed of the semivowel r plus i or u any more than it
is a diphthong comprising r̥ plus i or u. The pronunciation of r̥ as either ri
or ru is widespread but is to be avoided. Such a pronunciation involves a
touching of articulators (here, tongue and gum), which is strictly forbid-
den in the production of vowels. Additionally, the vowel in the syllable
ri is i, not r̥.
The actual pronunciation of the vowels r̥ and r̥̄ is subject to disagreement
between Sanskritists. Under the circumstances, it might be suggested that
one aims for a sound simulating a purring cat, without letting the tongue
touch the roof of the mouth. The consonant p can then be substituted with
k, to produce the Sanskrit root kr̥, meaning to do, make, fashion (The vowel
r̥ is by no means rare in Sanskrit, unlike its long equivalent r̥̄, which occurs
far less frequently). Two Sanskrit words containing r̥ are likely to be famil-
iar to intending students of Sanskrit. These are Kr̥ṣṇa and Ṛgveda (Note
that Ṛ should ideally have a circle below it rather than a dot. This is a font
limitation). The pronunciation of both Kr̥ṣṇa and Ṛgveda with an audible
i is widespread. This is to be avoided, if at all possible, because it reflects a
practice in which the vocal nature of r̥ is not acknowledged.
 The sound system 29

With respect to l̥ and l̥̄ , very little needs to be said, largely because the
former is rare in Sanskrit – rare to the point that a learner can easily com-
plete the first year of study without encountering it – and the latter is not
attested in Sanskrit, with the exception of two entries given by Monier-­
Williams (The first of these is simply the listing of the sound as the tenth
vowel of the sound system and the second is a noun which can refer either
to the deity Śiva or to ‘the mother of the cow of plenty’). In terms of its pro-
nunciation, l̥ requires, as with all vowels, the absence of touching between
any of the articulators in the oral tract. It cannot, therefore, be pronounced
the same way as the phoneme l, which Sanskrit classifies as a semivowel.
The sound is, in fact, produced by English speakers in the word walk, where
the tongue fails to touch the gum behind the top teeth, but is in the right
position to do so. In effect, it is the Sanskrit l (as semivowel) with no touch-
ing taking place. The result is a sound indistinguishable to the Southern
English pronunciation of the word awe, without the rounding of the lips.
The Sanskrit e and o are often mispronounced by beginners, amongst
whom the tendency exists to pronounce them as if they were the short
vowels in the English forget-­me-­not. In fact, the Sanskrit vowels e and o
are closer to four gate me note. The caveat here is that neither e nor o are
diphthongs in Sanskrit, meaning that they do not contain secondary vowel
sounds. Sanskrit possesses only two diphthongs, ai and au. Like e and o, the
diphthongs are long only. There are no short equivalents. Whereas they are
grouped together, being identifiable as diphthongs, ai and au can equally be
paired up with e and o, respectively.
The relationship between e and ai on the one hand, and o and au on the
other, is strong. It is reflected in the Devanāgarī writing system, in that the
letter for e is the basis of the letter for ai, when these are in initial position
(i.e. when they are the first letter of a word). Similarly, the initial form of o
is the basis of the initial form of au:

e=ए ai =ऐ
o=ओ au = औ

The pronunciation of the diphthongs is not problematic: ai rhymes with

high; au rhymes with how. Hindi speakers often pronounce ai as if it were
a shortened e, and au as if it were a short form of o. Care must be taken not
to conflate ai with e, or au with o. They are distinct phonemes. Sanskrit
does not contain triphthongs (three vowels sounds together). A word such
30 The sound system

as the Greek gaia would not be permissible in Sanskrit, although Sanskrit

can – and does – generate the word gaya (acquisition), formed as follows:
consonant (g) + vowel (a) + semivowel (y) + vowel (a).

3.1.3 Overview
Sanskrit identifies twelve vowels, although two of these (l̥ and l̥̄ ) are excep-
tionally scarce in the language, with the latter attested in only two diction-
ary entries. The vowels are classified according to whether they are primary
or secondary. There are additionally two diphthongs (ai) and (au) which
can, to all extents and purposes, be grouped with the secondary vowels. The
primary vowels have short and long forms, whereas the secondary vowels
(including the diphthongs) are all long. There are no short secondary vow-
els. Vowel length is a crucial distinction in Sanskrit, and the learner is urged
to pay great attention to it from the outset.
It is common practice for learners to conflate the vowels a and ā, i and
ī, u and ū, and care must be taken not to do so. Where a vowel is long, it is
said to take two beats to pronounce. Otherwise said, a long vowel ought to
be held for twice the length of its short equivalent. Unlike English, the short
and long vowels do not differ in quality but in duration. The English words
cut/cart, etc., on closer investigation, involve differences in the position
of the tongue. The movement of the tongue is responsible for producing a
vowel that has a different quality. One may say, therefore, that the differ-
ences between cut and cart are as much to do with the quality of the vowel
as to the fact that they are short and long, respectively. In Sanskrit, short
vowels and their long equivalents differ only as regards the amount of time
taken to produce them.
Of all the vowels in Sanskrit, simple or complex, r̥ and r̥̄ present the
speaker of English with the greatest difficulties. These are often referred
to as the ‘vocalic r’, a wholly accurate description and one which asserts
the fact that the phoneme is a vowel. It is a primary vowel, hence the pres-
ence of both a short and a long form. Given that it is a primary vowel, care
must be taken not to permit the inclusion of any other vowel sound in its
pronunciation. To pronounce r̥ as *ri or *ru is, in effect, either to assert
that it is some kind of diphthong, a phoneme in which two audibly distinct
vocalic sounds may be heard, or that it is nothing other than the syllable ri,
which is composed of the semivowel r and the short vowel i. Both of those
assertions are inaccurate. The vowel r̥ must contain only one sound, its long
equivalent likewise. Another common mistake amongst beginners is to per-
severe with the pronunciation of the complex vowels e and o as if they were
 The sound system 31

the vowels in the English word eggnog. Since they are secondary vowels in
Sanskrit, they are automatically long.

3.2 Consonants

3.2.1 Preliminaries
There are thirty-­three consonants in Sanskrit arranged in five categories
according to how they are pronounced. This number can be increased to
thirty-­five, if two further sounds are included: anusvāra and visarga (These
are investigated separately, since they merit separate discussion). The five
categories correspond to five distinct places of articulation, as follows, with
the place of articulation stated in the first column and the adjectives used to
describe the sounds in question in the second column:

Throat Velar (or guttural) e.g. core

Roof of the mouth Palatal sounds e.g. chore
Hard Palate Cerebral (or retroflex) Not attested in English
Teeth Dental sounds e.g. tore; thaw
Lips Labial sounds e.g. pore

For ease of exposition, and for the sake of clarity, the present work will
first focus on twenty-­five consonants equally distributed across the five
places of articulation and demonstrating the same pattern in the way in
which they are identified (This pattern is not in evidence with the other
eight consonants, anusvāra or visarga). The consonants in question are the
following:

Velar k kh g gh ṅ
Palatal c ch j jh ñ
Cerebral ṭ ṭh ḍ ḍh ṇ
Dental t th d dh n
Labial p ph b bh m

Note that whereas the IAST requires two letters to identify the conso-
nants kh, gh, ch, jh, etc., these are single phonemes in Sanskrit. Otherwise
32 The sound system

said, Sanskrit does not consider them to be the product of two separate
phonemes.
The eight remaining consonants are discussed separately, under the
heading of semivowels (3.3) and sibilants (3.4), with anusvāra (3.5.1) and
visarga (3.5.2) covered in the chapter dealing with additional sounds. In
Sanskrit, the sound of any consonant is stable and predictable in all cases.
The English consonant s can have a s-­sound (sit), a sh-­sound (sugar), a
z-­sound (wise), a zh-­sound (leisure) or, indeed, no sound at all (island). In
this respect, although the English alphabet contains fewer consonants than
the Devanāgarī, letters of the English alphabet do not always represent a
single sound. Once this is considered, it soon becomes evident that English
can represent a greater number of sounds than are present in Sanskrit, albeit
by using a smaller number of letters. Such variation within the realization
of phonemes is not possible in the Devanāgarī, where a letter has one (and
only one) sound and, conversely, a given sound is represented by one (and
only one) letter.
It is important to point out that the notion of what a consonant represents
differs between English and Sanskrit. The English alphabet is not the prod-
uct of a systematic and careful analysis of the sound system of the language.
English spelling is a haphazard affair when compared to Sanskrit. A word
may contain a vowel sound and yet be spelled using only letters identified
as consonants in the English alphabet (by, hymn, my, rhythm, why, etc.). In
addition to being used to indicate a vowel sound, a consonant in English
may also have no sound at all (knight, ballet, etc.). To complicate matters
further still, a single consonantal sound in English may be represented by
more than one letter: knight (the gh being, additionally, silent), rhythm, etc.
Conversely, multiple consonantal sounds may be represented by a single
letter, as in the word taxi. In Sanskrit, a single consonantal sound is always
represented by a single letter in the Devanāgarī, and each consonant letter
within the Devanāgarī represents a single, specific consonant, unvarying in
its pronunciation.
Given the one-­to-­one correspondence of a phoneme and its visual rep-
resentation in the Devanāgarī, no less than the fact that phonemes do not
exhibit variation in the way in which they are pronounced, Sanskrit presents
the learner with a highly consistent and predictable sound system and asso-
ciated writing system. Phonemes are classified as logically as possible, and
the writing system aims at representing the sound system as accurately as
possible. The only legitimate concern that a student of Sanskrit might have
is that the phonemic inventory of consonants in Sanskrit contains sounds
not attested in English.
 The sound system 33

3.2.2 Pronunciation

3.2.2.1 Velars
Consonants in Sanskrit are classified according to two criteria: (i) place of
articulation and (ii) manner of articulation. The first of these relates to the
position of the mouth and the use of the articulators of speech (the soft pal-
ate, the hard palate, the teeth, the lips, the tongue); the second relates to how
the sound is released, once the articulators are in a particular position. This
distinction may easily be grasped by looking at the first set of consonants:
the velars (These are also referred to as ‘gutturals’). It is useful to include
a vowel, so that the consonant can be articulated fully. For this purpose, a
short a is added to the consonant:

ka kha ga gha ṅa

The place of articulation for these consonants is identified as ‘velar’,

which is to say that sounds are produced by the back of the tongue making
contact with the soft palate (the velum). The place of articulation remains
constant for all five phonemes. At no point must the shape of the mouth or
the position of the articulators change. The fact that these phonemes belong
to the same place of articulation means they share the common feature
of being pronounced from the same place in the oral tract. The first two
phonemes in the sequence of velars are ka and kha: two phonemes distin-
guished in the IAST by the absence or presence of the letter h. The first of
these does not have a puff of air, whereas a puff or air is distinctly audible
with the second phoneme. This puff of air is referred to as the aspirate. Now
consider the following sentence:

That’s really thick curtain material.

The first k-­sound (in the word thick) contains no puff of air. Any such
puff of air is cut short by the onset of the following word. In the pronuncia-
tion of curtain, however, the puff of air is present. In Sanskrit, the puff of air
should be clearly audible. It is important to remember that, whilst English
makes nothing of the difference between the pronunciation of the letters
ck in thick and the c in curtain, such a distinction in Sanskrit is phonemic.
That is to say, the k without the puff of air and the kh with the puff of air
are deemed to be two separate phonemes within the language. A simple test
to see if aspiration is being produced is to place one’s palm in front of one’s
mouth and to ensure that a puff of air can be felt.
34 The sound system

To grasp the distinction between the k, which has no puff of air (i.e.
which is unaspirated), and the kh, which does (i.e. which is aspirated), a
pronunciation out loud of the following phrases ought to suffice. The first
of these contains the unaspirated k; the second contains the aspirated kh:

‘Clock out!’
Clock House

Once one can hear (and produce) the difference between the k in Clock
out! and the kh in Clock House, one will effectively have grasped the
distinction between the first two consonantal phonemes in Sanskrit. It is
important to stress the fact that kh is represented by two letters in the IAST
but is one phoneme in Sanskrit. Just as there is a distinction between k and
kh according to the absence or presence of the aspirate, so too this distinc-
tion comes into play as regards the following two consonants in the velar
series: g and gh. The same test as to the absence or presence of the aspirate
can be applied, using the following sentences:

‘Log out!’
Log House

This now accounts for the first four phonemes of the velar series
(k, kh, g, gh), leaving the fifth (ṅ) to discuss. Before doing so, one needs
to know k and kh, as a pair, may be distinguished from g and gh in a
way which is of primary importance. Both k and kh are voiceless (or
unvoiced) whereas both g and gh are voiced. Alternatively, one may say
that the voiced equivalent of k is g, and the voiced equivalent of kh is
gh. Learners tend to struggle with the concept of voice since, unlike the
notion of unaspirated and aspirated, there seems to be no easy test to
determine which is which.
A linguist would have no hesitation in explaining that a phoneme is
voiced when the vocal cords vibrate and that, conversely, a phoneme is
voiceless (or unvoiced) when the vocal cords are at rest. Whilst true, that
does not necessarily facilitate matters for the learner of Sanskrit. An easy
test is, however, quite literally at hand. If a hand is placed on the throat so
that the palm covers snugly the front of the throat and the top row of fingers
is held in place by the chin, one ought to feel a difference between k and g
with the following: ‘It’s c-­c-­cold!’ and ‘It’s a goal, goal, goal!’ By repeat-
ing c . . . c . . . c . . . followed by g . . . g . . . g . . . , a buzzing or a resonating
 The sound system 35

in the neck should be discernible with goal that is absent with cold. This is
an example of the vocal cords in action. They resonate with g, which is a
voiced phoneme, and are at rest with k, which is voiceless. Voice is a sepa-
rate matter to aspiration. Aspiration is always marked by the letter h in the
IAST. An aspirated phoneme can be voiceless in Sanskrit, as is the case with
kh, or voiced, as with gh.
The fifth phoneme in the velar series of consonants is ṅ. As evidenced
by the IAST, this phoneme is not subject to aspiration. Indeed, there is no
aspirated nasal phoneme in Sanskrit. Nasals, irrespective of the place of
articulation, are always voiced. If the nasal ṅ is reminiscent of any of the
preceding phonemes in the series, it is the unaspirated and voiced phoneme
of the series in question. In the case of ṅ, that is g (Pronouncing ṅ with a fol-
lowing vowel is difficult for speakers of English, so it is worth attempting
it with a preceding vowel: long, song, etc.). The nasal is distinct from the
other four phonemes of the series in that the air flow is different. In the case
of k, kh, g and gh, the air flow is through the mouth. One pronounces the
first four phonemes of the series by releasing the air from the mouth. If one
were to pinch one’s nostrils shut, they would sound precisely the same. This
is not the case with the nasals, where the air flow is through the nose and
the mouth simultaneously. It is not possible to pronounce a nasal if one’s
nostrils are blocked. A simple test is to pronounce the word ringer whilst
pinching one’s nostrils shut. If the air flow through the nose is completely
obstructed, one ought to hear the word rigger instead.
The following table identifies the consonants of the velar series accord-
ing to whether they are unaspirated or aspirated. It will be noted that these
terms do not apply to the nasal phoneme, where the term ‘nasal’ is sufficient
to distinguish the distinctiveness of the phoneme within the series. ‘Velar
nasal’ is a full and adequate description of the phoneme ṅ. The evidence of
air flow through the nose as well as through the mouth is obtained by pinch-
ing one’s nostrils shut during the pronunciation of Long House, resulting in
the pronunciation Log House.

Consonants: velars
Voiceless Voiced Voiced
Unaspirated Aspirated Unaspirated Aspirated Nasal
k kh g gh ṅ
‘Clock out!’ Clock House ‘Log out!’ Log House Long House
36 The sound system

3.2.2.2 Palatals
With the palatal consonants, the articulator responsible for producing the
phonemes within the series is the body of the tongue (as opposed to the
back of the tongue, as is the case with the velars). The tongue is raised,
restricting the air channel to a narrow space between the tongue and the
roof of the mouth. In forming the mouth to pronounce consonants belong-
ing to this series, one should feel the sides of the tongue being pressed and
held in place by the top teeth. The front part of the tongue then brushes
against the top of the mouth and traps the air as it attempts to hiss out. Try
pronouncing the word chocolate, but slowly and hesitantly, as if stammer-
ing over the word:

ch . . . ch . . . chocolate?

Think carefully about the position of the tongue during the pronunciation
of ch . . . ch . . . , since that is the position of the tongue for the pronunciation
of all the phonemes of the palatal series of consonants.
As with the velar series, aspiration is found in the palatals. The ch of
chocolate (assuming that there is not too much breathiness in the pronun-
ciation) is the first consonant of the palatal series, which appears as c in the
IAST. Note that there is a tendency amongst learners to pronounce the IAST
c as if it were pronounced k, no doubt as a habit formed by familiarity with
English spelling. The IAST, is not, however, based on English orthographic
conventions or indeed on English pronunciation. The IAST represents the
sounds of Sanskrit, not English. A simple way to remember that the IAST c
is not pronounced like k is to bear in mind that the IAST does not have two
letters for the same sound.
To obtain the unaspirated and aspirated phonemes which represent the
unvoiced palatals c and ch, one must remember that the first of these lacks
an audible puff of air. The same distinction arises between the third and
fourth phonemes of the series (j and jh), expect that these are voiced. In
the following table, the phonemes are identified in the same manner as the
velar series, with examples from English on the bottom row to help reflect
the differences between them. Again, it needs to be stressed that the articu-
lators must remain in the same position for all successive phonemes within
the series. To move the tongue to a different position for the palatal nasal
(ñ) is to disregard completely the basis on which the consonantal inventory
is classified in Sanskrit.
 The sound system 37

Consonants: palatals
Voiceless Voiced Voiced
Unaspirated Aspirated Unaspirated Aspirated Nasal
c ch j jh ñ
Churchill Church Hall Judging ‘Judge him!’ Munchkin

The palatal nasal ñ is often described, in Sanskrit grammars and general

descriptions on the sound system of the language, as being pronounced
like the first n in onion; and so it is, but care must be taken not to let the
y-­sound towards the end become too prominent. The phoneme is, after all,
a single sound, not the product of a nasal plus y, which is a separate pho-
neme (see 3.3). Think of where the tongue is positioned when it reaches
upwards in the mouth to prepare for the articulation of the first n in onion.
It is in the same place as it would be for the n in lunch or munchkin. The
tongue is in that position because the phoneme which follows the n, in
both words, is the palatal sound ch – which the IAST spells as c. Whilst
English orthography is oblivious as to differences in the pronunciation of
n, Sanskrit is extremely attentive to this. The position of the tongue for the
n in onion, lunch and munchkin is different to its position in the following
instruction:

Ban the bomb!

When n comes before a dental sound, it is in contact with the top teeth.

3.2.2.3 Cerebrals
The cerebral series of sounds in Sanskrit are not phonemes typically
found in Indo-­European languages and suggest a long period of contact
between Sanskrit, in its evolution in South Asia, and the languages of the
south of India belonging to the Dravidian language family. In Dravid-
ian languages, cerebral sounds occur widely. In Sanskrit, by contrast, the
voiced sounds ḍ and ḍh are infrequent, lending support to the suggestion
that they found their way into the language through early contact with
Dravidian (It should be pointed out that ṭ and ṭh, although more frequent
in Sanskrit than ḍ and ḍh, occur far less often than their dental equiva-
lents, t and th).
38 The sound system

The cerebrals are problematic to speakers of English in that they repre-

sent sounds which are pronounced with the tip of the tongue striking a point
on the hard palate much further back than is the case with the English t, d or
n. To English ears, the sounds are distinctly South Asian. Forming them is
not difficult. One has only to remember that the tongue needs to touch the
part of the mouth behind the teeth, not the gum directly behind the top teeth
(which is too far forward) and certainly not the teeth themselves. The cer-
ebrals are often referred to as retroflex sounds. This is a good description,
since, in forming these sounds, the tongue tip turns upwards and backwards
in the mouth.
There is no equivalent in English for the cerebral phonemes of Sanskrit,
so no examples from English can be given. It suffices, however, to remem-
ber that the position of the tongue is further back than would be the case for
the pronunciation of the English tuck, duck, etc. Just as with the velars and
palatals, the phonemes of the cerebral series are distinguished according
to aspiration and voice (ṭ, ṭh, ḍ, ḍh) and includes the nasal ṇ, pronounced
by allowing air to pass through the nasal cavity and through the oral cavity
simultaneously.

Consonants: cerebrals
Voiceless Voiced Voiced
Unaspirated Aspirated Unaspirated Aspirated Nasal
ṭ ṭh ḍ ḍh ṇ
(These sounds are not attested in English.)

3.2.2.4 Dentals
The dental series of phonemes in Sanskrit do not present the speaker of
English with any problems, although it must be stated from the outset that
their classification as dentals must be borne in mind. That is to say, they are
produced by the tongue being in contact with the top teeth. It is a common
error, frequent amongst learners who do not know any Indic languages, to
identify these as English dentals, thereby producing sounds not attested in
Sanskrit. The phonemes th and dh are not pronounced like the th in the
English thin and that, respectively.
 The sound system 39

With dental sounds in English, the tip of the tongue is below the top
teeth and extends beyond the teeth. In Sanskrit, by contrast, dentals are
pronounced with the tip of the tongue against the back of the top teeth. The
tongue does not poke out from underneath the teeth. Pronounced this way,
the dental sounds are audibly different to the cerebrals, where the tongue
tip is a considerable distance away from the teeth. Learners are occasion-
ally confused as to why Sanskrit contains two sets of t and d. That is to
see English as a sort of default model for the consonant systems of other
languages, a mindset which results in not grasping the essential point that
Sanskrit makes a phonemic distinction between two sets of t and d accord-
ing to how – or more specifically where – they are produced.
A shift in the place of articulation, in English, can be grasped by the pro-
nunciation of the word butter. This can differ widely according to whether
the speaker is English or American. The latter is likely to pronounce the
word butter almost as if it were spelled budder – that is to say, with the
tongue in a different part of the mouth. The place of articulation may not
be something to which an English speaker pays much attention but, in San-
skrit, it is a crucial factor.
A dental sound in Sanskrit, irrespective of aspiration or voice, involves
the tongue in contact with the top teeth and that is true, also, with the
nasal phoneme n. Consider the pronunciation of n in the word nun. Does
the tip of the tongue touch the teeth? If so, this is the Sanskrit dental n.
If not, the tongue is likely to be striking the gum behind the teeth (This
is known as the alveolar ridge). If the tongue is on the alveolar ridge, all
one must do is slide it even further back (and upwards) in the mouth to
produce the Sanskrit cerebral ṇ. Dental and cerebral phonemes are quite
distinct in Sanskrit. With the caveat that the tongue must be in contact
with the back of the top teeth, the pronunciation of the dental series is
straightforward, as indicated by the English words or phrases in the fol-
lowing table:

Consonants: dentals
Voiceless Voiced Voiced
Unaspirated Aspirated Unaspirated Aspirated Nasal
t th d dh n
Shorter Short hop Rounders Roundhouse Panther
40 The sound system

3.2.2.5 Labials
Along with the velars, the labials are the easiest phonemes for speakers
of English to grasp, since their pronunciation is not distinguishable from
English to any extent that the ear must be trained to hear a difference. The
same requirements nevertheless exist with respect to the way in which
this series of sounds is produced. The aspirate must be clearly audible in
the case of ph and bh, as identified in the English words contained in the
following table.

Consonants: labials
Voiceless Voiced Voiced
Unaspirated Aspirated Unaspirated Aspirated Nasal
p ph b bh m
Appalled Uphold Abandon Dab hand Glimmer

3.2.3 Overview
As regards the twenty-­five consonants investigated so far, there is a consist-
ent pattern. They are distributed within five series, indicating the place of
articulation, and all five series have the same pattern: the first two phonemes
are voiceless (or unvoiced), which is to say that the vocal cords do not
vibrate. The first of these, furthermore, does not have an aspirate whereas
the second one does. The third and fourth phonemes are both voiced and
follow the same distribution as the first and the second of the series (i.e.
unaspirated followed by aspirated). The fifth phoneme of the series is the
nasal appropriate to the series. Here, the air is channeled through the nose
as well as the mouth. All nasals are voiced.
With the velar series, the back of the tongue is in contact with the soft
palate. With the palatal series, the front of the tongue moves upwards in the
mouth, touching the hard palate and leaving only a narrow channel between
itself and the hard palate through which air can pass. The cerebral series
again involves the tongue but, on this occasion, the tip is quickly touched
against the hard palate and released. With the dental series, the tip of the
tongue is in contact with the top teeth. Only with the labial series is the
tongue not involved in the production of sounds. Here, it is the complete
closure of the lips which produces the sound.
 The sound system 41

3.3 Semivowels

3.3.1 Preliminaries
Semivowels are indicated in Sanskrit grammars as sharing the same place
of articulation as the consonants already identified, as follows:

y = palatal
r = cerebral
l = dental
v = labial

Whilst useful, this analysis of the semivowels is based on an approxima-

tion of where the tongue is in the mouth during the pronunciation of the
phonemes in question. There is no doubt, for example, that the tongue is in
a position with y that is not discernibly different to where it would be for
the pronunciation of c or j. Here, however, the tongue does not touch the
top of the mouth. Instead, the sides of the tongue are in contact with the top
teeth – hence there is touching, although it does not involve the front part of
the tongue, as would be the case with c and j. The y of Sanskrit is the y of
English in words such as yacht, yes and young.

3.3.2 Pronunciation
With the semivowel r, the tongue is momentarily in touch with the palate. The
tongue must be in the position it would occupy in anticipation of pronouncing
the English word rabbit, ensuring that the tongue tip extends to touch the part
of the mouth closest to it. The tongue tip should not touch the top teeth, nor
should it be too high up in the mouth. The alveolar ridge, located between
the teeth and the roof of the mouth, is the target location. It is important not
to trill the r in Sanskrit, as in Italian or Spanish, or produce the ‘rolling r’ of
French. The learner, if a speaker of English, should not struggle unduly with
the Sanskrit r, since it is sufficiently close to the English pronunciation of r,
albeit with contact between the tongue and the palate.
With l, the tip of the tongue is in contact with the back of the top teeth,
as is the case for the other dentals (t, th, d, dh, n). It is not uncommon for
l to be pronounced with the tongue touching the gum immediately behind
the top teeth rather than the teeth themselves. Every language permits slight
variation in the pronunciation of its phonemes and Sanskrit is no exception.
42 The sound system

The best advice that can be given is for beginners to pronounce l as they
would in English, making sure that the tip of the tongue is in contact with
either the back of the teeth or the gum behind them.
Variation in pronunciation is particularly marked with the fourth and
final semivowel, v. When Indians pronounce Sanskrit, this phoneme invari-
ably has the sound of the English w, hence the pronunciation of veda as
something which, to English ears, sounds like wader. Whilst v is identified
as a labial sound (the place of articulation being an approximation in the
case of semivowels, as noted), the other labial phonemes of Sanskrit all
involve contact between both lips (p, ph, b, bh, m). There is no such meet-
ing of the lips in the articulation of v. There must, in fact, be an absence of
touching, or else the sound produced is indistinguishable from b.
With non-­Indian learners of Sanskrit, the semivowel v is more often than
not pronounced as if it were the English v of vixen, voice, etc. This practice
ought to be avoided, since the English pronunciation of v involves some-
thing which does not happen in Sanskrit: the meeting of the top teeth with
the bottom lip. The use of the letter v in the IAST is arguably misleading,
since it tricks the eye into assuming an English pronunciation for the semi-
vowel v. It might have been better for the letter w to have been used instead,
although that would not have been ideal, either. Despite the Indian pronun-
ciation of v as the English w, the phoneme does not involve the rounding of
the lips as with u and ū. The lips are kept unrounded for the pronunciation
of v. Alternatively stated, the phoneme is pronounced like the English v,
with the crucial distinction that the teeth must not touch the lip.

3.3.3 Overview
Three of the semivowels (y, r and l) are not problematic to the speaker of English,
with only v calling for particular attention. The semivowel r is close to its English
counterpart but requires the tip of the tongue to be in contact with the gum behind
the upper teeth. There should be no trilling or rolling with r. A simple, quick
touch is all that is involved. The semivowel v is more of a challenge because care
must be taken not to use the teeth in the pronunciation of this phoneme. For those
who are familiar with Spanish, the letter b in the word beber comes close in terms
of pronunciation to the Sanskrit v. All semivowels are voiced.

3.4 Sibilants

3.4.1 Preliminaries
Strictly speaking, the sibilants in Sanskrit are three in number: ś, ṣ and
s. Whilst the phoneme h is not strictly a sibilant (‘whistling sound’) but
 The sound system 43

an aspirate, it will be included alongside the sibilants pure and simple.

In Sanskrit, in any event, all four phonemes are grouped together under
the term ūṣman (the heating ones; the ones producing vapour). The term
ūṣman is particularly fitting when one considers that all four phonemes in
the category of that name (i.e. the three sibilants and the aspirate) involve
an airflow which can be sustained for many seconds. These are the sounds
capable of producing a visible cloud in cold weather. The sibilants are
voiceless. Sanskrit does not, for example, have a voiced equivalent for ś,
heard in the word Asia, nor does it have a voiced equivalent for s, which
is the English z.

3.4.2 Pronunciation
The phoneme ś is straightforward for speakers of English, since it is indis-
tinguishable to the pronunciation of sh in words such as short and shy. The
tongue is in the same position in the mouth as it is for ch, j and jh. For that
reason, it is classified in Sanskrit as the palatal sibilant. Consider how one
might pronounce cat chew. It is not much different to how one would pro-
nounce cat shoe, were the pronunciation not to be accompanied by a short
pause between syllables in the case of the former.
The pronunciation of s is as unproblematic as ś. It is the same s as in
sort and sigh. The only sibilant with which English speakers struggle is
the cerebral ṣ. Here, the thing to remember is that the dot below the letter
in the IAST acts as a prompt. It is the same subscript dot which one finds
with the cerebral (or retroflex) consonants ṭ, ṭh, ḍ, ḍh and ṇ. The tongue tip
touches a part of the mouth which is higher than is the case for ś. The com-
ments made regarding the place of articulation of the cerebral consonants
apply (see 3.2.2.3).
Finally, as regards the phoneme h, this is technically an aspirate rather
than a sibilant – which is to say it is the same sound associated with the
aspirated consonants (kh, gh, ch, jh, ṭh, ḍh, th, dh, ph, bh). Quite simply,
h is the sharp breathing out heard in the expletive: ‘ha!’

3.4.3 Overview
Unlike the sibilants, the consonantal phonemes investigated in 3.2 can-
not be pronounced with sustained breath. They are produced by the
build-­up of air at some point in the oral tract and released in one go.
The phonemes in the fifth column (the nasals) are different, in that they
can be produced with a sustained airflow, by virtue of the fact that there
is airflow through the nose as well as through the mouth. The sibilants
44 The sound system

share the feature of a sustained airflow with the semivowels and the vow-
els but, crucially, the sibilants are not voiced. An easy way to remember
this is the fact that the expletive used in trying to get people to be silent
is: ‘shhh!’. As with the semivowels, the sibilants are allocated a place of
articulation which is approximate, as follows:

ś = palatal
ṣ = cerebral
s = dental

Unlike the semivowels, there is no labial sibilant. That would involve an

impossible situation, ruled out by articulatory constraints, in which the
breath would need to be released whilst the lips were in contact.

3.5 Additional sounds

Along with vowels and consonants, Sanskrit possesses two sounds and
two orthographic conventions which play an essential role in the pro-
nunciation of the language. The present work proposes that the sounds in
question should not be identified as consonantal phonemes. This view is
supported by the fact that they do not function as consonants within the
Devanāgarī. The first of these (anusvāra) relates to nasality and therefore
involves the nasals (ṅ, ñ, ṇ, n, m). The second (visarga) is convention-
ally pronounced as the aspirate h – but is not identical to it, since there
is an additional feature involved in its pronunciation. As for orthographic
conventions (the inherent vowel and virāma), these have a relationship to
each other which may be expressed as complementary, in that they both
relate to the phoneme a. Whilst virāma has a symbol, the inherent vowel
has no symbol at all. In brief, the inherent vowel and virāma are every bit
as important to the pronunciation of Sanskrit as the phonemes investigated
in the previous pages.

3.5.1 Anusvāra
When a word ends with the phoneme m, it is not always the case that the
letter m is marked in the Devanāgarī. Often, what one sees is a dot in its
place. This is indicated in the IAST by the letter ṃ. It is important to note
that, whilst the IAST places the dot underneath the letter m, anusvāra
is written at the top in the Devanāgarī (Some authors employ a vari-
ant form of the IAST which mimics the Devanāgarī practice, hence ṁ).
 The sound system 45

Consider the following examples in the Devanāgarī, both using the let-
ter ka:

Without anusvāra:
क
ka
With anusvāra: कं
kaṃ

The letter/syllable ka, with the anusvāra, is transliterated by the IAST

as kaṃ, although it is worth bearing in mind that the transliteration kaṁ is
occasionally found (This is a rare example of variation within the IAST, the
other being the infrequent representation of the diphthongs as āi and āu).
The following sentences clarify the use of anusvāra:

Rāmaḥ nagaraṃ gacchati

Rāma goes to (the) town.
Rāmaḥ nagaram agacchat
Rāma went to (the) town.

The final word in both sentences is the verb. In the first sentence, the
verb form (gacchati) begins with the consonant g. This is preceded by
the phoneme m, which takes anusvāra. In the second sentence, the verb
form (agacchat) begins with a vowel. It is preceded by m, which does not
take anusvāra. In the first sentence, nagaraṃ (town) uses anusvāra, whilst
this is not the case with respect to the second sentence. The explanation
of anusvāra is that it occurs when the phoneme which follows is a conso-
nant. Where a word-­final m is followed by a word beginning with a vowel,
there is no place for anusvāra. Alternatively said, anusvāra is dependent on
a word-­final m being followed by consonant.
It is worth remarking that, if a sentence ends with a word-­final m, there
can be no anusvāra, since the conditions of its use are not met. Furthermore,
a word-­final consonant at the end of a sentence requires a special sign to
indicate that it is not followed by the inherent vowel a. The sign in question
is virāma, investigated in 3.5.4.
There is one complicating issue as regards the pronunciation of anusvāra,
and this is connected to the place of articulation of the following phoneme. An
adjustment in pronunciation is preferred amongst those who chant or recite
Sanskrit. Effectively, anusvāra takes on the identity of the nasal belonging
46 The sound system

to the same place of articulation as the following phoneme. The spelling of

anusvāra does not normally change, irrespective of whether or not one makes
the adjustment in pronunciation from m to the class nasal of the following
phoneme. That said, there are traditions relating to the writing of Sanskrit in
which the spelling is indeed amended to reflect pronunciation changes. In
such traditions, anusvāra is replaced by the relevant nasal in the Devanāgarī:

Rāmaḥ nagaraṅ gacchati (etc.)

Anusvāra can (and frequently does) occur in the middle of a word. The
circumstances under which that happens are identical to those where it is
attested at the end of a word; that is to say, when m is followed by a con-
sonant, semivowel or sibilant. As with the adjustment which occurs with
word-­final anusvāra, its pronunciation is that of the class nasal of the fol-
lowing phoneme. Note the following examples:

śaṃkara Pronounced śaṅkara

(offering a blessing; epithet for śiva)
saṃcaya Pronounced sañcaya
(an accumulation)
saṃḍīna Pronounced saṇḍīna
(flying together)
saṃtoṣa Pronounced santoṣa
(contentment; joy)
saṃpūrṇa Pronounced sampūrṇa
(filled)

3.5.2 Visarga
Along with anusvāra, the Devanāgarī possesses a sign that has an impact
on pronunciation, albeit that it is not classified as a letter within the writing
system. This sign is visarga, whose name means a sending forth – in the
context of pronunciation, an emission of breath. Its pronunciation is con-
ventionally not distinguishable from the aspirate h, but it differs from the
aspirate in one crucial respect. Visarga stipulates for a glide vowel or echo
vowel. Take, for example, the two-­word phrase sukha duḥkha (pleasure
misery; the ups and downs of existence). Here, visarga is present in the sec-
ond word, indicated in the IAST as a dotted h. The aspirate is also present,
not by itself but as a feature within the unvoiced velar consonant, kh. Whilst
 The sound system 47

most users of Sanskrit might not distinguish between visarga (ḥ) and the
aspirate (h), there is nevertheless a technical distinction between the two.
Visarga specifies for the repetition of the vowel immediately preceding it.
With this is mind, one ought really to hear something resembling sukha
duḥukha. The u in the first syllable of both words has the same quality and
length, but the u-­sound following the visarga is a mere whisper and not
subject to any stress.
Students often struggle with visarga and there is a marked persistence
in producing a short a after it, irrespective of the vowel preceding it. This
is a bad habit and one that the learner is strongly advised not to acquire.
It is extremely rare for visarga to appear within a word, making sukha
duḥkha something of an anomaly, but it is frequently encountered at the
end of words. So pervasive is visarga that it is subject to a comprehensive
set of rules relating to modifications in its pronunciation (This set of rules
is visarga sandhi, the bane of the learner of Sanskrit in his or her first
year of study, if not beyond). It is not a bad mistake to pronounce visarga
as if it were none other than the aspirate h; certainly, this is more accept-
able than producing the wrong echo vowel after visarga. It is desirable, all
the same, that an effort be made to disambiguate between the two. Visarga
gives a certain cadence to the language which is unique. Consider the clos-
ing utterance of mantras: śāntiḥ śāntiḥ śāntiḥ. (peace, peace, peace). The
final phoneme of each word is the visarga. This means that the vowel i is
heard twice in each word: once before the visarga and again after it, as an
echo vowel: śāntiḥi śāntiḥi śāntiḥi (There is a convention, when reciting a
mantra, to draw out the last sound until the breath is exhausted, making the
final echo vowel a long one. This is in stark contradiction to the articulation
of the echo vowel elsewhere).
Since visarga is flanked by a vowel – the same vowel, invariably – this
has led to speculation amongst linguists that it may originally have dif-
fered from the aspirate in being a voiced sound. There has similarly been
speculation that visarga differed, in antiquity, from the aspirate in that it
had a different quality, namely that of being a scraping or hissing sound as
is heard in the Scottish word loch (Speakers of German might recognize
this sound in the pronunciation of words such as ich and milch, although
there is admittedly variation in the way that German speakers produce the
sound in question). This is the stuff of linguistic speculation. All that need
concern the student of Sanskrit is that visarga produces an echo vowel
whereas the aspirate does not. What if visarga is preceded by a diphthong
rather than a vowel (either simple or complex)? The answer is an obvi-
ous one: it is the second element of the diphthong which surfaces as the
echo vowel. The word aśvaiḥ (by/with horses) results in the pronunciation
48 The sound system

aśvaiḥi, not *aśvaiḥa. The echo vowel should be kept short, even if the
vowel preceding visarga is long. Remember that the echo vowel does not
take stress – nor should it be given any, since it does not produce an extra
syllable in the word.

3.5.3 The inherent vowel

Before looking at the final symbol recognized in the Devanāgarī, some-
thing needs to be said about the notion of the inherent vowel, since the final
symbol is virāma (cessation), which indicates the removal of the inherent
vowel. So what is the inherent vowel? It is none other than the phoneme a.
When the Devanāgarī represents a consonant, semivowel or sibilant, the
phoneme in question is automatically accompanied by a short a. In effect,
the Devanāgarī is not an alphabet but a writing system designed to indicate
syllables; and every syllable requires a vowel. The first consonant identi-
fied in Sanskrit is k – yet the Devanāgarī does not represent the letter k. It
represents the syllable ka:

क
ka

The principle is not dissimilar to the English alphabet. One pronounces

b, c and d as ‘bee’, ‘see’ and ‘dee’. For f, a vowel sound precedes the conso-
nant: ‘eff’; and so on and so forth. Where Sanskrit and English are markedly
different is that, if one wants to write the phoneme k rather than the syllable
ka, one must add a sign to achieve this.
When the sound k is articulated, it is followed by something which
sounds distinctly vowel-­like but indeterminate in quality. The indetermi-
nate vowel is, in fact, the most common vowel sound in English, in that it
accounts for most a and e sounds which do not receive stress. For example,
one hears the indeterminate vowel in the initial a of about, around, assume,
etc., and in the second e of emerald, pepper, telephone (The word emerald
contains the indeterminate vowel twice: emerald).
Sanskrit does not recognize an indeterminate vowel. The closest thing
that the language possesses to this is the short vowel a. It is unsurpris-
ing, therefore, that this is this sound which the Devanāgarī writing system
chooses in lieu of the inherent vowel: the vowel that accompanies all non-­
vocalic sounds unless indicated to the contrary. It is important for learn-
ers to avoid producing the indeterminate vowel in Sanskrit. The following
 The sound system 49

word, which means gold in Sanskrit, contains three instances of the short
vowel a – and the short vowel must be audible and identical in all cases:

कनक
kanaka

Vowels in Sanskrit do not undergo a change in pronunciation according

to where they occur in a word. Their quality and length remain constant.
It is also essential to avoid the pronunciation heard amongst speakers of
modern Indian languages, where the final a in a word is omitted altogether
(*kanak). Whilst this might be a pronunciation feature of the modern Indian
languages, it is not one associated with Sanskrit.

3.5.4 Virāma
The principle of the Devanāgarī is that individual letters represent syllables.
The implication is that, whereas vowel letters indicate individual vowels
with no other phonemic material present, the letters for the consonants auto-
matically include the inherent vowel. In describing the sound system of
Sanskrit, the present work uses the IAST which, unlike the Devanāgarī,
allows one to identify individual phonemes rather than having recourse to
a writing system in which the syllable is the basic sound unit. The principle
adopted by the Devanāgarī has certain consequences and it is fitting, in the
concluding parts of a chapter dealing with the sound system of Sanskrit, to
investigate precisely what these consequences are.
As shown in 3.5.3, the Devanāgarī letter for the first consonant, in terms
of the traditional running order of the consonants, involves the phoneme k.
However, the sound k does not by itself constitute a syllable, since it lacks a
vowel, and the Devanāgarī, as noted, is a writing system aimed at represent-
ing syllables. This is where virāma comes into play. Virāma is a diagonal
slash placed at the base of a letter. It instructs the reader to remove the inher-
ent vowel implied by the Devanāgarī:

Without virāma:
क
ka
With virāma: क्
k
50 The sound system

The use of virāma is restricted to the end of a sentence and is not avail-
able, in authentic Sanskrit, anywhere else within the sentence. This leads to
the inevitable question: how, then, can the inherent vowel be removed when
two (or more) consonants occur in a cluster, without intervening vowels?
The solution adopted by the Devanāgarī is to create a conjunct form, merg-
ing the consonants to produce a shape which indicates that the inherent
vowel is not present. Usually, conjunct forms are instantly recognizable;
sometimes, however, they are not. An investigation of the strategies of con-
junct formation requires familiarity with the basic forms of the Devanāgarī,
to which 4.2 is dedicated.
Virāma is not used if a consonant is followed by a vowel other than a.
In this event, the vowel in question replaces the inherent vowel, thereby
creating a syllable, in adherence to the principles of the Devanāgarī (This is
explored in greater detail in 4.2.2).

k+a k+ā k+i k+ī k+u k+ū

क का कि की कु कू
k+e k + ai k+o k + au

के कै को कौ

3.6 Exercise: true or false?

The following exercise contains twelve questions aimed at allowing the
reader to test himself or herself on the information presented so far. The
questions are based on misunderstandings frequent amongst learners, so
one ought to make a note of which questions are answered incorrectly and
reread the relevant chapter to resolve the misunderstanding. The answers
are given on the same page, for convenience’s sake.
 The sound system 51

Questions
(i) The IAST indicates all long vowels with a macron.
(ii) There are only two diphthongs: ia and ua.
(iii) Diphthongs are classified as two phonemes joined together.
(iv) All semivowels are voiced, and all sibilants are voiceless.
(v) The phonemes r̥ and l̥ are semivowels.
(vi) The aspirate (h) is voiceless, therefore all aspirated phonemes must
also be voiceless.
(vii) All nasals are voiced.
(viii) The symbols ṇ and ṅ are interchangeable.
(ix) The symbols ṃ and ṁ are interchangeable.
(x) The echo vowel after visarga is always the same as the vowel before
visarga.
(xi) Capital letters are not permitted in the IAST.
(xii) । ॥
The only punctuation marks are and .

Answers
(i) False. Both e and o are long, even though they are not marked with a
macron.
(ii) False. The only two diphthongs in Sanskrit are ai and au.
(iii) False. Diphthongs are classified as individual phonemes.
(iv) True.
(v) False. They are vowels.
(vi) False. There are as many voiced aspirated consonants (gh, jh, ḍh, dh,
bh) as there are voiceless aspirated consonants (kh, ch, ṭh, th, ph).
(vii) True.
(viii) False. The first of these is the palatal nasal; the second is the velar nasal.
(ix) True. That said, a text ought to use one or the other, not both.
(x) True.
(xi) False. The IAST does not require capital letters, but many authors
use them in the spelling of proper nouns (etc.) and at the start of sen-
tences. Capital letters are used by the Harvard-­Kyoto Convention to
identify specific phonemes.
(xii) True.
4
THE WRITING SYSTEM

4.1 Origin of the Devanāgarī

The Brāhmī writing system is the ancestor of all the scripts of India, irre-
spective of whether they are Indic languages (and therefore Indo-­European)
or members of another family, such as Dravidian. Urdu is the exception,
since it uses a modified form of the Perso-­Arabic script. It is worth not-
ing that Panjabi, Sindhi and Kashmiri can also appear in the Perso-­Arabic
script, although they possess Brāhmī-­based writing systems (The choice of
script tends to depend on whether literacy practices are being undertaken by
a Muslim or a Hindu speaker of the language in question – or a Sikh, in the
case of Panjabi).
The Brāhmī is the oldest-­known writing system for the Indic lan-
guages, with inscriptions dating to around 250 bce still visible on pillars
erected by the Emperor Aśoka, a convert to Buddhism. Aśoka’s edicts
were carved in the vernacular – the Prakrit – and were aimed at allow-
ing anyone who was literate and able to read the Buddhist content of
the text. Attempts to connect the Brāhmī with the Indus Valley script
have not been persuasive, since the forms of the Brāhmī do not resemble
those of the Indus Valley script, with the exception of a few symbols
which would appear to be the result of nothing more than coincidence.
In any event, the Indus Valley writing system used hundreds of symbols

DOI: 10.4324/9780429325434-5
 The writing system 53

not attested in the Brāhmī, and there is more than a millennium separat-
ing the collapse of the Indus Valley Civilization and the monumental
edicts of Aśoka.
The suggestion that the Brāhmī arose from contact with Greek (more in
the nature of stimulus diffusion than the actual borrowing of Greek letters)
seems ill-­founded. Greek expansion into western and central Asia, made
possible by the destruction of the Persian Empire in 331 bce, reached its
height less than a century before the Aśokan inscriptions in Brāhmī. This
period is too short to account for the development of the Brāhmī into a
mature writing system, using quite distinct principles from the alphabetic
Greek. Whatever its place of origin and however ancient the Brāhmī may
prove to be, no evidence has yet come to light of it being used to record
Sanskrit. That said, the abugida principles of the Brāhmī confirm it, beyond
doubt, to be the ancestor of the Devanāgarī – the writing system with which
Sanskrit is primarily associated.

4.2 The Devanāgarī investigated

4.2.1 Preliminaries
The Devanāgarī is one of the principal writing systems of South Asia,
descended from the Brāhmī script and written from left to right. Although
the Devanāgarī is frequently referred to as an alphabet, it is technically an
abugida. The difference between an alphabet and an abugida is that, whereas
the alphabet treats vowels and consonants as having equal value in terms of
their representation, the abugida places an emphasis on the representation
of the syllable, where the consonant takes pride of place and the vowel is
indicated as a secondary feature. The combination consonant + vowel is less
complex in terms of composition than the combination vowel + consonant,
as the following example demonstrates:

क ka
अक् ak
With ak, the Devanāgarī uses the initial form of the vowel a, followed
by the body shape of the consonant k. Virāma is then applied to remove the
54 The writing system

inherent vowel which would otherwise produce the syllable ka. Omitting
the virāma would result in the disyllabic aka.
In Sanskrit, a word cannot contain two vowels together. The diphthongs
ai and au are not exceptions to this rule since they are individual phonemes.
Sanskrit does, however, allow for a semivowel to occur before or after a
vowel representing what in other languages may be indicated by a diph-
thong. For example, the diphthongs ia and ua are absent in Sanskrit but note
the following:

यव
yava (barley)

The word yava contains two syllables, ya and va. The first phoneme
in the word is indicated by y, which is then followed by the inherent
vowel a, not visually represented when in non-­initial position in the
word; the second syllable contains v, similarly followed by the inher-
ent vowel a. At this juncture, it is desirable to see the Devanāgarī in its
entirety. To this end, the abugida is set out on the following page. The
running order of the letters, as presented, is found in Sanskrit dictionar-
ies. The Devanāgarī starts with the vowel a and ends with the aspirate h,
as indicated.

Vowels
Primary a ā i ī u ū r̥ r̥̄ l̥ l̥̄
Secondary (+ diphthongs) e ai o au

Consonants
Velar k kh g gh ṅ
Palatal c ch j jh ñ
Cerebral ṭ ṭh ḍ ḍh ṇ
Dental t th d dh n
Labial p ph b bh m
Semivowels y r l v
Sibilants ś ṣ s
Aspirate h
 The writing system 55

Vowels
Primary
अ आ इ ई उऊ ऋॠ ऌ ॡ
Secondary (+ diphthongs) ए ऐ ओ औ
Consonants
Velar क ख ग घ ङ
Palatal च छ ज झ ञ
Cerebral ट ठ ड ढ ण
Dental त थ द ध न
Labial प फ ब भ म
Semivowels
य र ल व
Sibilants श ष स
Aspirate ह
4.2.2 Vowels
The Devanāgarī differs from an alphabet in another respect: it possesses two
sets of signs for vowels according to where in the word the vowel occurs. If the
vowel is word-­initial, the Devanāgarī uses a full form. This is also the form in
which a vowel would appear if it were to be written by itself. The difference
between the forms is not one of uppercase as opposed to lowercase letters. The
Devanāgarī does not make such a distinction. Initial forms are more complex
in terms of their shape than their medial (non-­initial) equivalents. It is worth
remembering that the inherent vowel has no form at all when it is not initial.
Fortunately, it is the only vowel to lack a form in medial position. Were it
the case the other vowels followed suit, the Devanāgarī would be a vowel-­
defective writing system, as is the case with the old Semitic alphabet – that is to
say, the original alphabet – and some of its living descendants, such as Hebrew
and Arabic (and alphabets adapted from Arabic, like Persian and Urdu).
The initial forms of the vowels display a strategy which, whilst not
entirely consistent, greatly assists the learner. Since Sanskrit recognizes the
relationship between short vowels and their long equivalents, that relation-
ship is visually reflected in the Devanāgarī. Effectively, the form of the
56 The writing system

initial short vowel provides a template for the long vowel, which is distin-
guished from the short vowel by the addition of an element. The element in
question is either a vertical stroke or a hook. The following table shows the
initial forms of the simple vowels in more detail:

Short vowel Long vowel Strategy

अ आ
The long vowel adds a vertical stroke to the right
a ā of the short form

इ i ई ī
The long vowel adds a hook above the
short form

उ u ऊ ū
The long vowel adds a hook to the right of the
short form

ऋ r ॠ r̥̄
The long vowel adds a hook to the bottom right of
the short form

ऌ l ॡ l̥̄
The long vowel adds a hook to the bottom right of
the short form

As regards the complex vowels and the diphthongs, they too display a
strategy, but one that is different to the simple vowels. The distinction is
not one of length, since the complex vowels and the diphthongs are long.
There are no short equivalents. Here, the distinction is between long vowel
and diphthong, and it is appropriate that the strategy used to distinguish
between the phonemes is not the same strategy employed with the simple
vowels. With the complex vowels and diphthongs, the strategy involves the
addition of a diagonal slash above the body shape of the diphthong, even if
a diagonal slash is already present.

Complex vowel Diphthong Strategy

ए ऐ
The diphthong adds a diagonal slash above the
e ai body shape for e

ओ o औ au
The diphthong adds a further diagonal slash to
the body shape for o

It is regrettable that the Devanāgarī recycles the body shape of the letter
a to produce not only the long equivalent, ā, but also the complex vowel
o and the diphthong au. This is not helpful to the learner. A simple way to
remember which phoneme is being indicated is to bear in mind that o and au
are both long and, as such, employ the body shape of ā. There are two verti-
cal strokes in the formation of the body shape. The diagonal slash – either
 The writing system 57

single or double – is a property of long vowels and diphthongs only. It is not

used with short vowels (This theme reoccurs with the medial forms).

अ = a आ = ā ओ = o ऐ = ai औ = au
It seems to be desirable, at this point, to discuss the OM sign which,
whilst not a letter of the Devanāgarī, is frequently encountered. There are
two ways of representing OM: by using the symbol or by using the letters
of the Devanāgarī. This situation is analogous to the choice presented in the
use of & or and. Whilst deemed in Hinduism to be the primordial sound,
OM is composed of three elements according to the Devanāgarī: a + u + m
(Note that a + u does not produce the diphthong au but, rather, a complete
fusion of the simple vowels, resulting in the complex vowel o).

ॐ = ओं oṃ (with anusvāra) or ओम् om (with m and a virāma)

The medial forms of the vowels, as noted, are different to the initial forms.
These are less complicated in their formation and exemplify the point that
the Devanāgarī is an abugida rather than an alphabet. Here, the vowels are
written around the consonant. Most vowels, in this form, appear above or
below the consonant, but there are two notable exceptions. These exceptions
relate to the short vowel i and the long equivalent, ī. On the following page, a
table is given in which both the initial forms are given (above) and the medial
forms (below). The medial forms are accompanied by a dotted circle which
indicates the location of the consonant vis-­à-­vis the vowel in its medial form.
As was seen with a and ā, the short a does not possess a medial form, since
the concept of the inherent vowel comes into operation. With ā, a vertical
stroke is added to the right of the consonant, just as there was an additional
vertical stroke in the initial form, to distinguish it from its short counterpart.
With i and ī, the strategy is different. Short i is indicated, unlike short a. It
appears as a vertical stroke above which there is a crest – and it appears before
the consonant. It is the singular exception to the rule that the Devanāgarī is
written from left to right. The medial form of i is a source of frustration to
anyone who writes Sanskrit, where the tendency exists to write the consonant
without leaving space for the short i. The medial form of ī is much more in
keeping with the principles of the Devanāgarī: it occurs after the consonant.
With u and ū, the strategy is different again. Here, the vowel is indicated
below the consonant. With u, the medial form resembles the number 6 which
has fallen backwards; with ū, the number 6 has toppled forwards. The easiest
way to remember the distinction is to remember the sequence: short and long;
58 The writing system

backwards and forwards. As for the remaining simple vowels (r̥, r̥̄, l̥ and l̥̄ ), yet
another strategy is employed, but a simple one. With r̥, a little hook, somewhat
resembling the letter c in English, is added below the consonant. As for its long
counterpart, r̥̄, two little hooks are added. The medial form of l̥ mimics the
initial form, without the top line and the neck attaching the top line to the body
shape of the letter, and the medial form of l̥̄ is not found in Sanskrit. If it were,
it would resemble its short equivalent, with an extra little hook (For the medial
forms of e, ai, o and au, see the following notes accompanying the table).

Vowels
Primary vowels
(Initial and medial forms)

अ a आ ā इ i ई ī उ u ऊ ū
n/a ◌ा ि◌ ◌ी ◌ु ◌ू
ऋ r̥ ॠ r̥̄ ऌ l̥ ॡ l̥̄
◌ृ ◌ॄ ◌ॢ ◌ॣ
Secondary vowels (+ diphthongs)
(Initial and medial forms)

ए e ऐ ai ओ o औ au
◌े ◌ै ◌ो ◌ौ
Note on the secondary vowels (+ diphthongs)
The medial forms of e, ai, o and au follow a certain logic, although it must be
said that they are responsible for the greater part of reading errors amongst
learners. With this in mind, the following points ought to assist matters:

(i) All secondary vowels and diphthongs in Sanskrit are long.

(ii) The initial forms of o and au use the body shape of short a – but with an
additional vertical stroke (resulting in ā) which remains visible in the
medial form.
(iii) The diagonal slash, present in all four medial forms, is reserved to e, o
and the diphthongs.
 The writing system 59

(iv) The key to remembering these medial forms is to focus on e and o. The
first of these uses the diagonal slash only; the second one has the diago-
nal slash above a vertical stroke. The associated diphthongs, ai and au
respectively, merely add an additional diagonal slash.

4.2.3 Consonants
As discussed in 3.2, the Devanāgarī presents twenty-­five consonants out
of thirty-­three in an order which is much more patterned than is the case
with the vowels. These twenty-­five consonants are accommodated within
a structure in which each component of the structure, based on the place
of articulation, adheres to a principled and logical pattern. First comes the
phoneme which possesses neither voice nor aspiration (k, c, ṭ, t, p), fol-
lowed by the phoneme which, whilst still voiceless, contains aspiration (kh,
ch, ṭh, th, ph). This accounts for all the voiceless consonants of Sanskrit,
except for the sibilants. Then follow the phonemes which are the voiced
equivalents of the first two, similarly distinguished by the unaspirated pho-
neme followed by its aspirated equivalent: g, gh; j, jh; ḍ, ḍh; d, dh; b, bh.
The fifth phoneme in the sequence is the nasal for the place of articulation
in question (ṅ, ñ, ṇ, n, m). All nasals are voiced and unaspirated.
In contrast to the situation with the vowels, the Devanāgarī does not
indicate a visual relationship between consonants. Aspirated phonemes
and their unaspirated equivalents, for example, have quite distinct letter
shapes (except for two pairs of consonants, as discussed in the following
paragraph). Whilst marvellously classified and arranged with respect to
their sound, the consonants are a lot less ‘tidy’ in their visual represen-
tation. Indeed, learners invariably encounter difficulties in distinguishing
between consonants as they are represented by the Devanāgarī – a task not
facilitated by the similarities between various letters. Note the similarities
in form between the following pairs, as this is where confusion regularly
occurs (The inherent vowel is present and therefore indicated in the IAST).

घ gha ध dha
ङ ṅa ड ḍa
ज ja ञ ña
ट ṭa द da
भ bha म ma
60 The writing system

There are some similarities between consonants which assist the learner.
Such similarities are, however, limited to two pairs of consonants:

ट ṭa ठ ṭha
प pa फ pha
With ṭa/ṭha and pa/pha, a strategy takes place which is not dissimi-
lar to one encountered with a number of vowels. With ṭa, the tail of
the body shape does not complete a circle, unlike ṭha. In other words,
there is an addition to the body shape of a letter to create a distinction
between two phonemes which are in some sense paired (With the vow-
els, the pairing is the relationship between short and long counterparts).
This is also the case with pa and pha, where we see a strategy hardly to
be differentiated from u and ū. There, a hook is added to the right-­hand
side of the body shape of pa to create pha. The pairing of these conso-
nants is not one of length, since Sanskrit does not classify consonants
as short or long, but a distinction between unaspirated and aspirated
counterparts.
Variation between fonts is relevant when discussing the consonants. This
will be explored in more detail (4.4.2). For now, it is opportune to look at the
variation which may occur between consonants which are frequently con-
fused for others in the Devanāgarī. The present work uses a font (Sanskrit
Text) in which the body shapes of the consonant pairs gha/dha and bha/ma
are disambiguated by the presence of a curl on the top left of dha and bha.
Whilst these curls might be present in most (if not indeed all) contempo-
rary fonts, the Devanāgarī one sees in textbooks from the period before the
second half of the twentieth century does not exhibit them. The resulting
problems in disambiguation are compounded by the fact that the typeface
in older Sanskrit textbooks often has poor legibility. When the curls in both
dha and bha are replaced by a short top line, the similarity in form with gha
and ma, respectively, is increased. The only disambiguating factor is that
with gha and ma, there is a continuous topline but with dha and bha, the
top line is interrupted.

Old typeface Sanskrit Text IAST

ध dha

भ bha
 The writing system 61

Consonants
Velars
Voiceless Voiced Voiced
Unaspirated Aspirated Unaspirated Aspirated Nasal

क ka ख kha ग ga घ gha ङ ṅa
‘Clock out!’ Clock House ‘Log out!’ Log House Long House

Palatals
Voiceless Voiced Voiced
Unaspirated Aspirated Unaspirated Aspirated Nasal

च ca छ cha ज ja झ jha ञ ña
Churchill Church Hall Judging ‘Judge him!’ Munchkin

Cerebrals
Voiceless Voiced Voiced
Unaspirated Aspirated Unaspirated Aspirated Nasal

ट ṭa ठ ṭha ड ḍa ढ ḍha ण ṇa
(These sounds are not attested in English.)

Dentals
Voiceless Voiced Voiced
Unaspirated Aspirated Unaspirated Aspirated Nasal

त ta थ tha द da ध dha न na
Shorter Short hop Rounders Roundhouse Panther

Labials
Voiceless Voiced Voiced
Unaspirated Aspirated Unaspirated Aspirated Nasal

प pa फ pha ब ba भ bha म ma
Appalled Uphold Abandon Dab hand Glimmer
62 The writing system

4.2.4 Semivowels and sibilants

The semivowels and sibilants, as with the other consonants (but not the
vowels), are accompanied by virāma when the inherent vowel is removed.
The examples on the following page do not contain virāma. Accordingly,
the IAST accompanying the Devanāgarī includes a short a. The Devanāgarī
letters for the semivowels and sibilants are unique, in that they are not
deemed to be modifications of the letters representing any of the vowels
or the consonants. That said, there are similarities in form which need to be
addressed. In this respect, the Devanāgarī letters representing the syllables
ra, va and ṣa deserve a mention.
The semivowel ra has a similar form to the initial stroke of the body
shape for the letter kha. In fact, the letter kha, in many fonts (such as Nir-
mala), is disconcertingly close to the Devanāgarī spelling of rava.

Nirmala Sanskrit Text IAST

र र ra

ख ख kha

रव रव rava

With kha, the initial (left-­hand) stroke of the body shape should curl
over to the right at the base. It is not necessary for this stroke to touch the
body shape to the right of it but, as is the case in many fonts (including
Sanskrit Text), that is precisely what it does. This is good practice to adopt
in handwriting, given that the touching of the two elements comprising the
body shape of kha serves to disambiguate kha from rava.
As regards va and ṣa, these are only distinguished from ba and pa,
respectively, by the absence or presence of a diagonal slash within the body
shape of the letter.

व va ब ba
ष ṣa प pa
 The writing system 63

Semivowels and sibilants

Velars
Semivowel (voiced) Aspirate (voiceless)

ह
n/a ha
Hear

Palatals
Semivowel (voiced) Sibilant (voiceless)

य श
ya śa
Yoghurt Fish

Cerebrals
Semivowel (voiced) Sibilant (voiceless)

र ष
ra ṣa
Rare Hushed

Dentals
Semivowel (voiced) Sibilant (voiceless)

ल स
la sa
Follow Snake

Labials
Semivowel (voiced) Sibilant (voiceless)

व
va n/a

(See 3.3.2.)
64 The writing system

4.3 Conjuncts and conjunct formation

4.3.1 Preliminaries
Conjuncts are often referred to as ‘conjunct consonants’ or ‘compound
consonants’. This indicates that the phenomenon in question is one which
involves consonants only. A conjunct is the outcome, in the Devanāgarī, of
representing two or more non-­vowel sounds together, without an interven-
ing vowel. Sanskrit, as explained, has thirty-­three consonants, composed of
the twenty-­five investigated in 3.2 (distributed across five places of articu-
lation and distinguished according to whether they are voiceless/voiced,
unaspirated/aspirated), four semivowels, three sibilants and the aspirate. All
consonants are involved in conjunct formation, vowels are not. Anusvāra
and visarga do not form conjuncts.
Conjuncts can be straightforward to read, in cases where the Devanāgarī
letters suddenly undergo a transformation from their basic forms but remain
legible all the same. That said, conjuncts can take on a form in which the
individual elements involved are not discernible. At that point, no amount
of peering at a page illustrating the basic forms of the Devanāgarī, attempt-
ing to investigate possible contenders, is likely to shed any light. Conjuncts
of this type (see 4.3.5) are, fortunately, in the minority. They must be learnt
as unique forms, such as the following:

क् k+ ष ṣa = क्ष kṣa e.g. मोक्ष mokṣa (liberation)

Why do conjuncts exist? Quite simply, because the Devanāgarī assumes
the presence of the inherent vowel a in the pronunciation of any letter indi-
cating a consonant, unless the inherent vowel is removed (The removal is
achieved with virāma, as seen in 3.5.4). All well and good, but this does beg
the question as to why, if the removal of the inherent vowel is easily under-
taken by the insertions of virāma, there should be any need for the com-
pounding of letters. There is no obvious answer to this question. The answer
is quite likely to be connected to philosophical perspectives on the primor-
dial nature of the syllable. Certainly, virāma is used by the Devanāgarī, illus-
trating the point that a consonant, vowel or semivowel without the following
inherent vowel is easily represented by the writing system. The Devanāgarī
does not, however, technically allow for the use of virāma unless the letter to
which it is attached is the final letter of the sentence.
All of the writing systems descended from the Brāhmī use conjunct
forms, with the notable exception (amongst the Indic languages) of the
 The writing system 65

Gurmukhī script devised for Panjabi. The Gurmukhī effectively limits

conjunct formation to two phonemes, r and v, reducing these to subscript
letters whose forms are immediately legible. Further south, the Dravidian
languages have adopted the principle of conjunct forms. Here too, there is a
singular exception, with Tamil showing innovation in the creation of a sign
(a dot above the letter as opposed to a slash below the letter, like virāma) to
remove the inherent vowel. Conjuncts are, therefore, a feature in the writing
systems of most of the languages of India. They are by no means restricted
to the Devanāgarī.
Attempting to enumerate the conjunct forms is a pointless exercise, and
it suffices to say that the number of conjuncts attested in the Devanāgarī
runs into the high hundreds. It is certainly not the case that one has but to
calculate the sum of 33 x 33 (33 representing the total number of letters
for the consonants). Such a thing ignores two crucial points: (i) a given
consonant does not automatically generate a conjunct form with all other
consonants, and (ii) conjuncts are not limited to two elements, but may con-
tain three, four or even five. Despite the high number of conjunct forms,
there are strategies which make the recognition of conjuncts considerably
less arduous than might otherwise be the case. These strategies will now be
explored.

4.3.2 Removal of the vertical stroke

A cursory glance at the Devanāgarī reveals that most of the letters represent-
ing consonants are typified by having a vertical stroke on the right-­hand
side of the body shape of the letter (where ‘body shape’ indicates the form
of the letter without the horizontal top line). This vertical stroke provides
the means for the first and most extensively used strategy in the formation
of conjuncts. Quite simply, if the first element in a conjunct contains a verti-
cal right-­hand stroke, this is removed and the remaining strokes of the body
shape of the letter are merged with the following element. Occasionally,
when the following consonant also contains a vertical stroke, the vertical
stroke of that letter is adopted by the body shape of the previous element.
This results in the second element within the resulting conjunct being in
some way displaced. Consider the following examples:

न् n + त ta = न्त nta e.g वेदान्त vedānta (end of the Vedas)

ख् kh + य ya = ख्य khya e.g. ख्यात khyāta (named)
66 The writing system

(All examples of conjuncts will include the inherent vowel a at the end,
so that the conjunct can be articulated. Removing the inherent vowel would
require the final element of the conjunct to be marked with a virāma.)
With nta and khya, there is no displacement of the second element (-­t
and -­y, respectively), which keeps its shape intact. Note, however, the fol-
lowing conjunct, where the second element (-­n) is displaced and, in the
process, the horizontal stroke in the body shape of the letter dips by 45°:

ग् g + न na = ग्न gna (Often confused with ग्र gr)

From the outset of investigating conjunct forms, it is important to note
that variation exists according to the Devanāgarī font selected by an author.
In fairness, this is not a modern phenomenon, given the great variation in
conjuncts in the Sanskrit manuscript tradition. It seems quite fitting that
computer fonts should mimic the situation as it has always been with
respect to the writing of Sanskrit (see 4.4.2 for more on font variation). The
conjunct gna appears in the Mangal font as the following:

ग्न gna

With the Mangal font, it is evident that the strategy is much more in
keeping with the examples for nta and khya, where the first element merely
forfeits its right-­hand vertical stroke. By contrast, it could said that the other
font (Sanskrit Text) generates a conjunct in which elements are subject to
a stacking strategy. Such a strategy certainly exists and it is to this that the
discussion now turns.

4.3.3 Stacking
The removal of the vertical stroke in letters containing a vertical stroke is
the main strategy in conjunct formation. This is to be expected, given that
twenty-­two out of the thirty-­three eligible letters are accommodated, where
they constitute the first element of a conjunct. There are eleven letters, as
follows, where such a strategy is not available:

Velar: k, ṅ, h
Palatal: ch
Cerebral: ṭ, ṭh, ḍ, ḍh, r
Dental: d
Labial: ph
 The writing system 67

(Note that the aspirate h and the semivowel r are given in terms of their
place of articulation and that there are no sibilants involved.)
Where one of these eleven letters constitutes the first element of a con-
junct, the strategy adopted by the Devanāgarī is to stack them on top of the
following element within the conjunct. Although not invariably the case,
such a strategy results in the second element losing part of the shape it
possesses when written separately. The loss can involve the top line (as
in the conjunct ṅga) additionally involve a 45° anticlockwise rotation (as
with ddha) or take the form of the removal of part of the body shape of the
second element (hma).

ङ् ṅ + ग ga = ङ्ग ṅga e.g. गङ्गा gaṅgā (the Ganges)

द् d + ध dha = द्ध ddha e.g. बुद्ध buddha (the Buddha)
ह् h + म ma = ह्म hma e.g. ब्राह्मण brāhmaṇa (a Brahman)
From the foregoing examples, it can be noted that the first and second
elements do not allow for a space to be left between them. This a point
worth noting. Despite what certain fonts might indicate, conjuncts require
the shapes of the letters involved to be in contact with each other. Otherwise
said, the elements must touch.
Conjuncts can be very subtle in their representation. With ṅga, one sees
the base of the letter ṅa, albeit not flat, standing in place of the top line
which would be a feature of the letter ga. With ddha, the letter da lends its
tail to dha, which then treats it as if it were its vertical right-­hand stroke,
rotated. The conjunct hma involves neither of these things but, instead, sees
the loop of the body shape for ha acting like a tendril and attaching itself
to the square body shape of the letter ma, which loses the left-­hand vertical
stroke in the process. The word brāhmaṇa contains two conjuncts: hma,
as already seen, and bra. Conjuncts involving the semivowel r have unique
forms that are both simple and regular.

4.3.4 Conjuncts involving r

As with the other consonants, the Devanāgarī letter ra represents a
syllable (i.e. r plus the inherent vowel a) unless virāma is added to it,
removing the inherent vowel and reducing the syllable down to the bare
phoneme, r. This phoneme is extremely prevalent in conjunct formation
68 The writing system

and possesses its own strategy. Effectively, it takes one of two signs,
according to whether it is the first element in a conjunct (r-­) or the second
or subsequent element (-­r).
When r is the first element in a conjunct, it takes the form of a semicir-
cular loop above the top line of the letter of the second element. Remember
that conjuncts are read from top to bottom when they are stacked, so the
placing of the loop in this fashion is completely consistent with this princi-
ple. It is important to note that, where the letter following r has a right-­hand
vertical stroke, the loop is positioned directly above this.

र् r + ग ga = र्ग rga
There is one complicating factor in this process, and it is one which
causes difficulties to the majority of learners as regards learning to read
with confidence. The loop will occupy a vertical line that stands on its own
in preference to one which forms the body shape of a letter. In other words,
it will more readily attach to the medial form of the long vowel ā, if such an
option presents itself. The loop can be displaced quite far to the right of the
word, causing the reader to fail to pronounce r at the proper time.

मार्गाः mārgāḥ (roads)

(This is not pronounced *māgrāḥ, since that would involve a conjunct in
which -­r were the second element, for which the loop is not used.)
The displacement of the loop to the right can be considerable, such is the
fondness which r-­displays for occupying a vertical stroke.

कार्त्स्न्य kārtsnya (entirety)

When there is no vertical stroke following r-­, the loop has no choice but
to settle over the letter it precedes.

मर्द marda (friction)

By contrast with the strategy employed by r-­, -­r (where r constitutes
the second element in a conjunct) is much more straightforward. In this
instance, -­r takes the form of a diagonal line added to the first element, at its
base. Where the body shape of the preceding letter has a vertical stroke, the
diagonal line attaches itself to it.
 The writing system 69

ग् g + र ra = ग्र gra
क् k + र ra = क्र kra
In cases where the letter preceding -­r does not contain a vertical stroke,
a pseudo-­vertical stroke is provided, but rotated through 45°. This results in
what resembles an upside-­down ‘v’ at the base of the preceding letter.

ट् ṭ + र ra = ट्र ṭra
छ् ch + र ra = छ्र chra
द् d + र ra = द्र dra
(The letter da has a tail which may appear either as a curl or as a straight
line, depending on handwriting or the choice of font. In either event, it is
straightened to accommodate the conjunct sign for -­r.)
There are two conjuncts involving -­r, namely hra and śra, that are worth
a separate mention. Although the letter ha does not possess a vertical stroke,
it attracts the same conjunct sign for -­r as if it did. Also, the conjunct śra,
despite possessing a vertical stroke, does not simply add the conjunct sign
for -­r to the body shape of the letter śa.

ह् h + र ra = ह्र hra
श् ś + र ra = श्र śra
With śra, the body shape of the letter ś undergoes a radical change,
where the left-­hand stroke (and the top line above it) is reduced to what
resembles a ribbon. This is a reminder that, like ra, other letter shapes can
also become modified in forming conjuncts, to the extent that they become
distinct in form.

4.3.5 Unique forms

In 4.3.2 and 4.3.3, the main conjunct strategies were explored. In 4.3.4, it
was demonstrated that the letter ra undergoes a transformation according to
whether the phoneme r is the first element within a conjunct or the second
(or subsequent) element. Further, 4.3.4 identified the letter śa as having
70 The writing system

a distinctive form when it was the first element in the conjunct śra. The
reader will be pleased to know that the major conjunct strategies described,
together with those used by r, account for the overwhelming number of con-
juncts encountered in the Devanāgarī. The conjunct śra may have appeared
to add an unwelcome complication to matters, but it should be noted that
śa, unlike ra, reserves its transformation to when it is the first element of
a conjunct – and even then, the transformation in question does not always
occur. Where it is the second element of a conjunct (which is rare), it retains
its body shape.

श् ś + र ra = श्र śra (with transformation)

श् ś + व va = श्व śva (with transformation)
श् ś + म ma = श्म śma (without transformation)
श् ś + य ya = श्य śya (without transformation)
ञ् ñ + श śa = ञ्श ñśa (with ś as second element)
There are a number of conjunct forms which are unique and have simply
to be learnt as they are encountered. These unique forms may not account
for more than about one in twenty of all conjuncts, but some are extremely
frequent. Great care should, therefore, be taken from the outset of one’s
study of Sanskrit to learn them as they appear in a text. Whilst most text-
books aimed at the learner use virāma to simplify the process of reading the
Devanāgarī, which is useful and serves a purpose, the unique forms cannot
be derived through an application of the strategies discussed. They do not
follow any identifiable strategy. On the following page, some of the most
common unique conjuncts are given. It is, however, merely an introduction
to unique conjunct forms, the recognition of which must form part of ongo-
ing study.
It is not uncommon, when starting to read Sanskrit, to come across con-
junct forms which appear to differ from those printed in a textbook with
which one may have become familiar. The difference is a consequence of
font selection and the learner needs to be aware of this (see 4.4.2). There is
no ‘right or wrong’ conjunct form, where legitimate differences (i.e. variant
forms) exist, as the following examples illustrate:
 The writing system 71

क् k+त ta = क्त kta e.g. भक्ति bhakti (devotion)

क् k + त ta = क्त kta e.g. भक्ति bhakti
As regards the two ways of representing the conjunct kta in bhakti, it
is important to note that the first example contains a unique form, which is
to say that it is not the product of the removal of the vertical stroke or of
a stacking strategy. For that matter, the second example neither removes a
vertical stroke (ka does not have a right-­hand vertical stroke to remove), nor
does it adopt a stacking strategy, yet the body shapes of ka and ta remain
clearly discernible. Both examples are licit. The same is not, however, true
with the conjunct kṣa, for which there is only one licit form.

क् k + ष ṣa = क्ष kṣa e.g. मोक्ष mokṣa (liberation)

Here, the option does not exist to clip the right-­hand stroke of ka and to
attach the modified body shape to the left of the following letter.
There are four further unique forms which the learner is strongly advised
to memorize alongside the basic forms of the Devanāgarī, given that they
are not subject to an analysis according to the strategies investigated. These
conjunct forms have a high frequency:

ज् j + ञ ña = ज्ञ jña e.g. ज्ञान jñāna (knowledge; thinking)

द् d + य ya = द्य dya e.g. विद्या vidyā (knowledge; learning)
त् t + त ta = त्त tta e.g. चित्त citta (noticed)
(Contrast this with क्त kta, with which it is easily confused)

त् t + र ra = त्र tra e.g. कु त्र kutra (where?)

(Contrast this with न्न nna, with which it is easily confused)

4.3.6 Multiple conjuncts

Conjuncts are not restricted to two elements but may contain three or four.
Words containing three conjunct elements are by no means rare, although
72 The writing system

four-­element conjuncts are substantially less common. The maximum num-

ber of conjunct elements in a word is five – but such a thing is extremely
rare. The forms which these conjuncts take are not any more complex than
those present in the two-­element conjuncts. Quite simply, a multiple con-
junct works in precisely the same way as a two-­element conjunct as regards
the strategies employed in its formation.
With three-­element conjuncts, the first and second elements generate the same
form as the two-­element conjunct. The second element, in turn, joins with the third
in precisely the same manner as one would expect with a two-­element conjunct.

त् t + स् s = त्स् ts + न na = त्स्न tsna

With tsna, the process of creating the conjunct is straightforward. The
letter ta possesses a right-­hand vertical stroke which is removed, with the
remaining body shape merging with that of the letter sa. The letter sa also
possesses a right-­hand vertical stroke, and this is similarly removed when
it conjoins with na. In the example given, the horizontal stroke of the body
shape of na rotates anticlockwise by 45°. Such a phenomenon has already
been encountered. The conjunct exhibits minimal difference with font vari-
ation (The font in the example which follows is Mangal).

त t् + स s् = त्स्ts + न na = त्स्न tsna

Conjunct forms which adopt the strategy of stacking continue to do so in
multiple conjuncts, as is evident in the following example (Note the rotation
of the va by 45°, using the tail of da as a pseudo-­vertical stroke).

द् d + ध् dh = द्ध् ddh + व va = द्ध्व ddhva

Not all fonts are currently capable of generating three-­element conjuncts.
In such cases, virāma is deployed (The font is Nirmala).

द् d + ध dh् = द्ध् ddh + व va = द्ध्व ddhva

As regards unique conjunct forms, these too are represented in multiple
conjuncts in precisely the same way as with two-­element conjuncts, as the
following example illustrates:
 The writing system 73

क् k+ त् t = क्त् kt + व va = क्त्व ktva

As explained in 4.3.5, where a variant exists for a two-­element conjunct,
that is the form which will be represented in a multiple conjunct (The font
is Nirmala).

क् k+ त् t = त् kt + व va =  ktva

Four-­element conjuncts act no differently to three-­element conjuncts
in terms of the procedure followed. The first element joins with the sec-
ond, producing the relevant conjunct; the second element then joins with
the third, and so forth. The only difficulty likely to be encountered by the
learner is when a unique form occurs between the second and third ele-
ments, or between the third and fourth. The approach here is to bear in
mind that unique forms will be produced wherever the letters that generate
them occur in the relevant sequence. In the following example, the first and
second elements (ṅ + k) create a conjunct form which then undergoes sub-
sequent transformation, since the second and third elements (k + ṣ) generate
a unique form:

ङ् ṅ + क् k= ङ्क् ṅk + ष् ṣ = ङ्क्ष् ṅkṣ + व va = ङ्क्ष्व ṅkṣva

Finally, we come to a five-­element conjunct which, whilst alarming as
a concept, is remarkably straightforward in design. It is unsurprising that
the conjunct in question contains a semivowel (two, in fact), since that
is something which one may regularly expect with conjuncts at the point
where more than three elements are involved. Sanskrit does not favour com-
plex clusters any more than English does. In the word kārtsnya (totality;
entirety), there is a cluster comprising three elements (-­tsn-­) surrounded by
two semivowels (r-­; -­y).

र् r + त् t + स् s + न् n + य ya = र्त्स्न्य rtsnya
All the consonants contain right-­hand vertical strokes and accord
with the strategy in 4.3.2. As for the phoneme r-­, it is the first element
in the conjunct and has the form as discussed in 4.3.4, over a vertical
stroke.
74 The writing system

4.4 Variant forms

4.4.1 Orthographic variants

Before returning to the issue of variation within the Devanāgarī, according
to the font used, it is worth looking at letter shapes frequently encountered
in Sanskrit texts. These letter shapes are occasionally referred to as ‘older
forms’ and, certainly, they are the forms which appear in Sanskrit manu-
scripts. Some textbooks use the forms in question, although those which
do are increasingly in the minority. These forms are worth learning from
the outset, to be saved a surprise when coming across them at a subsequent
point in one’s study of Sanskrit.
The short a has a more intricate shape in the older form (OF) than the one
employed by the Sanskrit Text font (ST). In essence, it rather resembles the
letter pa, with three strokes attached to the left-­hand side of the body shape.
If a text uses this form of the letter a, it follows that the initial forms for ā,
o and au also contain the form in question. In addition to initial a, ā, o and
au, there are also two consonants, jha and ṇa, possessing older forms. It is a
moot point as to whether r̥ and r̥̄ have older forms or whether the differences,
as they appear in certain texts, are merely an example of font variation. In
any event, being able to recognize the Devanāgarī letters for those vowels
in the font used in the present work allows the reader to identify the variant
forms for r̥ and r̥̄. On that basis, the orthographic variants for r̥ and r̥̄ are not
illustrated.

OF ST IAST

अ a

OF ST IAST

झ jha

OF ST IAST

ण ṇa
 The writing system 75

4.4.2 Font variation

There are many fonts available for the Devanāgarī. Two of these (Mangal
and Nirmala) appeared in 4.3.6. In principle, a good grasp of the Devanāgarī
gained in one font ought to suffice with respect to helping the eye adjust to
reading other fonts. Those who learn English, whose first language does not
use the Roman alphabet, are faced with a considerable number of variant
forms, as the following indicates with just two letters (The names of the
fonts are given in brackets).

a(Viner Hand) a (Script MT) a (Calibri)

g (Viner Hand) g (Script MT) g (Calibri)
It is unsurprising, given the examples from the Roman alphabet, that
there is variation between fonts available for the Devanāgarī. The varia-
tion is considerably reduced when one considers that the Devanāgarī does
not have the distinction, unlike the Roman alphabet, of uppercase and low-
ercase letters (It is worth noting that Latin did not originally possess such
a distinction – as was the case with ancient Greek also). That said, font
variation in the Devanāgarī can become slightly problematic where con-
juncts are concerned. As can be seen with the following examples, con-
junct formation strategies can vary according to the font used. Note how
both Mangal and Aparajita opt for the removal of the vertical stroke of
the first conjunct element whilst Sanskrit Text uses the stacking strategy:

Mangal Sanskrit Text IAST

न न्न nna

Aparajita Sanskrit Text IAST

ठ ष्ठ ṣṭha

Given that the learner of Sanskrit is increasingly more likely to want

to type Sanskrit than to write it, helped by the existence of the requisite
software, it pays to think carefully about font selection. Nirmala appears to
76 The writing system

be popular and it is certainly less ornate than some other fonts, although it
represents conjuncts as they appear in Hindi rather than Sanskrit: the stack-
ing strategy is avoided in favour of one which places the conjunct elements
side by side, even where the body shape of the letter for the first element
does not have a final right-­hand vertical stroke or where a unique form is
attested in the Devanāgarī.

Nirmala Sanskrit Text IAST

क क्क kka

Nirmala Sanskrit Text IAST

त क्त kta

The choice of font is entirely up to the student, who is free to choose

whichever font represents the easiest way to distinguish the basic let-
ter shapes of the Devanāgarī. In 4.4.3, various fonts are demonstrated.
This ought to be sufficient to allow the learner to decide which font is
the most suitable. Fonts such as Aparajita are ornate and move away
from anything resembling Devanāgarī in its written form. For that point
alone, the learner is advised to refrain from making Aparajita the font
of choice.
In selecting a font, it is important to consider that much of the printed
material for Sanskrit texts, especially if it comes from India, is highly
unlikely to use a font resembling Aparajita or Mangal. Both are indicative
of fonts used for Hindi in newspapers, advertisements, shop signs and
banners. Most of the Sanskrit literature printed by India’s most promi-
nent provider of such, Motilal Banarsidass, uses not only the older forms
illustrated in 4.4.1 but a font which closely resembles Sanskrit Text. From
that perspective, making Sanskrit Text one’s ‘go to’ font has considerable
merit.

4.4.3 Sample text in various fonts

The learner is encouraged to accustom the eye to differences that exist
between some of the more frequently used Devanāgarī fonts and to put into
 The writing system 77

practice the ability to identify the Devanāgarī letters, including conjunct

forms. The IAST version accompanies a text in various fonts, allowing the
learner to match the Devanāgarī to its transliteration.
The Sanskrit text is the opening verse of the Bhagavadgītā, in which
Dhr̥ tarāṣṭra, head of the Kaurava clan, addresses his minister, Saṃjaya.
Dhr̥ tarāṣṭra asks Saṃjaya to describe the field of combat on which the armies
of the Kauravas and the Pāṇḍavas are assembling on the eve of war. This is
the great war which forms the basis of the Indian epic, the Mahābhārata (A
translation of the verse is given in 9.3.1).

Sanskrit Text

धर्मक्षेत्रे कु रुक्षेत्रे समवेता युयुत्सवः ।

मामकाः पाण्डवाश्चैव किमकु र्वत सं जय ॥
Nirmala

धर्मक्षेत्रे कुरुक्षेत्रे समवेता युयुत्सवः ।

मामकाः पाण्डवाश्चैव किमकुर्वत संजय ॥
Mangal

धर्मक्त्
षे रे कुरुक्षेत्रे समवेता यय
ु तु ्सवः ।
मामकाः पाण्डवाश्चैव किमकुर्वत संजय ॥
Aparajita

धर्म क्षेत्रे कुरुक्षेत्रे समवेता युयुत्सवः ।

मामकाः पाण्डवाश्चैव किमकुर्वत संजय ॥
IAST
dharmakṣetre kurukṣetre samavetā yuyutsavaḥ
māmakāḥ pāṇḍavāś|caiva1 kim|akurvata2 saṃjaya
78 The writing system

Notes
1 What appears to be one word in the Devanāgarī is actually a combina-
tion of three: pāṇḍavāḥ + ca + eva. The visarga (ḥ), when followed by
c, transforms to ś. This is a part of a regular process known as visarga
sandhi, which will be investigated in 8.2. Since ś and c are then in con-
tact, with no intervening vowel, a conjunct is produced. The conjunct
causes pāṇḍavāś + ca to be written together (Remember that virāma
can only legitimately be deployed right at the end of a sentence). Fur-
ther, the words ca + eva cannot be placed together, since *ae is not
a diphthong recognized in Sanskrit. Here, vowel sandhi is applied,
replacing the impermissible *ae with a diphthong that is permitted: ai
(see 8.4).
2 Bearing in mind that the Devanāgarī does not permit the use of virāma
unless at the end of a sentence, the upshot is that a word which ends
with m will be subject to one or other of the following: (i) if followed
by a consonant, it will become anusvāra (see 3.5.1); (ii) if followed
by a vowel, the vowel in question will take the medial form and the
two words will then be written together (see 4.2.2). The latter situation
occurs with kim + akurvata. Remember that short a does not have a
medial form.

There is agreement, in all four Devanāgarī fonts, as to the unique

conjunct form for kṣa. In addition, the strategies for conjuncts involv-
ing r-­ (dharmakṣetre; akurvata) and -­r (dharmakṣetre; kurukṣetre)
are also consistent across fonts. Furthermore, the conjuncts ts
(yuyutsavaḥ) and ṇḍ (pāṇḍavāś) show no differences in terms of strat-
egy, although the elements t-­ and -­s are not in actual contact with each
other in Mangal and Aparajita. Look carefully, however, at the conjunct
śc (pāṇḍavāśca). Here, Sanskrit Text, Nirmala and Aparajita adopt the
stacking strategy, whilst Mangal places the letter ś to the left of ca
rather than above it. In Mangal, the ś does not have the special, ribbon-­
like form it often has when it is the initial element of a conjunct. The
right-­hand vertical stroke of ś has been removed, pure and simple. This
strategy is perfectly licit, in that other conjunct forms allow for the
removal of the right-­hand vertical stroke of ś, as seen with śma and śya
in 4.3.5.
 The writing system 79

4.4.4 Other writing systems

There are many writing systems in India – many more than is the case in
Europe, where the overwhelming majority of languages are written in the
Roman alphabet. Diacritical marks are easily added to the letters of the
Roman alphabet to indicate sounds specific to a given language and for
which a basic letter shape is not deemed sufficient. French, for example,
regularly employs three diacritical marks over the letter ‘e’ (è, é, ê) and
possesses a much rarer fourth (ë, as in Citroën). For certain languages of
Eastern Europe, the Cyrillic script is (or has been) used. This is the script
generally associated with Russian. There is only one language that uses
neither the Roman alphabet nor Cyrillic: Greek.
The proliferation of different scripts in India is notably different to the
situation in Europe. This is in no small part because the printing press came
late to South Asia. It was established by the British in the 1780s, in Calcutta,
and intended for English. One has, however, to consider the other forces
at work which delayed widespread literacy and its associated practices.
Sanskrit occupied a position of cultural preeminence amongst the Hindu
population, and the language tradition was heavily identified with Brahmins
amongst whom oral transmission was the norm. Also, the dominant power
in South Asia, until the British challenged it, was the Moghul Empire. For
Moghul rulers, literacy practices tended strongly to favour Persian, where
manuscript traditions prevailed.
Notwithstanding the dominance of Sanskrit, Persian and English dur-
ing the early modern period, there nevertheless existed a literacy tradition
amongst the vernacular languages of India (The Tamil tradition, in the south
of India, is one such example and an ancient one). In the north of India, the
Bhakti movement had provided a strong incentive for the writing of ver-
nacular languages, in sharp distinction to the oracy practices of Sanskrit, on
the one hand, and the high culture which found expression in manuscripts
written in a non-­Indian language, on the other. Vernacular languages were
not in any meaningful sense subject to the type of prescriptivism associated
with language planning. These languages were left to develop organically,
and a diversity in both language and script was the natural outcome. In
the following, a small sample is given of five phonemes as they appear
in the chief writing systems of contemporary North India. A comparison
with the Devanāgarī is useful, since it sheds light on how Devanāgarī letters
can be formed.
80 The writing system

Other writing systems

Devanāgarī Panjabi Gujarati Oriya Bengali

ग ਗ ગ ଗ গ
ga

Devanāgarī Panjabi Gujarati Oriya Bengali

थ ਥ થ ଥ থ
tha

Devanāgarī Panjabi Gujarati Oriya Bengali

न ਨ ન ନ ন
na

Devanāgarī Panjabi Gujarati Oriya Bengali

य ਯ ય ଯ য
ya

Devanāgarī Panjabi Gujarati Oriya Bengali

ल ਲ લ ଲ ল
la

Note
The Devanāgarī writing system is one amongst many in India. The other
major writing systems of northern India are those associated with Panjabi,
Gujarati, Oriya, Bengali, Assamese and Urdu – all official languages, as
per the Indian Constitution. Bengali and Assamese share the same writing
system. Urdu is distinct in that it is not descended from Brāhmī but uses the
Perso-­Arabic script.
 The writing system 81

4.5 Writing the Devanāgarī

Most textbooks on Sanskrit begin by introducing the Devanāgarī from
the first chapter or lesson, having dedicated a few pages in the prelimi-
nary chapters to letter formation. Whilst this has the virtue of obliging the
learner to work on Sanskrit in the Devanāgarī, it can take learners a number
of months to adjust fully to the writing system. The result is that lessons
can, for many, be extremely challenging, in that the learner is faced with a
grammatically complex language presented in a script which has yet to fall
into place. Conjuncts frustrate matters, leaving some learners with the over-
riding sensation that Sanskrit is simply beyond them, at which point they
disengage with their study of the language.
Letter formation is important for those who intend to write Sanskrit man-
ually, although it is increasingly the case that learners seek out the requisite
software which allows them to generate the Devanāgarī electronically. On
the basis that it is useful to be able to practise the Devanāgarī by hand, a
few pages will be devoted to illustrating how the letters are formed. As with
the writing of any script, it is important not to be overly prescriptive. There
are only a couple of principles to remember, when writing the Devanāgarī.
Once these have been accepted, whatever strategy the learner employs to
form the letters is acceptable. The key point is to make the letters distinct,
where similarities exist, so that ambiguity does not arise.
Some of the other major writing systems for the Indic languages, particu-
larly Gujarati, offer insights into letter formation. Gujarati script differs from
the Devanāgarī in one crucial aspect: it does not possess the horizontal top
line, the mātrā (measure), which is a hallmark of the Devanāgarī. In essence,
the Gujarati script is derived from the writing practices of merchants who
devised a script which could be written hastily. A glance at the Devanāgarī
and Gujarati letters for ga shows that the Devanāgarī has two vertical strokes
hanging down from the mātrā, whereas Gujarati has two strokes and no
mātrā. In the Gujarati letter, the left-­hand shape, resembling a semicircle, is
written first. One goes from top to bottom, clockwise. Having formed the first
element, the right-­hand element is then added. This is precisely the case with
the Devanāgarī, which then completes the letter by adding the mātrā, from
left to right. These are the guiding principles in writing the Devanāgarī: top to
bottom, left to right, with the mātrā written as the final stroke.
The principles of letter formation can be explored very easily by look-
ing at the letters for consonants in the first two series (velar and palatal).
These contain all the information necessary to write the other consonants,
semivowels, sibilants and vowels. The principles apply in all cases. The
first letter in the velar series represents the syllable ka, which requires four
82 The writing system

strokes (With the exception of initial ā, r̥, r̥̄ and o, all the Devanāgarī letters
are formed by three, four or five strokes). The letter ka is a good model with
which to explore the principles in question.
The letter ka is formed by the following four strokes, in sequence: (i)
the leftmost element of the body shape of the letter, formed anticlockwise;
(ii) the middle element of the body shape, from top to bottom; (iii) the
right-­hand element of the body shape, from left to right (where the pen
does not have to be lifted after the second stroke but can travel upwards
to the middle of the letter and across to the right); (iv) the mātrā, written
from left to right.

If we look at the first series of consonants (the velars), bearing in mind

the principles applied with the formation of ka, letter formation becomes
straightforward (The formation is given from top to bottom rather than, as
with the example of ka, left to right, almost by way of reinforcing the prin-
ciples underlying letter formation).

ka kha ga gha ṅa

    ड
क ख ग घ ङ
क ख ग घ ङ
As noted earlier, the letter ga is formed with three strokes, as opposed
to the four needed for ka (and indeed gha and ṅa). That said, the formation
of the velars can be shown in three phases: the initial stroke or strokes –
proceeding always from left to right – terminating with the mātrā, similarly
written from left to right. Note that, with kha, there is a sickle shape to the
left of the circle which one finds in ka. Since that is the leftmost element, it
is written first. As for ṅa, there is a dot to the right of the body shape of the
letter. This is written after the body shape and before the mātrā.
The same principles of letter formation are applied to consonants from
the second series (the palatals). The leftmost element is written first, fol-
lowed by the vertical stroke – preceded, in the case of jha and ña, by a
 The writing system 83

connecting stroke. The mātrā is written last. The only letter which merits
special attention is cha, where there is no vertical stroke in the letter but,
instead, a short neck above the loop of the body shape. This is added after
the body shape and before the mātrā.

ca cha ja jha ña

    
च छ ज झ ञ
च छ ज झ ञ
If the learner can keep in mind the principles demonstrated, there is no
letter for which formation remains unclear. The dentals ta and na occa-
sionally elicit questions, in that the leftmost element in the body shapes of
these letters requires a stroke which travels upwards. This is true but one
may, with both ta and na, give priority to the principle of proceeding from
top to bottom (with the exception, naturally, of the mātrā, which must
remain the final stroke). If that is the strategy preferred by the learner,
the starting point in letter formation is the centre of the body shape, going
from right to left to form the ‘leg’ of the ta or the notched horizontal line
of the na, and proceeding from top to bottom in the process. Practice
makes perfect and the point is perhaps best iterated that a prescriptive
approach to the formation of the letters of the Devanāgarī serves no pur-
pose at all.

4.6 Exercises

4.6.1 Identify the consonant

The reader may now consolidate the current chapter by attempting two
exercises. The first exercise relates to the consonants identified in 3.2, since
the knowledge of the first twenty-­five consonants involves, aside from
the ability to pronounce them, an awareness of whether they are voiced
or voiceless, aspirated or unaspirated (This is relevant to Chapter 8, where
sound changes are explored). The answers are given on the same page, for
convenience’s sake.
84 The writing system

Questions
Identify the following consonants according to their description.

Labial Velar Dental Palatal Cerebral

– V, +A – V, -­A +V, +A Nasal +V, +A

Labial Cerebral Dental Palatal Velar

Nasal – V, -­A Nasal – V, +A Nasal

Labial Palatal Labial Cerebral Velar

+V, -­A – V, -­A +V, +A – V, +A +V, +A

Velar Palatal Dental Labial Cerebral

– V, +A +V, +A – V, -­A – V, -­A +V, -­A

Velar Dental Cerebral Palatal Dental

+V, -­A +V, -­A Nasal +V, -­A – V, +A

Answers

Labial Velar Dental Palatal Cerebral

– V, +A – V, -­A +V, +A Nasal +V, +A
फ pha क ka ध dha ञ ña ढ ḍha
Labial Cerebral Dental Palatal Velar
Nasal – V, -­A Nasal – V, +A Nasal
म ma ट ṭa न na छ cha ङ ṅa
Labial Palatal Labial Cerebral Velar
+V, -­A – V, -­A +V, +A – V, +A +V, +A
ब ba च ca भ bha ठ ṭha घ gha
Velar Palatal Dental Labial Cerebral
– V, +A +V, +A – V, -­A – V, -­A +V, -­A
ख kha झ jha त ta प pa ड ḍa
Velar Dental Cerebral Palatal Dental
+V, -­A +V, -­A Nasal +V, -­A – V, +A
ग ga द da ण ṇa ज ja थ tha
 The writing system 85

4.6.2 Transliteration
A second exercise asks the reader to attempt transliteration from the
Devanāgarī to the IAST. The ability to read both systems accurately is
important, given that a good knowledge of the IAST indicates a good grasp
of the sound system of Sanskrit. Until the time when one is fully able to read
the Devanāgarī (and thereby dispense with the IAST), the IAST represents
the best means of accessing information about the language. The answers
are given on the following page.

Questions
Give the IAST for the following words:

(i) Words without conjuncts.

कनक
दे व
नृप
मानुष
रामायण
ऋषि
(ii) Words with conjuncts.

वेदान्त
अश्व
ग्राम
मार्ग
मोक्ष
सं स्कृत
86 The writing system

Answers
(i) Words without conjuncts.

कनक kanaka (gold)

दे व deva (god)

नृप nr̥pa (king)

मानुष mānuṣa (man)

रामायण rāmāyaṇa (the Rāmāyaṇa)

ऋषि r̥ṣi (a sage; rishi)

(i) Words with conjuncts.

वेदान्त vedānta (end of the Vedas)

अश्व aśva (horse)

ग्राम grāma (village)

मार्ग mārga (road)

मोक्ष mokṣa (liberation)

सं स्कृत saṃskr̥ta (Sanskrit)

This exercise allows the reader to identify any revision requirements. See
4.2.2 if the revision involves the representation of the vowels and 4.3 if a
reminder is needed as to conjunct formation. Note that the words rāmāyaṇa,
vedānta and saṃskr̥ta have not been spelled with an initial capital letter.
This accords with the more orthodox use of the IAST, in which capital let-
ters are unnecessary and therefore avoided. It is a minor point, in that the
capital letters do not affect the IAST one way or another.
Only once the reader is confident with the material covered up to this
point is progression to Chapter 5 advised.
5
THE SANSKRIT WORD

5.1 Preliminaries
In this chapter, the aim is to assist the learner to migrate from the investiga-
tion of individual sounds and syllables to the exploration of complete words
in Sanskrit. Although a few words were introduced in the previous chapters,
this was by way of consolidating the ability to read them. The tactic is now
to subject a short list of words to a greater analysis, looking at how they are
formed as well as what they mean. The intention is to multitask by doing
three things at once: building up a number of words that are likely to be
encountered early in the study of Sanskrit (thereby beginning the process
of acquiring vocabulary); exploring the primordial nature of the verbal root,
which lies at the heart of the Sanskrit word; and teasing out information
from the verbal root to see what may be deduced about the literal meaning
of the words in question.
A Sanskrit text can often be subject to a baffling number of translations
which appear to differ, one from the other, to the point where the reader
begins to wonder which translation (if indeed any of them) comes close to
capturing the original meaning. Sanskrit is subtle. It can embed layers of
meaning into a single word which, to any translator, represents a challenge.
It should always be borne in mind that Sanskrit is the product of an ancient
culture. There are all too often multiple choices in rendering a Sanskrit word
into English – but with each word failing to catch entirely the concept which
the word embodies. The Italian expression traduttore, traditore comes to

DOI: 10.4324/9780429325434-6
88 The Sanskrit word

mind: translator, traitor. In essence, translation from Sanskrit aims at an

approximation in meaning but, if concepts in Sanskrit do not directly map
onto these which may be expressed in the target language, something is
invariably lost.
In the following pages, twelve words are investigated. Each of these are
theoretically derived from a verbal root, of which more will be said in Chap-
ter 7, which looks at the Sanskrit verb and at conjugation. The verbal root
is indicated by an appropriate symbol (√). Each word is a noun and is given
according to its citation form – the form in which it appears in a Sanskrit
dictionary before any grammatical information has been added. The San-
skrit noun and its declension are dealt with in the following chapter. For
now, it suffices to see how much information may be extracted from the
Sanskrit word before grammar is even considered.

5.2 Analysis of words

(i) दे व deva (√ div)

The word deva (god) is an easy one to read. It contains an initial con-
sonant (d) above which the vowel e appears. This is nevertheless note-
worthy in that it shows the assertion to be false that, when reading
the Devanāgarī, one reads from the top to the bottom of a syllable.
Doing so would result in the reading of the first syllable as *ed. The
Devanāgarī syllable is correctly identified by reading the body shape of
the letter and by looking to see if anything is added above the topline,
below the body shape or to the right of it. The only exception to this is
the short medial i, which is written before the consonant, semivowel or
sibilant. Deva is derived from the verbal root div (to shine; be bright)
and is cognate with the Latin deus, Italian dio and French dieu. Eng-
lish, borrowing heavily from Latin (often via French), has words such
as divine and divinity, both of which can be connected with the word
deva. Whereas the word for the singular God (or one of many gods, in
a non-­monotheistic belief system) may not bring to mind a shining or
resplendent entity in other languages, it does so overtly in Sanskrit. A
deva is a resplendent being.
Philologists have reconstructed the possible form and meaning of
the word deva as it may have been uttered in the ancestor of the Indo-­
European languages (Sir William Jones’s ‘common source’). This
reconstructed form, named ‘Proto-­Indo-­European’ (PIE), contains the
root *dyew (sky; heaven), suggesting that the concept of a shining,
 The Sanskrit word 89

supernatural being was connected to the luminescence of the sky or

to daylight. Latin sheds some light on this – no pun intended – in the
word diēs (day). It is worth mentioning that the Latin Iuppiter (which
appears in English as Jupiter), albeit borrowed into Latin from an ear-
lier Italic language in which the dropping of initial d-­ was a feature,
allows philologists to hypothesize that the name for the Father of the
Gods, in the Roman pantheon, was Father of the Sky: *dyew patēr
(Note that variations in PIE reconstructions are almost as numerous as
the researchers engaged in such an activity. Since reconstructed forms
are not actually attested, in that PIE predates literary records, all forms
are routinely given with a preceding asterisk). What the Sanskrit deva
shows us is that the English words divine, deity and day are related, as
are the Latin and Greek Iuppiter and Zeus, respectively, and that there
is a clarity in Sanskrit with respect to its structure which permits such
insights.

(ii) कमल kamala (√ kam (?))

The Sanskrit word kamala is one of a number of words for the lotus.
Its symbolism is pervasive in Indic culture. It is a thing of beauty which
grows from the mud and is, therefore, a most suitable symbol for the tri-
umph of human consciousness, transcending both humble origins and the
chores of everyday existence. Little wonder, then, that the lotus is a sym-
bol associated strongly with the Buddha and with any number of devas
within Hinduism, frequently portrayed as accompanied by the lotus or,
indeed, rising from it. The noun kamala is neuter but, with the appropri-
ate adjustment to the ending, a feminine noun can be created (see the
following page and 6.1 as regards grammatical gender). Kamalā is not
an uncommon name in India. It is an epithet for the goddess Lakṣmī.
It may be that the word kamala is derived from the verbal root kam
(to love; long for), a possibility alluded to by Monier-­Williams, in list-
ing one of the meanings as desirous, lustful. This would be consistent
with the reddish pink of the lotus, a colour associated with blushing or
desire. If Monier-­Williams is correct, it could be argued that kamala
encodes the concept of worldly desire which the Buddha, being seated
on a lotus, has transcended. That said, the exercise of searching for the
etymology of a word must be taken with a pinch of salt. Etymology
can be – and frequently is – subject to assumptions engrained in popu-
lar tradition but for which a linguistic analysis offers only a tentative
suggestion. There are no PIE connections to speak of with the word
kamala, which is but one of many words in Sanskrit for a lotus.
90 The Sanskrit word

As regards the verbal root kam, it is found in several words within

Sanskrit and is at the core of the word kāma (as in Kāmasutra, the
ancient Indian treatise on love-­making). It is to be noted that the word
kāma is a noun (sensuality; desire) and that the vowel in the verbal
root has been mutated in the creation of the noun (kam → kāma).
This is a common feature in Sanskrit, as can be seen in the word vāda
(a saying; speech) from √ vad. In this manner, a considerable number
of nouns may be derived from the same verbal root. The mutation of
the root vowel is not always obvious, requiring the learner to investi-
gate and to be familiar with a phenomenon referred to as ‘vowel muta-
tion’, ‘vowel gradation’, ‘vowel strengthening’ or guṇa and vr̥ddhi.
This phenomenon is further explored in Chapter 7, since it is crucial
in the derivation of a stem from a verbal root, which is the first step in
conjugation.

(iii) विद्या vidyā (√ vid)

The spelling of the word vidyā is considerably more complex than that
of kamala in three respects: it contains the medial form of the vowel i
which is exceptional in being written before the phoneme with which
it forms a syllable; second, it contains a conjunct (dya) which falls
into the category of being a unique form (as described in 4.3.5); and
third, it contains the long vowel ā, where kamala only has the short
a, not indicated by the Devanāgarī unless it is in word-­initial position.
The word vidyā is derived from the verbal root vid (to know), cognate
with the German wissen (which has the same meaning) and glimpsed
in the English expression to be out of one’s wits. It is most likely cog-
nate with Latin vidēre (to see), given that seeing appears to be closely
linked with the concept of knowing. Consider, for example, how one
may respond, in English, to a statement containing information: ‘I see’
(i.e. certain information has been presented to me and I understand its
significance). Care must be taken not to conflate vid (to know) with
another verbal root which is spelled the same way and, consequently,
is pronounced the same way in Sanskrit. That is the verbal root vid (to
be found; exist). Sanskrit does not have many homophones – far fewer
than other languages – but it is still the case that there are words which
differ in meaning yet have the same pronunciation and spelling.
The existence of a final, long ā in the noun vidyā is ample proof that
not all noun stems end in a short a (A noun stem is the form which a
noun has before grammatical endings are applied. This is of prime con-
sideration in the process of declension, investigated in Chapter 6). The
 The Sanskrit word 91

words deva and kamala both end in a short a; but deva is referred to
as a ‘masculine short -­a stem’ and kamala as a ‘neuter short -­a stem’.
Before approaching Chapter 6, it is worth drawing to the reader’s atten-
tion that Sanskrit possesses grammatical gender. A noun is masculine
(deva), neuter (kamala) or feminine (vidyā). Only feminine nouns end
in -­ā and -­ī, although the situation is not as clear in cases where the
stem ending is -­a. Additionally, not all nouns end in -­a -­ā, or -­ī. It is
important for the Sanskrit student to approach the acquisition of vocab-
ulary with gender in mind. All nouns must be learnt together with their
gender. Whilst this is not an issue with vidyā, given that the long -­ā
identifies it as a feminine noun, short -­a can indicate either a masculine
or a neuter noun. Nouns are listed in the Index, along with their gender.

(iv) रात्री rātrī (√ ram (?))

The word rātrī is one of several Sanskrit words for night (Another,
which occurs frequently and appears in the current work, is niśā). It is
clearly not cognate with the English night, German nacht or Latin nox –
all of these identifiable with the PIE reconstruction *nókʷts. As with
kamala, the verbal root is speculative. Monier-­Williams’s Sanskrit-­
English Dictionary suggests √ ram (to rest) as ‘probable’ rather than
certain. Indeed, the learner of Sanskrit is urged to be cautious with
respect to assuming that all nouns may be derived from verbal roots.
The Sanskrit grammatical tradition demonstrates an exceptional acuity
as regards ascribing nouns to verbal roots, but that is not to say that
all nouns can be so derived or, indeed, that the Sanskrit grammatical
tradition claims this to be the case. Whatever the origin of the word
rātrī, the important point to note (as with kamala) is that Sanskrit has
more than one word for a phenomenon where English, or any number
of other languages, might simply have one. It comes as no surprise,
therefore, that Sanskrit vocabulary is extensive. With rātrī, it should be
noted that the Devanāgarī contains the medial form of long ā (as does
vidyā) and the medial form of long ī, which differs from the short i of
vidyā. There is also a conjunct (tra), which is similar in form to the
conjunct nna (see the final example given in 4.3.5).

(v) अग्नि agni (√ ag (?))

The word agni equates with the English fire and is cognate with Latin
ignis (Note the English words ignite; ignition). Although contested,
it is possible that it is derived from √ ag (to move tortuously) which
would mean that the root captures the concept of movement rather than
92 The Sanskrit word

the heating or burning qualities of fire. Such speculation is possible

because Sanskrit preserved, as has been noted, a list of the verbal ele-
ments from which a large element of the language was deemed to be
comprised. This list, the Dhātupāṭha, was transmitted orally, along
with the structural outline and rules of the language. It formed an inte-
gral part of a grammatical tradition of considerable antiquity which
culminated in Pāṇini, the author of the world’s first known systematic
grammar. Whilst Pāṇini’s dates are subject to much speculation, a sug-
gestion as to the period 500–350 bce is in order. The precise origin of
the Dhātupāṭha has not been established. It is often ascribed to Pāṇini
himself, although it seems more plausible that the concept of verbal
roots predates Pāṇini, as a result of the assiduous attention paid to the
language of Vedic ritual practice.
The word agni offers an insight into how there is not always agree-
ment within the Sanskritic tradition as to verbal roots. In antiquity, √
aj (to drive; propel) was proposed as the verbal root; this is indeed
possible but appears less plausible. Such a proposal would certainly
suggest a cognate with Latin agere (to drive; act), yet this would move
even further away from describing fire in terms either of its action or
its qualities. Another proposal suggests agri. As mentioned in Monier-­
Williams, agri was created to explain the etymology of agni and is not
a serious contender. Whatever the verbal root of agni, be it √ ag or √
aj, it is an important word in the Ṛgveda, the oldest literary work of
ancient India. There, it is the name of a deity which, along with Indra
and Soma, belongs to the early belief system of the people which his-
tory and tradition have termed ‘Aryan’ (in Sanskrit, ārya).
As regards spelling, agni contains the initial form of a and the
medial form of i, seen in vidyā). It is worth noting that medial i is writ-
ten immediately before the conjunct. It does not split the conjunct by
inserting itself between the elements comprising it. It is also worth not-
ing that the conjunct gna bears a striking resemblance to the conjunct
gra. Care must be taken not to conflate the forms.

ग्न gna ग्र gra

(vi) गुरु guru (√ gr̥̄)
The association in meaning between a verbal root and a noun derived
from it may not always be evident. This is the case with guru (teacher;
 The Sanskrit word 93

preceptor). The verbal root is √ gr̥̄ (to make known; teach), yet the
word guru defines someone who has substance or a thing which is
weighty (In Sanskrit, a long syllable is termed guru, for example, and
a short syllable is laghu: light). The notion of weightiness is a far cry
from the concept of teaching – unless one assumes that the act of teach-
ing involves the imparting of information which has substance. In that
case, it is possible to argue for a semantic connection between weight
and learning. The word gravitas lends support to the argument that a
guru is weighty not because of body fat but, rather, on account of the
learning acquired. This is not an outlandish argument, given that such
a connection exists in other Indo-­European languages. One speaks of
a person who has gravitas, which comes from the Latin gravis (heavy;
serious).
The spelling of guru in the Devanāgarī is straightforward. It
is disyllabic (gu + ru) and contains no conjuncts. That said, the
reader is asked to refer to the second syllable and to pay close
attention to the fact that the short medial u is not at the base of
the body shape for the letter ra. Both short u and long ū, when
medial (a term which also includes the representation of vowels
when they are word final), have a distinct form when they follow
the phoneme r:

रु ru रू rū
(vii) कर्म karma (√ kr̥)
The Sanskrit word karma is widely known by speakers of English
who have never studied Sanskrit, since it has entered English in such
expressions as: ‘That’s karma for you’ – an expression which in some
sense signifies that an action has met with its appropriate reward. In
Sanskrit, however, the word means nothing other than the sum of one’s
actions, from √ kr̥ (to make; fashion; do; create). There is no sense
that justice has been meted out. Karma is simply action, devoid of any
divine justice. The term has philosophic overtones, all the same. One’s
actions may be prescribed and proscribed according to what it is one
does or is supposed to do. ‘Bad karma’, in this context, represents a
course of action which is inappropriate for a person, all circumstances
being considered. If there is any judgement pursuant on one’s actions,
that is dharma.
94 The Sanskrit word

(viii) धर्म dharma (√ dhr̥)

Just as karma is one of the most widely known of Sanskrit words
amongst non-­Sanskritists, dharma is a prime contender with respect
to being one of the most debated as regards meaning. A great number
of words are candidates in the attempt to translate it. The word comes
from √ dhr̥ (to protect; preserve; restrain). Notions of dharma as a
form of judgement do not entirely capture the concept, since the uni-
verse is dispassionate; but to translate the word as law – not a human
law but one which is dispassionate as, for example, the laws of nature
or those of physics – similarly fails to capture the concept. Dharma
is the force which keeps the universe balanced as regards righteous-
ness. On a more mundane note, the verbal root can be used to signify
nothing other than protection or preservation, as with the name of the
blind king in the Mahābhārata, Dhr̥tarāṣṭra: the one who protects
(dhr̥ta) the kingdom (rāṣṭra).

(ix) मोक्ष mokṣa (√ muc)

The concept of the liberation of the Self – the ātman – is funda-
mental to Hinduism, where the union of the Self (often couched in
Judaeo-­Christian terms as the soul) with the universe ensures that
one’s consciousness becomes part of the infinite. For Hindus, it is a
consummation, as expressed by Hamlet, devoutly to be wished. It is
the great liberation (mokṣa) from the cycle of birth, death and rebirth.
The verbal root could not be clearer and is not subject to any disagree-
ment. It is √ muc (to let loose; release). The reader might wonder at
the fact that the verbal root contains the phoneme c, which is neither
audible nor visible in the word mokṣa; or, for that matter, that mokṣa
contains o whereas the root does not. In both cases, Sanskrit phonol-
ogy provides the answer.
It has already been noted that, when verbal roots are used to gener-
ate nouns, vowel mutation often occurs (Chapter 7 explains this in
more detail). In a nutshell, vowel mutation (guṇa) results in the fol-
lowing changes:

Simple vowel a or ā i or ī u or ū r̥ or r̥̄ l̥

Mutation a e o ar al
 The Sanskrit word 95

Whilst these changes are not attested in √ vid → vidyā, √ ram →

rātrī and √ gr̥̄ → guru, they predict √ div → deva, √ kam → kamala
and √ ag → agni (Admittedly, mutation does not result in a visible
change, where a verbal root contains a). Mutation is, however, visible
with √ kr̥ → karma and dhr̥ → dharma. It is to be noted that √ vid →
veda does indeed accord with the mutations presented, as does √ gr̥̄ →
garīyas (heavier). With the word buddha (the next word to be inves-
tigated), the verbal root is √ budh. The derivation of the noun buddha
does not accord with the information in the table; but it does so with the
word bodhisattva (one who helps others attain nirvāṇa).
Returning to √ muc, one may predict that, if the root is subject to
any mutation in the derivation of a noun, the mutation would result
in *moc. Sanskrit phonology, at this stage, then transforms the final c
to kṣ, following a complex sound change, predicated around the fact
that a fully formed word cannot end in -­c. The result is neither *moca
nor *mok but a hybrid of both which preserves the fricative sh-­sound
inherent in the palatal c: mokṣa.

(x) बुद्ध buddha (√ budh)

The word buddha connects linguistically with the contemporary con-
cept of being ‘woke’ – a term which, whilst current at the time of writ-
ing, has yet to show whether it will stand the test of time. Buddha is an
epithet for the person born Siddhārtha Gautama, whose birthplace is
generally accepted to have been Lumbini, in the south of Nepal. There
can be no doubt as to the fact that the teachings of the Buddha have
endured, having spread out from northern India to what is now Afghan-
istan and central Asia and, courtesy of the Silk Road, to Tibet, Mongo-
lia, China, Korea and Japan. Buddhism also expanded further south, to
Sri Lanka, Burma, Thailand, Laos, Cambodia, Vietnam and Indonesia.
Within the vast area in which Buddhist thought became known, San-
skrit and the related Pali were the languages in which such thought
found expression.
Whereas to be ‘woke’ in the twenty-­first century is to be aware of
social injustice, the Buddha was awoken to the suffering of existence
and how such suffering could be ended. His epithet does not call him a
saviour or redeemer but comes from the verbal root budh (to awaken),
an everyday word with no overtly spiritual meaning. The difference
between √ budh and buddha is an interesting one and sheds light
96 The Sanskrit word

on the pervasiveness of sound changes in Sanskrit. Buddha may be

termed a proper noun, since it is a name, albeit an epithet. It is also,
technically speaking, a past passive participle, which usually take the
form of a suffix -­ta after the verbal root (More will be said about the
past passive participle in 7.7, viii and 9.3.1). Siddhārtha Gautama is
not, however, referred to as the *Budhta, so something has clearly hap-
pened to alter the pronunciation. The explanation is twofold.
When elements which constitute a word (i.e. morphemes: units of
meaning) come together in Sanskrit, they are often subject to a sound
change. As seen with the sound system (Chapter 3), most consonants
are classified as voiceless or voiced, unaspirated or aspirated (Nasals
and semivowels are voiced; sibilants are unvoiced. Aspiration is not
a feature of nasals, semivowels or sibilants). The subject of sound
changes within Sanskrit is the source of the greatest frustration for
learners of the language in their first few months of study. Fuller inves-
tigation of sound changes is the subject of Chapter 8. For now, it suf-
fices to note that, in the bringing together of √ budh and the participle
-­ta, the following changes take place:

budh (where dh is voiced) + -­ta (where t is voiceless)

First step: t changes to its voiced equivalent, d.
Result: *budh+da

So far, so good, but *budhda is still not the final form. The aspirate (h)
is now trapped, in that it cannot be released without a vowel to assist it.

Second step: displace the aspirate so that it comes before a vowel.

Result: buddha

Fortunately, not all Sanskrit words undergo such complex sound

changes. The learner should, all the same, be linguistically ‘woke’ to
the fact that sounds changes are an extremely important feature of the
language. When morphemes come together, voice (and often aspira-
tion) can trigger a change in sound and spelling.

(xi) निर्वाण nirvāṇa (nir-­+ √ vā)

Nirvāṇa is the final destination of the Self for Buddhists, just as mokṣa
represents the goal in Hinduism. It is derived from √ vā (to be blown out;
extinguished; made calm). The reader will notice that something comes
 The Sanskrit word 97

before vā; this is a prefix, as opposed to -­ta, which is a suffix. Prefixes

are common in Sanskrit, but it is well beyond the scope of the present
work to discuss them. Prefixes can either modify the concept contained
in the verbal root – sometimes a little, often quite considerably – or have
no real effect on the root meaning. Only exposure to Sanskrit vocabu-
lary will clarify the extent to which prefixes can modify meaning. For
the learner who is new to Sanskrit, the study of prefixes is best put to
one side until more basic concepts have been grasped.
The word nirvāṇa contains a conjunct in which r is the first element
(rv). Accordingly, r takes the form of a crest, written above a vertical
stroke. The vertical stroke is not that of the body shape of the letter va
but, instead, the following vertical stroke which represents the medial
form of ā. The crest of the r prefers not to share a vertical belonging to
the body shape of a letter. If a medial ā is present, the crest will appear
above it.

(xii) उपनिषद् upaniṣad (upa-­ + ni-­ + √ sad. Note that sadhu is not
derived from √ sad but from √ sādh: to attain a goal)
The Upaniṣads (frequently appearing in the English spelling Upani-
shads) constitute a culturally important body of texts relating to specu-
lative thought in which the nature of the ultimate reality and the Self
are the focus. They are post-­Vedic works, in that they come after the
Ṛgveda in terms of their composition. As such, the term Vedānta (end
of the Vedas) is often applied to them. As with the word Buddha, the
verbal root does not imply anything spiritual. Here, the root means
nothing other than to sit; be seated (√ sad). As with nirvāṇa, there is
a prefix – two of them, in fact: upa-­ and ni-­. Between them, the pre-
fixes serve to modify the verbal root by giving some information on
the state of sitting, which is defined as being close to (upa-­) and below
(ni-­) something or someone. This is highly descriptive of the traditional
social setting in which the Upaniṣads were transmitted in antiquity:
the guru would be seated, close enough to his students to be audible to
them, and the students would be seated below him, showing the requi-
site level of respect.
There is a sound change, from sad to ṣad, but this is not of con-
cern for the moment. It is not a sound change triggered by voice or
aspiration. In terms of its spelling, the Devanāgarī is straightforward:
no conjuncts and no conventions to consider, such as the fondness of
the crested r (in conjunct state) for a particular type of vertical stroke.
98 The Sanskrit word

Attention should be paid to the virāma, at the end of the word, which
ensures that the pronunciation is not *upaniṣada. There is, however, a
small point which is nevertheless worth raising. Just as Sanskrit does
not permit c in word-­final position, as we saw with mokṣa, so too d is
not a permitted final. The term ‘permitted final’ relates to a number of
phonemes with which any given word may end. Vowels are all permit-
ted finals but, as concerns the consonants, semivowels and sibilants,
only the following are technically allowed: k, ṭ, t, p, ṅ, n, m, r (Strictly
speaking, therefore, the word upaniṣad, if occurring by itself or fol-
lowed by a word beginning with a voiceless phoneme, is upaniṣat).
Anusvāra and visarga are also permitted word-­finally. On that note, fair
notice should be given to the reader that visarga is subject to a system-
atic number of sound changes. These will be investigated in Chapter 8.
The following section will focus on Sanskrit grammar, on which matter
a few preliminary comments are highly desirable.

5.3 Approaching Sanskrit grammar

Whatever one’s knowledge of complex grammatical systems and however
well one may know an Indic language or one such as French (which has
grammatical gender) or German (which has both grammatical gender and
case), Sanskrit presents the learner with ample challenges. It is, as Sir Wil-
liam Jones aptly remarked, “more perfect than the Greek, more copious
than the Latin”. Examples have been used, in the earlier chapters of the
present work, to illustrate cognates with Sanskrit. Often, the example was
drawn from Latin. This is not to say that Latin, or indeed any other lan-
guage, is necessary for the study of Sanskrit. Familiarity with grammar can
be acquired from scratch. It does not have to be applied from an existing
knowledge, gained by the formal acquisition of another language. For the
learner for whom grammatical terminology is a thing of mystery, a glossary
of terms is provided by the present work. The reader who feels slightly out
of his or her comfort zone is strongly advised to consult the glossary as and
when required.
A good grasp of the sound system is crucial to the study of Sanskrit.
Words are formed and are subject to change in Sanskrit according to the
principles of the sound system. To ignore the differences between voiced
and voiceless sounds is to court disaster; but the voiced or voiceless nature
of phonemes is just one of many phenomena of which the Sanskrit stu-
dent should be aware. The key sound changes will be explored (Chapter 8).
First, however, must come something which allows one to see what Sanskrit
 The Sanskrit word 99

grammar looks like, beginning with the noun (Chapter 6) and moving on to
the verb (Chapter 7). The Sanskrit noun is masculine, neuter or feminine,
which means that the word for any given thing has to be learnt along with
its gender. That is just the starting point. As for the verb, arguably the most
complex component of Sanskrit grammar, gender is not an issue, but the
Sanskrit verb is composed of ten classes, each class exhibiting a different
pattern, either in how the verbal root generates a stem on which the personal
endings (I, you, he/she/it, etc.) can be added, or in the way in which the verb
is conjugated. The following two chapters of the present work are dedicated
to presenting the Sanskrit noun and the Sanskrit verb as simply as possible,
giving the reader the confidence to continue with his or her study of what
is a challenging but immensely rewarding language. One should not rush
into Sanskrit grammar. Time taken to grasp the basic principles is time very
well spent.
6
THE SANSKRIT NOUN
Declension

6.1 Preliminaries
In the previous chapter, a dozen words were investigated, allowing for a
discussion on various things which they were able to reveal, both about
the culture with which Sanskrit is associated and as regards certain purely
linguistic matters, including the Devanāgarī writing system. The words in
question were all nouns. They related to things or to people, whether tangi-
ble (such as kamala, guru, buddha) or abstract (vidyā, karma, dharma).
They did not, however, all have the same endings. Whilst several of the
words ended in -­a, there were some for which the final phoneme was
another short vowel (-­i, -­u), a long vowel (-­ā, -­ī) or a consonant (-­d/ -­t).
These endings are important, since they indicate something about the noun
which is central to the process of declension, where a noun takes various
endings according to the grammatical information it contains.
In English, one distinguishes a difference between tree and trees. The
former indicates a singular tree; the latter, two or more. The distinction,
grammatically speaking, is that tree appears in the singular number whilst
trees indicates the plural number. The term ‘number’ is as simple as that, in
English: it identifies whether something is in the singular or in the plural.
The ending -­s is the phoneme in English which distinguishes tree from trees.
As such, linguists and grammarians refer to it as the ‘plural morpheme’. It is
a unit of meaning which can be added to a word to contribute to its overall
sense. English is not consistent in its application of the plural morpheme.

DOI: 10.4324/9780429325434-6
 The Sanskrit noun 101

Whilst a singular cat is made plural by its addition (cats), as too with dog
and dogs, the plural morpheme is not applied to the end of the word mouse
(i.e. the plural is mice, not *mouses).
Sanskrit possesses not only the singular and plural numbers but also the
dual number, which indicates two (and only two) of something. The dual
is an ancient feature in Indo-­European languages. Latin did not use it and
Greek contained a few examples but had largely abandoned it by the classi-
cal period. It is residual in modern English, although its function has been
lexicalized, meaning that a separate word is provided to indicate its pres-
ence, with the noun itself appearing in the plural: a pair of cats; both mice, a
brace of pheasants. In Sanskrit, both nouns and verbs possess dual endings,
which are quite distinct from the singular and the plural.
The present work does not investigate the dual number. Whilst it may be
a basic feature of Sanskrit, declension presents the learner with sufficient
challenges with singular and plural endings. The dual is far less frequent
in Sanskrit texts than the singular and the plural, so it makes good sense
to keep matters as simple as possible for a learner taking the first steps
into the language. Only once a knowledge of basic declension has been
firmly grasped with respect to singular and plural forms ought the learner
to approach the dual. The presence of cases in Sanskrit (of which more will
be said in 6.2) inflates the number of possible forms for the noun to sixteen:
eight for the singular; eight for the plurals. Adding the dual would result in
twenty-­four forms, which makes grasping declension a daunting prospect at
the outset of one’s study of Sanskrit.
Returning to the words which were investigated in 5.2, it was noted that
they display a number of different endings. These endings indicate, in most
(but not all) cases, whether a noun is masculine, neuter or feminine. This
is not grammatical number but grammatical gender. Many Indo-­European
languages have maintained the concept of grammatical gender. French has
two genders: masculine (e.g. le chat: the cat) and feminine (la souris: the
mouse). German has three: masculine (der Hund: the dog); feminine (die
Katze: the cat); neuter (das Schaf: the sheep). In French and German, it is
not possible to determine the gender of a noun from the final sound or let-
ter in the noun. A final -­t in a French noun, for example, does not indicate
the masculine gender any more than a final -­s indicates feminine gender (la
jument: the mare; le fils: the son). As with number, Sanskrit exhibits more
consistency.
A short, final -­a in the stem form of a noun (which is the citation or
dictionary form, before any grammar has been added) indicates that a
noun is either masculine or neuter. It cannot be feminine. The word deva
102 The Sanskrit noun

is masculine, whilst kamala is neuter. That may not be terribly useful with
respect to being able to distinguish the gender of a noun ending in -­a, but it
does at least rule out the feminine as a possibility. As for niśā and rātrī, they
are both feminine. The endings -­ā and -­ī are reserved for feminine nouns.
Things are not always cut and dried in terms of distinguishing the gender of
a noun in Sanskrit: short -­i and -­u can be masculine, neuter or feminine –
although neuter nouns ending in -­i and -­u are rather limited. All told, the
Sanskrit noun is reasonably explicative in terms of indicating gender. This
is as well, given the intricacies of Sanskrit declension.

6.2 Cases and case functions

Grammatical gender is not a matter which is likely to cause the reader any
conceptual difficulty – especially if the reader has any exposure, however
minimal, to such languages as French, German, Italian or Spanish (which
remain the most widely learnt languages for English speakers in Europe and
the United States). As for those already familiar with any of the Indic lan-
guages, such as Panjabi, Gujarati and Hindi, a knowledge of what grammat-
ical gender represents will already be in place. The same cannot be said with
case, which is by no means present in all languages that have grammatical
gender. Whilst German has cases, French, Italian and Spanish do not. Case
endings were lost in the development of French, Italian and Spanish from
Latin, just as modern English lost the cases which Old English possessed,
replacing them by prepositions (of, from, to, etc.).
Case indicates the various forms which a noun undergoes as it reflects its
function within a sentence. Whilst it is true that modern English no longer
has case as part of its grammatical system, the vestiges of case are still evi-
dent. If we investigate the sentence: she sees me and I greet her, the words
she and her are different, although they appear to relate to the same person.
This is true, also, of the words me and I. Speakers of English would have no
difficulty in identifying the following sentence as ungrammatical:

*Her sees I and me greets she.

She/her and I/me are personal pronouns and these, in English, are the
remnants of case as it existed in Old English. She is the one controlling the
verb to see, but I am the one controlling the verb to greet. The recipient of
the action of seeing is me, whilst the recipient of the action of greeting is
her. The person or thing controlling the verb is the subject of the sentence
(she; I). The one who is the recipient of the verb is the object of the sentence
 The Sanskrit noun 103

(me; her). In Sanskrit, case endings applied to the end of a noun allow us
to know who or what is the subject of the sentence and who or what is the
object (Sanskrit also possesses personal pronouns, which are explored in
6.6). The case which indicates the subject is the nominative case, whilst
the object of a sentence is marked in the accusative case – unless the accu-
sative is overridden by another case indicating further information about
the object. These are the first two cases (of which there are eight) that the
learner should explore.
The following is a complete sentence in Sanskrit which uses both the
nominative and the accusative:

रामः अश्वं पश्यति

Rāmaḥ aśvaṃ paśyati
Rāma sees the horse.

Rāma is the subject of the sentence, since he controls the verb of seeing
(paśyati: he sees; is seeing). The recipient of the action of seeing is the horse
(aśva). Rāma is, accordingly, indicated by the nominative case and aśva by
the accusative. Both words have an ending which signifies the relevant case.
With Rāma, it is -­ḥ, with aśva, it is -­m. The reader will note that that ending
-­m appears as a dotted letter in the IAST. This is anusvāra, which indicates
that a word-­final -­m is followed by a sound other than a vowel. The word
order of the sentence is subject, object, verb. Regrettably for the learner,
Sanskrit recognizes a phenomenon called sandhi, of which more will be said
in Chapter 8. For the time being, indulgence is asked of the reader to accept
that the sentence, once sandhi has been applied, reads as follows:

रामो ऽश्वं पश्यति

Rāmo ’śvaṃ paśyati
Rāma sees the horse.

Sandhi is momentarily deferred for the sake of exposition, but it cannot

be ignored for long. Authentic Sanskrit (as opposed to Sanskrit that has
been simplified to ease the learner into the language) is fully sandhified.
Sanskrit word order can be extremely flexible, given that case end-
ings serve to identify the relationship of a noun within the sentence. The
104 The Sanskrit noun

following sentence would mean the same as the previous one – and there
are no sandhi changes to make:

अश्वं रामः पश्यति

aśvaṃ Rāmaḥ paśyati
Rāma sees the horse.

The extent to which the second sentence might differ in terms of nuance
from the first is something which is connected to stylistics rather than gram-
mar. From a grammatical point of view, the subject and object are clearly
identifiable, and the interpretation is the same. Stylistically, one could posit
that a displacement of the words Rāmaḥ and aśvaṃ, so that the object
appears first, might indicate an emphasis which would be reflected in the
English translation: It is the horse that is being seen by Rāma (as opposed to
some other creature or thing). That might be a suitable translation, accord-
ing to the context in which the sentence occurs, but it ought to be stressed
that such a translation reflects a passive construction, not an active one (i.e.
something is seen by someone). Passive constructions in Sanskrit, as with
English, have a different structure. To explore these at this juncture would
obscure matters rather than clarify them.
Aside from the nominative and accusative cases, Sanskrit possesses six
others, which need now to be investigated. It is worth, always, learning San-
skrit according to patterns, as these facilitate the absorption and retention
of new information. This is certainly true of the running order of the cases,
which the learner is strongly encouraged to learn in the following sequence:

Nominative
Accusative
Instrumental
Dative
Ablative
Genitive
Locative
Vocative

The information which cases provide, aside from the nominative and
accusative, is indicated by a number of prepositions in English (As regards
the nominative and accusative, reflecting the grammatical subject and
object, respectively, English employs word order: Mary sees James puts
 The Sanskrit noun 105

the subject before the verb and the object after it, in an active construction.
Mary is therefore the subject and James is the object). If the learner can
keep in mind the functions of the individual cases, together with the prepo-
sitions which English uses to represent them, a good start will have been
made in grasping Sanskrit declension.
Starting with the instrumental, this is the case which English expresses with
the prepositions by or with. The clue to remembering this is to bear in mind
what purpose an instrument serves in English. If it is a musical instrument, one
makes music with it; if a surgical instrument, a surgical procedure is undertaken
by means of it. If Rāma goes to town on horseback, Sanskrit determines that he
goes (gacchati) to town (nagara) by means of the horse (i.e. he uses the horse as
an instrument of travel). Accordingly, the horse appears in the instrumental case.

रामः अश्वेन नगरं गच्छति

Rāmaḥ aśvena nagaraṃ gacchati
(With sandhi: Rāmo ’śvena nagaraṃ gacchati)
Rāma goes to town by horse.

It is important to point out that the sentence contains one subject (Rāma)
and two objects (the town and the horse). The town is the direct object, since
it is the recipient of the action of going (It is the thing ‘being gone to’). As for
the horse, it is an indirect object, since it is not directly related, grammatically
speaking, to the action of going. Care needs to be taken to avoid something that
is the cause of confusion amongst many learners. If Rāma goes to town (with or
without a horse), he moves towards it. The preposition to is somewhat ambigu-
ous in English. One sees it in the sentence: I gave a book to Mark. In Sanskrit,
however, that example of to captures the notion of giving or of doing something
for someone’s benefit – and that is not the accusative but the dative case.
The dative case, as noted, is the case of giving. This is an exact transla-
tion of the Latin cāsus datīvus. In Sanskrit, the person or thing which is the
recipient of the action of giving is marked with the dative case. If Rāma
gives (yacchati) food (anna) to the horse, the sentence which states this in
Sanskrit is as follows:

रामः अश्वाय अन्नं यच्छति

Rāmaḥ aśvāya annaṃ yacchati
(With sandhi: Rāmo ’śvāyānnaṃ yacchati)
Rāma gives food to the horse.
106 The Sanskrit noun

It is not the food which is marked as dative, since it is not the recipient of
the action of giving. The food takes the accusative case. Rāma remains the
subject, marked by the nominative case, because he is the one performing
the verb. The recipient of the action of giving is the horse.
Both Rāma and the horse are masculine short -­a stems, whilst nagara
and anna are neuter short -­a stems. This does not make any difference as
regards the instrumental or dative cases, where the singular endings are
-­ena and -­āya, respectively, with both the masculine and the neuter short
-­a stems. This is true of the next three cases also (the ablative, genitive and
locative). Once one knows the singular case endings for masculine short
-­a stems, one also knows the singular case endings for the neuter short -­a
stems, except for the nominative. This is a pattern to which the learner’s
attention is drawn. As previously stated, patterns are useful in the learning
of Sanskrit.
The ablative case corresponds to the English proposition from. It is one
of the easiest cases to grasp, conceptually, for a speaker of English. Both the
masculine and neuter short -­a stems take the case ending -­āt in the singular.
The word nagara (town), as mentioned, is neuter:

रामः अश्वाः च नगरात् आगच्छन्ति

Rāmaḥ aśvāḥ ca nagarāt āgacchanti
(With sandhi: Rāmo ’śvāś ca nagarād āgacchanti)
Rāma and the horses come from the town.

Note the word ca, in the example given, which is the English and. It
does not come between the two things which it joins together but after both.
Rāma and the horse is thus Rāma horse and in Sanskrit. The definite arti-
cle (the) does not exist as such in Sanskrit, although Sanskrit possesses a
demonstrative pronoun which can express definiteness (The demonstrative
pronoun is investigated in 6.5). Note, also that the verb gacchati is prefixed
with ā-­ which gives the meaning to come rather than to go. The verb also
contains the plural ending -­anti rather than -­ati, since there is one Rāma and
at least three horses involved in the action of coming from town. Sanskrit
makes it clear that there is not a single horse or indeed two, since the case
ending for horse is the nominative plural, not the nominative singular or the
nominative dual.
There is something worth remarking, as concerns the ablative case end-
ing in the plural. This is -­ebhyaḥ, a form which is shared by the dative case
 The Sanskrit noun 107

ending for both the masculine and neuter short -­a stems. Whereas this might
seem to the learner to be a bonus, in that fewer forms must be committed
to memory, the ideal situation would be one in which all case endings were
distinct, one from the other. The sharing of endings results in a grammatical
ambiguity where one is reliant on context to determine which case is being
represented.
The genitive ending, in the singular of both masculine and neuter short -­a
stems is -­asya. English, too, has a genitive ending, visible by what linguists
and grammarians refer to as the ‘genitive apostrophe’. The pen belonging
to Mary may also be expressed as Mary’s pen. Although frequently misun-
derstood, and consequently omitted, in contemporary English – even by the
press – the genitive apostrophe is also used in the plural: the horses’ stable
does not refer to a stable for a singular horse (that would be the horse’s sta-
ble) but for two or more horses.

रामस्य सूतः नगरात् आगच्छति

Rāmasya sūtaḥ nagarāt āgacchati
(With sandhi: Rāmasya sūto nagarād āgacchati)
Rāma’s charioteer comes from the town.

Note that the word sūta (charioteer) is a masculine short -­a stem, for
which the conjugational endings are the same as aśva – and, for that mat-
ter, Rāma (although a plural form for Rāma is unlikely ever to be encoun-
tered in a Sanskrit text). Since the subject is in the singular, the verb form
is āgacchati. The plural form of the verb (āgacchanti), in the previous
sentence, reflected the fact that there were two subjects equally control-
ling the verb; Rāma as the first subject and at least three horses as the
second (i.e. a singular subject and a plural subject). Had there been only
two horses, the plural form of the verb would still have been appropriate,
since there would have been three entities (one human and two equine),
but the declensional ending for the horses would have necessitated a dual
form. As concerns the dual number, the learner is encouraged to bear in
mind its existence from the outset, although it is best to focus on the sin-
gular and plural forms. Once the latter are learnt, the dual can then easily
be incorporated. At the very beginning of one’s study of Sanskrit, clarity
is always best.
There are two remaining cases in Sanskrit: the locative and the vocative.
The locative locates an action as its name suggests. The English prepositions
108 The Sanskrit noun

in, on, at or through are the ones most likely to be used to translate the San-
skrit locative. Consider the following example, in which the subject of the
sentence is not, on this occasion, Rāma but his charioteer. Again, the verbal
ending is -­ati, because the subject is in the singular (This is the present
tense, which is investigated in Chapter 7). Here, the verb is to be (bhavati:
he/she/it is) rather than to see, to go or to come. The locative singular ending
-­e relates to the town:

रामस्य सूतः नगरे भवति

Rāmasya sūtaḥ nagare bhavati
(With sandhi: Rāmasya sūto nagare bhavati)
Rāma’s charioteer is in (the) town.

This brings us to the last of eight cases, which is the vocative. This is the
case of calling out – just as one’s vocation is one’s calling. If Rāma’s horse
were capable of articulate speech, he might wish to attract Rāma’s attention.
Since Rāma is a short -­a masculine stem, the horse would have but to use
the vocative case (in the singular, since there is only one Rāma). The voca-
tive singular, for short -­a stems, has the same appearance as the stem form.
The vocative singular (o Rāma!) is Rāma. Similarly, if Rāma were to call
to his horse, he would say aśva (o horse!).
The existence of eight cases may seem daunting to the learner, but
there is some consolation to be had in knowing that Sanskrit is extremely
methodical in their application. Declensional patterns are repetitive. Once
the learner has grasped a handful of patterns, the others can be learnt by
analogy. In other words, one works with what one already knows, noting the
differences that occur between a familiar declensional pattern and one that
is less familiar. There are more similarities between declensional patterns
than there are differences. Bucknell (1994 – see Chapter 10) identifies forty
declensional patterns as regular, noting an additional thirty-­three that show
irregularities. In the following four pages, two declensional patterns (or
paradigms) will be discussed for short -­a stems; one for masculine nouns,
one for neuter nouns. These exhibit minimal differences. We will then turn
to the declensional patterns for the long -­ā and long -­ī stems (feminine),
where the differences with the masculine and neuter paradigms are more
pronounced.
 The Sanskrit noun 109

6.3 Sample paradigms: short -­a stems

On the following page, two paradigms are given: one for a short -­a mas-
culine stem (deva) and one for a short -­a neuter stem (kamala). Departing
from a practice employed so far, only the endings are bolded, allowing for
them to be more legible. Where there is no difference between the endings
for the masculine and neuter declensional forms, there is no use of italics.
Italics indicate the existence of a difference between the declension of deva
and kamala. These differences occur in the nominative singular and nomi-
native, accusative and vocative plurals.
The reader will note something important about the nominative sin-
gular ending for kamala. It is the same as the accusative singular form.
Unless one knows that a short -­a stem is neuter, one is likely to assume
that kamalam indicates the accusative singular only, with the consequence
that a sentence containing kamalam will be understood as one in which
kamalam must be the object. The following sentence therefore presents the
learner with a conundrum. The lotus (kamala) grows (rohati) in the garden
(udyāna). The word udyāna is not problematic, in that the ending clearly
indicates the presence of the locative singular case ending -­e, whether the
stem is masculine or neuter (It is, in fact, neuter).

kamalam udyāne rohati

A/the lotus grows in the garden.

If one has assumed that kamala is masculine, then the case ending -­m
indicates an object. Yet the lotus is the thing controlling the verb of grow-
ing, meaning that it must be the subject, which calls for a nominative case
ending.
Two other points are worth noting. First, the dative and ablative singular
endings are distinct with both the masculine and neuter short -­a stems but
are the same in the plural. Second, the masculine vocative singular does
not have a final visarga (-­ḥ) and the neuter vocative singular does not have
a final -­m. In effect, the vocative singular forms in both genders have the
same appearance as the citation form of the noun: deva and kamala. With
both deva and kamala, the vocative plural endings are the same as the
nominative plural endings. This pattern is repeated time and again in San-
skrit declension and explains why the Sanskrit grammatical tradition did not
identify the vocative as a separate case.
110 The Sanskrit noun

Sample paradigms: short -­a stems

deva (god) Masculine
Case Singular Plural
Nominative devaḥ devāḥ
Accusative devam devān
Instrumental devena devaiḥ
Dative devāya devebhyaḥ
Ablative devāt devebhyaḥ
Genitive devasya devānām
Locative deve deveṣu
Vocative deva devāḥ

kamala (lotus) Neuter

Case Singular Plural
Nominative kamalam kamalāni
Accusative kamalam kamalāni
Instrumental kamalena kamalaiḥ
Dative kamalāya kamalebhyaḥ
Ablative kamalāt kamalebhyaḥ
Genitive kamalasya kamalānām
Locative kamale kamaleṣu
Vocative kamala kamalāni

6.4 More paradigms: long -­ā and long -­ī stems

Long -­ā and long -­ī stems have declensional endings that are quite distinct
from the short -­a stems. Whilst this might strike the learner as dispiriting,
the good news is that long -­ā and long -­ī stems exhibit a number of similari-
ties to each other. Knowledge of one assists greatly with the acquisition of
the other, albeit that there are a few more differences than exist between the
short -­a masculine and short -­a neuter stems. Long -­ā and long -­ī stems are
feminine.
 The Sanskrit noun 111

As with the nominative and vocative endings in the short -­a stems, the
nominative and vocative singular endings in the long -­ā and long -­ī stems
are not the same as each other. Whereas the vocative singular endings in the
short -­a stems are identical to the citation form of the noun, that is not so
with the long -­ā and long -­ī stems (niśā → niśe; rātrī → rātri). The citation
form is, instead, seen in the nominative singular of both the long -­ā and long
-­ī stems. There is pattern with the vocative plural endings that is consistent
with the short -­a stems: the nominative and vocative plural endings are the
same (devāḥ/devāḥ; kamalāni/kamalāni; niśāḥ/niśāḥ; rātryaḥ/rātryaḥ).
Another pattern which the long -­ā and long -­ī stems share with the short
-­a stems is that the declensional forms are the same in the dative and abla-
tive plural (-­ebhyaḥ in the short -­a stems, -­bhyaḥ in the long -­ā and long -­ī
stems). There is a difference in the sibilant between the locative plural end-
ings niśāsu and rātrīṣu, and a difference in the nasal between the genitive
plural endings niśānām and rātrīṇām. These relate to internal sandhi rules
which are investigated in 8.5.
Note the addition of y after the stem (i.e. niśā+y) where the declensional
endings begin with a vowel. This prevents the outcome a+ā, ā +a or ā+ā,
which Sanskrit does not permit. A similar violation to Sanskrit phonology is
sidestepped in rātrī, where the ī in the stem is substituted by the semivowel
y, preventing the impermissible diphthongs *īa and īā, not to mention the
triphthong *īai. Note, also, that the -­ā in the stem form niśā is short in the
instrumental singular (i.e. niśayā, not *niśāyā).
If the learner can master the short -­a stems (masculine and neuter) and
the long -­ā and long -­ī stems (feminine), a great stride will have been taken
into Sanskrit declension. These paradigms merit careful perusal.

More paradigms: long -­ā and long -­ī stems

niśā (night) Feminine
Case Singular Plural
Nominative niśā niśāḥ
Accusative niśām niśāḥ
Instrumental niśayā niśābhiḥ
Dative niśāyai niśābhyaḥ
Ablative niśāyāḥ niśābhyaḥ
Genitive niśāyāḥ niśānām
Locative niśāyām niśāsu
Vocative niśe niśāḥ
112 The Sanskrit noun

rātrī (night) Feminine

Case Singular Plural
Nominative rātrī rātryaḥ
Accusative rātrīm rātrīḥ
Instrumental rātryā rātrībhiḥ
Dative rātryai rātrībhyaḥ
Ablative rātryāḥ rātrībhyaḥ
Genitive rātryāḥ rātrīṇām
Locative rātryām rātrīṣu
Vocative rātri rātryaḥ

6.5 The demonstrative pronoun tat

Sanskrit possesses a number of words to signify this one, that one, it. These
are words which refer to someone or something whose identity has been
established in the context of an utterance or sentence and, in this respect,
they function as such words do in English. For example, if one has been
discussing the cost of an item, one might choose not to refer to the item by
using its noun form again but, instead, using the appropriate pronoun, as the
following example illustrates:

The seafood pasta looks nice, but it is too expensive.

English is sensitive to gender if the entity is a person – less so, if the

entity is an animal (Babies tend also to be indeterminate as regards gender).

Mary is upset with James. She doesn’t like him to drink too much.
Jim’s dog is enormous. He is/It is a real beast.

(He makes it ambiguous as to whether it is Jim or his dog being

discussed.)
 The Sanskrit noun 113

The words he/she/it, substituting for nouns, are pronouns. Sanskrit pos-
sesses these as well as English. In fact, the Sanskrit demonstrative pronoun
tat is unlikely to evade the learner for long, so pervasive is it in all manner
of texts. It is ‘demonstrative’ in that it points to someone or something,
although it should be noted that it also happens to be the third person pro-
noun. As such, tat can be understood to be that one, this one, those, these,
he, she, it, they (etc.) according to the context. Since Sanskrit has gram-
matical number and case, it follows that tat may have many different forms,
according to how many entities are involved and what the case relationship
happens to be between the entity or entities in question and the other words
within the sentence.
When a tat-­form is used for a god, already present in the discourse, the
tat-­form in question must be masculine, because the Sanskrit noun is mas-
culine (deva: god). Just as the masculine short -­a stem deva can take sev-
eral case endings in the singular (and also the dual and plural), so too can
the tat-­form. The exception is the vocative case, which does not apply with
tat. The endings of tat-­forms often resemble those of the nouns to which
they refer. There are differences all the same and these must be learnt quite
early in the study of Sanskrit.

The demonstrative pronoun: tat with short -­a stems

tat + deva (god) Masculine
Case Singular Plural
1
Nominative saḥ devaḥ te devāḥ
Accusative tam devam 2 tān devān
Instrumental tena devena taiḥ devaiḥ 5
Dative tasmai devāya tebhyaḥ devebhyaḥ 4
Ablative tasmāt devāt 3 tebhyaḥ devebhyaḥ 4
Genitive tasya devasya teṣām devānām 2
Locative tasmin deve teṣu deveṣu
Vocative n/a n/a
114 The Sanskrit noun

tat + kamala (lotus) Neuter

Case Singular Plural
Nominative tat kamalam tāni kamalāni
Accusative tat kamalam tāni kamalāni
Instrumental tena kamalena taiḥ kamalaiḥ
Dative tasmai kamalāya tebhyaḥ kamalebhyaḥ
Ablative tasmāt kamalāt tebhyaḥ kamalebhyaḥ
Genitive tasya kamalasya teṣām kamalānām 2
Locative tasmin kamale teṣu kamaleṣu
Vocative n/a n/a

Notes
1 The form saḥ has a special set of rules. The visarga is dropped in all cases
except if saḥ comes at the end of a sentence. Also, if followed by short a,
it changes to so and the short a-­is dropped (e.g. saḥ aham → so ’ham).
2 If a tat-­form ends with -­m and is followed by any consonant, semivowel
or sibilant, anusvāra applies (e.g. taṃ devam; tesāṃ kamalānām, etc.).
3 If a tat-­form ends with -­t and is followed by a voiced consonant, the ending
changes to -­d (e.g. tasmād devāt). This is a consonant sandhi rule (8.3).
4 If a tat-­form ends with -­aḥ and is followed by a voiced consonant, semi-
vowel or h, the ending changes to -­o (e.g. tebhyo devebhyaḥ). See 8.2.
5 If a tat-­form ends with -­aiḥ and is followed by a voiced consonant,
semivowel or h, the ending changes to -­air (e.g. tair devaiḥ). See 8.2.

If a tat-­form is used alongside the noun to which it refers, it lends the noun a
degree of definiteness. The noun phrase saḥ devaḥ (which becomes sa devaḥ,
with the dropping of the visarga, as mentioned in the first note on the previous
page) can be translated as that god, this god or simply the god, according to
context. Tat may function, therefore as the Sanskrit equivalent of the definite
article in English. Strictly speaking, tat is a neuter form which identifies neuter
nouns in both the nominative and accusative singular. Its name is retained irre-
spective of whether it is being used for masculine and feminine nouns. Many
Sanskrit grammars refer to it as tad, which is perfectly acceptable. The only
difference between tat and tad is whether it is followed by a phoneme which
is voiceless (in which case it is tat) or voiced (requiring the form tad). With the
 The Sanskrit noun 115

nominative and accusative singular noun kamalam, beginning with the voice-
less k, it has the form tat: tat kamalam (that/this/the lotus).
The learning of tat may seem arduous, but it is worth the perseverance. Not
only is tat frequently encountered but it also gives the student of Sanskrit fur-
ther declensional patterns for free. Once learnt, tat-­forms provide the basis for
the declension of another demonstrative pronoun, etat/etad, which is deemed
to refer to an entity closer to the speaker than tat (This one here rather than that
one there) and is identical to tat, with the addition of e-­at the beginning (There
are two minor differences, relating to the change in the sibilant in the nomina-
tive singular masculine and feminine forms. The predicted *esaḥ and *esā,
respectively, are not attested and appear instead as eṣaḥ and eṣā). Importantly,
tat-­forms have the same endings as yat/yad: the relative pronoun, indicating
who(m), whose, which, etc. The use of the relative pronoun is best seen in
context. It appears in the last of four readings from the Bhagavadgītā (9.3.4).
In terms of when the learner might wish to consider committing tat-­forms
to memory, the answer is ‘as soon as practicable’. First, however, the learner
must be secure as regards his or her grasp of the declension of short -­a stems
(both masculine and neuter) and at least the long -­ā feminine stem. Most tat-­
forms have the same endings as the stems in question. The dual forms of
tat, as with the declensional paradigms presented, have not been included.
These are less frequent than the singular and plural forms and, on that basis,
may profitably be deferred until the singular and plural forms have become
familiar. In the study of Sanskrit, rushing to learn everything at once is never
advisable.

The demonstrative pronoun: tat with long -­ā/-­ī stems

tat + niśā (night) Feminine
Case Singular Plural
Nominative sā niśā 1 tāḥ niśāḥ 3
Accusative tām niśām 2 tāḥ niśāḥ 3
Instrumental tayā niśayā tābhiḥ niśābhiḥ 5
Dative tasyai niśāyai tābhyaḥ niśābhyaḥ 4
Ablative tasyāḥ niśāyāḥ 3 tābhyaḥ niśābhyaḥ 4
Genitive tasyāḥ niśāyāḥ 3 tāsām niśānām 2
Locative tasyām niśāyām 2 tāsu niśāsu
Vocative n/a n/a
116 The Sanskrit noun

tat + rātrī (night) Feminine

Case Singular Plural
Nominative sā rātrī tāḥ rātryaḥ 3
Accusative tām rātrīm 2 tāḥ rātrīḥ 3
Instrumental tayā rātryā tābhiḥ rātrībhiḥ 5
Dative tasyai rātryai tābhyaḥ rātrībhyaḥ 4
Ablative tasyāḥ rātryāḥ 3 tābhyaḥ rātrībhyaḥ 4
Genitive tasyāḥ rātryāḥ 3 tāsām rātrīṇām 2
Locative tasyām rātryām 2 tāsu rātrīṣu
Vocative n/a n/a

Notes
1 The form sā stays unchanged. There is no final visarga.
2 If a tat-­form ends with -­m and is followed by any consonant, semi-
vowel or sibilant, anusvāra applies (e.g. tām niśām; tām rātrīm,
etc.).
3 If a tat-­form ends with -­āḥ and is followed by a voiced consonant
or semivowel, the visarga is simply dropped (e.g. tā niśāḥ, etc.).
See 8.2.
4 If a tat-­form ends with -­aḥ and is followed by a voiced consonant
or semivowel, the ending changes to -­o (e.g. tābhyo niśābhyaḥ).
See 8.2.
5 It a tat-­form ends with -­iḥ and is followed by a voiced consonant,
semivowel or h, the ending changes to -­ir (e.g. tābhir niśābhiḥ).
There is one exception: if the following phoneme is r, the visarga is
dropped and the preceding vowel is lengthened (e.g. tābhī rātrībhiḥ).
See 8.2.

6.6 Personal pronouns

As with tat, the personal pronouns indicate a noun, but they do so without
stating the noun. In English, there are only a few forms for the personal pro-
nouns, because there are few case functions in English. As demonstrated in
 The Sanskrit noun 117

6.2, whilst English may not formally recognize grammatical case, it is nev-
ertheless sensitive to the fact that a noun or its pronoun may be the subject
of a sentence, the object or a possessor. These are, of course, the nomina-
tive, accusative and genitive cases, respectively, of Sanskrit. The following
tables match the personal pronouns of the first and second person to their
Sanskrit equivalents. In having eight cases (seven of which have person
pronoun forms), not to mention a dual number, the number of forms is much
greater in Sanskrit:

First Person Singular Plural Singular Plural

Nominative I we aham vayam
Accusative me us mām asmān
Genitive my/mine our(s) mama asmākam

Second Person Singular Plural Singular Plural

Nominative you you tvam yūyam
Accusative you you tvām yuṣmān
Genitive your(s) your(s) tava yuṣmākam

It is relevant to mention that the singular forms of the second person in

English are identical to the plural ones. Modern English started to abandon
the singular forms (thou, thee, thy/thine) in favour of the plurals. Although
attested in writing of the Tudor and Jacobean periods and occasionally still
heard in dialect English, use of the older forms of the singular decreased
during the seventeenth century. Had they been retained, the similarities
between English and Sanskrit would have been much more evident. The
older English second person singular forms began with th-­, the Sanskrit
ones have t as the initial phoneme, as do the French forms (toi/tu, te, tien).
Be that as it may, ample similarities remain between English personal pro-
noun forms and Sanskrit ones. Note how the first person singular shows a
predominance of forms beginning with m in both languages and the English
plural forms have w/u/ou-­/ in contrast to the v/a of Sanskrit (The second
person plural forms show an even greater similarity between English and
Sanskrit).
118 The Sanskrit noun

Personal pronouns
First Person
Case Singular Plural
Nominative aham vayam
Accusative mām (mā) asmān (naḥ)
Instrumental mayā asmābhiḥ
Dative mahyam (me) asmabhyam (naḥ)
Ablative mat asmat
Genitive mama (me) asmākam (naḥ)
Locative mayi asmāsu
Vocative n/a n/a

Second Person
Case Singular Plural
Nominative tvam yūyam
Accusative tvām (tvā) yuṣmān (vaḥ)
Instrumental tvayā yuṣmābhiḥ
Dative tubhyam (te) yuṣmabhyam (vaḥ)
Ablative tvat yuṣmat
Genitive tava (te) yuṣmākam (vaḥ)
Locative tvayi yuṣmāsu
Vocative n/a n/a

Notes
1 Some grammars of Sanskrit cite the first person singular and plural
ablative forms as mad and asmad, respectively, and the second person
singular and plural ablative forms as tvad and yuṣmad, respectively.
Such grammars are likely, also, to cite tat as tad, etat as etad. This is
perfectly in order.
 The Sanskrit noun 119

2 The dual forms are not shown. The first person dual equates, in con-
cept, to the two of us; the second person dual to the two of you. The dual
forms may be ignored until the learner encounters them in a reading
passage. They do not occur in the present work.

If a personal pronoun qualifies a noun, as with tat, it precedes the noun in

question (Tat is, after all, the personal pronoun for the third person as well
as a demonstrative pronoun). There are, however, shortened forms of various
pronouns that are placed after the noun which they qualify. These are known
as ‘enclitic’ forms, in that they cannot come before the noun in question. The
reader has already encountered an enclitic. That was the conjunction ca (and),
placed after the words that it joins together (see 6.2, with reference to the com-
ments made relating to the ablative case). In the table on the previous page,
these enclitic forms are indicated in brackets. They do not have unique forms
and can, as a result, cause ambiguity to arise. Context is needed, with the enclitic
forms of the personal pronouns, to figure out the case that is being represented.
With the non-­enclitic form:

मम पुत्रः जीवति
mama putraḥ jīvati
(With sandhi: mama putro jīvati)
My son (putra) lives (jīvati).

With the enclitic form:

पुत्रः मे जीवति
putraḥ me jīvati
(With sandhi: putro me jīvati)
My son lives.

Enclitic forms are by no means rare. The greeting exchanged daily

between millions of Hindus is namaste. That is not a single word but two:
namas (salutation; homage) and the enclitic form of the second person
singular dative (te: to you). The non-­enclitic form, whilst grammatically
correct is not used as variant form of the greeting (*tubhyaṃ namas). As
for the other common greeting, namaskāra, the word namas is the same
but a personal pronoun is not present. The word kāra is derived from the
verbal root kr̥ (to do; make; create; fashion). Whilst namaste is, therefore,
120 The Sanskrit noun

a salutation being offered to another, namaskāra is simply a statement

that salutation is being made. This brings the reader to the point where an
investigation of matters relating to verbs becomes incumbent, although not
without a couple of exercises beforehand, by way of consolidation of the
present chapter.

6.7 Exercises

6.7.1 Questions on declension

The reader is asked to attempt the following questions. The answers
are given on the next page, for convenience’s sake. It may be benefi-
cial to attempt the questions twice: referring to the relevant tables on
the first occasion and a second time from memory, to consolidate the
information.

Questions
(i) What are the functions of the instrumental, dative, ablative and geni-
tive cases?
(ii) Give the case and number of the following forms:

devaiḥ; kamalasya; niśāyām; rātrībhyaḥ.

(iii) Give the case and number the following forms tat-­forms:

tam; tām; tān; tāni.

(iv) Two tat-­forms do not begin with the phoneme t. Which ones are they?
(v) Give the case and number of the personal pronoun forms vayam and
tvayā.
(vi) What are the personal pronoun forms for my/mine and your(s)
(singular)?
(vii) What are the enclitic forms of the first person and for which cases are
they available?
(viii) What are the enclitic forms of the second person and for which cases
are they available?
 The Sanskrit noun 121

Answers
(i) The instrumental signifies with; by; by means of. The dative is the
case of giving. (Note that it is the English: He gives this to you, not the
to of: He goes to the town). The ablative indicates from. The genitive
indicates possession (of; belonging to).
(ii) devaiḥ Instrumental plural.
kamalasya Genitive singular.
niśāyām Locative singular.
rātrībhyaḥ Dative or ablative plural.
(iii) tam Accusative singular (masculine).
tām Accusative singular (feminine).
tān Accusative plural (masculine).
tāni Nominative and accusative plural (neuter).
(iv) saḥ (masculine singular) and sā (feminine singular).
(v) vayam Nominative plural of the first person.
tvayā Instrumental singular of the second person.
(vi) mama (my/mine); tava (your(s), singular).
(vii) Singular: mā (accusative); me (dative); me (genitive).
Plural: naḥ (accusative, dative and genitive).
(viii) Singular: tvā (accusative); te (dative); te (genitive).
Plural: vaḥ (accusative, dative and genitive).

6.7.2 Sentences for translation

These sentences are aimed at activating the vocabulary that appeared in
earlier chapters as well as the current one. In Sanskrit, the article (a/the)
is implied by the context. New words (or reminders) are given in brackets.
The appropriate verb forms are as stated.
No knowledge of sound changes (sandhi) is presupposed, since this has
yet to be explored (Chapter 8). The translations on the following page nev-
ertheless give both the unsandhified and sandhified versions of the sen-
tences. It is recommended that the reader compare both versions to see if
anything can be deduced, at this stage, regarding sandhi.
122 The Sanskrit noun

Sentences

(i) जनाः अन्नेन जीवन्ति।

(jana: person. Masculine short -­a stem)

(ii) बुद्धः ग्रामात् आगच्छति।

(grāma: village. Masculine short -­a stem)

(iii) ताः सं स्कृतं पठन्ति।

(paṭhanti: they recite; are reciting)

(iv) सा रामस्य पुत्राय कमलानि यच्छति।

(yacchati: he/she/it gives; is giving)

(v) कृ ष्णः अर्जुनं युद्धे पश्यति।

(yuddha: battle. Neuter short -­a stem)

(vi) नगरस्य जनाः कृ ष्णं रामं च पूजयन्ति।

(pūjayanti: they worship; are worshipping)

Translations

(i) janāḥ annena jīvanti

(With sandhi: janā annena jīvanti)
People live by means of food.

(ii) Buddhaḥ grāmāt āgacchati

(With sandhi: Buddho grāmād āgacchati)
(The) Buddha comes from (the) village.

(iii) tāḥ saṃskr̥taṃ paṭhanti

(With sandhi: no change)
They (feminine plural) recite Sanskrit.

(iv) sā Rāmasya putrāya kamalāni yacchati

(With sandhi: no change)
She gives lotuses to Rāma’s son.
 The Sanskrit noun 123

(v) Kr̥ṣṇaḥ Arjunaṃ yuddhe paśyati

(With sandhi: Kr̥ṣṇo ’rjunaṃ yuddhe paśyati)
Kr̥ṣṇa sees Arjuna in (the) battle.

(vi) nagarasya janāḥ Kr̥ṣṇaṃ Rāmaṃ ca pūjayanti

(With sandhi: no change)
(The) people of (the) town worship Kr̥ṣṇa and Rāma.

Note how an ending in -­aḥ, when followed by a voiced consonant, changes

to -­o (sentence ii). When followed by a short a-­, -­aḥ changes to -o and the
following a-­ disappears (sentence v). An ending in -­āḥ, when followed by
a voiced phoneme – be it a consonant or vowel – simply drops the visarga
(sentence i) and there is no mutation of the long -­ā to -­o. These are visarga
sandhi changes, as discussed in 8.2. Also, a voiceless consonant at the end
of a word, if followed by a vowel, is voiced (sentence ii). This is consonant
sandhi, discussed in 8.3.
7
THE SANSKRIT VERB
Conjugation

7.1 Preliminaries
The Sanskrit verb represents the most complex aspect of the language, and
a full investigation of conjugation lies well beyond the scope of the pre-
sent work. To attempt to present a synoptic view of conjugation would only
serve to baffle the beginner. It is crucial, nevertheless, to understand how
the verb functions. In the discussion which follows, the focus will be on
exploring the Sanskrit verb in the present tense and the imperfect. There are
many things to be said about the verb and a knowledge of formal grammar
is not presupposed on the part of the reader.
To begin with, it is worth distinguishing between tense and mood. Tense
is a description of when something is happening – the past, the present or
the future. Mood defines the way an action is taking place. Sanskrit has
five tenses: present (which includes the indicative, imperative and optative
moods), imperfect, perfect, aorist, future. The present tense in the indicative
mood represents the logical and sensible first step in the study of the San-
skrit verb. The indicative is used in the making of a statement or the asking
of a question. You go is a statement; are you going? is a question. In both
instances, the tense is the present.
Along with five tenses, Sanskrit also possesses an active, middle and
passive voice. English makes the distinction between an active and passive
voice but does not formally recognize a middle voice. An English speaker
would not fail to identify a difference between the following sentences:

DOI: 10.4324/9780429325434-8
 The Sanskrit verb 125

Peter washes the dog.

The dog is washed by Peter.

In the first sentence, the emphasis is placed on the one who is perform-
ing the action. It is Peter who is actively washing the dog. In the second
sentence, the emphasis has shifted from Peter to the dog, who is being
passively washed. The middle voice is more difficult to grasp, in that the
emphasis may be on the one performing the action expressed by the verb,
but the focus is on the action that is taking place. In the Sanskrit grammati-
cal tradition, a distinction is made between the active voice, which is termed
as a word/voice for another (parasmaipada), and the middle voice, which
is a word/voice for oneself (ātmanepada).
There is something distinctly reflexive about the middle voice which,
according to traditional Indic terminology, strongly indicates that the mid-
dle voice would originally have signified the personal attachment of the one
performing the verb to its outcome. This can be captured by the following
examples from English and French:

Peter washes [the dog]/Pierre lave [le chien] (Active voice).

Peter washes [himself]/Pierre [se] lave (Middle voice).

Whatever the original semantic distinction in Proto-­ Indic or Vedic

between the active and middle voice, Sanskrit is deemed to have lost any
such distinction. Classical Sanskrit retains the grammatical distinction all
the same. A verb root, along with being allocated to a conjugational class,
is also indicated as occurring in the active or middle voice. Very often, a
verb root is identified as capable of being expressed in both voices. Most
grammars and dictionaries of Sanskrit use the capital letters P (parasmai-
pada) and Ā (ātmanepada) to signify this, although some grammars and
dictionaries prefer the capital letter U (ubhayapada: a word/voice for both)
to indicate that the verb may appear in either voice.
As with the noun, the Sanskrit verb uses the concept of number. Verbs
are performed in the singular (by one entity), the dual (two and only two
entities) or the plural (three or more entities). Unlike the noun, case does
not come into play in conjugation and there is no notion of gender. The
third person singular of the verb to go (he/she/it goes) remains the same
whether the entity is masculine, feminine or neuter. Whereas a noun under-
goes declension according to its stem, the verb is a root from which a stem
must be derived before it can be conjugated.
126 The Sanskrit verb

7.2 Root, stem and vowel mutation

In Sanskrit, verbal roots are classified as belonging to one of ten classes,
usually indicated in Sanskrit grammars by Roman numbers. A verbal root
forms a stem according to the strategies appropriate to the class to which
it belongs. It is important to note that the stem represents a modification of
the verbal root to which personal endings are applied, yielding the neces-
sary grammatical information about who is performing the verb. Personal
endings are not applied directly to a verbal root.
In English, only the third person tends to show what appears to be a
personal ending, with personal pronouns being deployed to give the requi-
site information regarding the subject of the verb. Consider the following
example:

Present simple (English)

Person Singular Plural
First I go we go
Second you go you go
Third he/she/it goes they go

As the English example demonstrates, the third person singular ending is

distinct; elsewhere, go is the conjugated form (This is the pattern in English,
with the notable exception of the verb to be, where the first person singular
also has a unique form: I am; you are; he/she/it is; we are; you are; they
are). In the Sanskrit conjugation of the verb to go, by contrast to English, all
endings are distinctive. This makes the personal pronoun redundant, given
that each form can be identified by its ending:

Present indicative – active voice (Sanskrit)

Person Singular Plural
First gacchāmi gacchāmaḥ
Second gacchasi gacchatha
Third gacchati gacchanti

Two things are worth noting about the Sanskrit conjugation of the verb
to go in the present tense. First, English distinguishes between the present
 The Sanskrit verb 127

simple (I go, you go, etc.) and the present continuous or progressive (I am
going, you are going, etc.). When translating a sentence from Sanskrit to
English, the appropriate form of the present tense must be selected. Sec-
ond, Sanskrit often uses a personal pronoun, albeit that it is not necessary
for meaning. On such occasions, one ought to be sensitive to the fact that
this is indicative of emphasis: gacchāmi already means I go/I am going. If
the appropriate personal pronoun is included (aham: I), a different reading
suggests itself:

ahaṃ gacchāmi
I am the one who goes/is going.

Conjugation of a verb in Sanskrit does not rely on identifying the infini-

tive form of a verb, as is the case in English (see 7.5 for more on the infini-
tive). Instead, Sanskrit uses the verbal root to generate a stem. Some of
these modifications involve vowel mutation, a phenomenon briefly dis-
cussed in 5.2, ix. Four of the ten verb classes are referred to as ‘thematic’,
in that they share a common theme (Class I, IV, VI and X). The other verb
classes (II, III, V, VII, VIII, IX) use different strategies to generate a stem.
The present work will focus on the thematic verb classes. To do otherwise
would be to present the reader with a mass of information on conjugation
strategies, each of which would require investigation in the context of sound
changes, irregularities, etc. The thematic group is always studied first, since
it involves four of the ten classes at a stroke and accounts for substantially
over half of all the verb forms encountered by the student of Sanskrit (Class
I is unusually large and contains the lion’s share of the most frequently
occurring verbs).
The concept of vowel mutation (or vowel strengthening) states that
a vowel may undergo changes according to ‘grade’. This grade is effec-
tively the fusion of a short a into the vowel in question, to produce a
qualitative difference in the first instance (guṇa). With the strongest
grade, a further short a is added to guṇa, which produces a quantita-
tive distinction (vr̥ ddhi). It will be noted that e and o are not simple
vowels but complex ones. They appear in the table of vowel mutation
as guna-­forms, thereby reinforcing their classification by the Sanskrit
grammatical tradition as vowels which were in some way less ‘pure’ than
a/ā, i/ī, u/ū, r̥ /r̥ ̄ and l̥ . The diphthongs, likewise, are present. They are
vr̥ ddhi-­forms.
128 The Sanskrit verb

Simple vowel a or ā i or ī u or ū r̥ or r̥̄ l̥

guṇa a e o ar al
vr̥ddhi ā ai au ār (n/a)

Note that a does not change its quality at any point, but merely its length. This
is because the strengthening agent is a itself. Only in the vr̥ ddhi grade does short
a change to long ā. As for r̥ , r̥ ̄ and l̥ , the pure vowel is replaced by the closest
semivowel, since *ar̥ , *ar̥ ̄ and *al̥ are not recognized as diphthongs in Sanskrit.

7.3 Thematic verbs

Verbs belonging to classes I, IV, VI and X share a common theme, as noted.
Two of these additionally use vowel mutation in the process of generating
the verbal stem. These are Class I and Class X (Classes IV and VI do not
involve a mutation of the root vowel). Vowel mutation is best seen with
reference to an actual example and, with this is mind, the verb root bhū (to
be) is a good starting point.
√ bhū is extremely common in Sanskrit and therefore serves as an excel-
lent introduction to the thematic group. It is an interesting example, since it is
not entirely transparent as regards the process of vowel mutation. That said,
once the principle is grasped, other examples of vowel mutation easily fall into
place. The root vowel is subject to mutation and, since the vowel in question is
neither a nor ā, strengthening produces something both audibly different to ū.
Applying guṇa to bhū produces *bho, which is not attested in the conju-
gational paradigm of the verb in the present tense. The form *bho would, in
any event, lead to a violation in that the thematic strategy for a Class I verb
is to add a short a after the modified root and before the personal endings.
This would result, in principle, in the following:

*bho + a + ti (personal ending of the third person singular)

Clearly, something is required to prevent the violation *oa from taking

place. This is resolved by the application of one of a set of rules relating to
instances where a complex vowel or diphthong precedes a vowel within a
word. Taking *bho as the example, the relevant rule is as follows:

Root with guṇa Before a vowel Stem

bho bhav bhava
 The Sanskrit verb 129

The set of rules in question also transforms -­e → -­ay; -­ai → -­āy; -­au
→ -­āv. They prevent the creation of impermissible diphthongs, such
as *oa, by changing the complex vowels and diphthongs into the seg-
ment: simple vowel + semivowel (with a long vowel, as regards the
diphthongs).
The addition of a after the modified verbal root is the theme which cre-
ates the notion of a thematic group. Classes IV, VI and X also employ a,
but not always by itself. In Class IV and Class X, the semivowel y is also
present.
Thematic verbs (Class I): the present tense
√ bhū (to be) Class I
Strategy: -­a is added to the root but the root undergoes mutation.
Stem: bhava

Active voice (parasmaipada)

Person Singular Plural
First bhavāmi bhavāmaḥ
Second bhavasi bhavatha
Third bhavati bhavanti

Middle voice (ātmanepada)

Person Singular Plural
First bhave bhavāmahe
Second bhavase bhavadhve
Third bhavate bhavante

Notes
1 The personal endings (-­mi, -­si, -­ti, etc.) are added to the stem form of
the verb. There are two exceptions to note. First, the final -­a in the stem
in lengthened to -­ā in the first person (both singular and plural) except
for the first person singular of the middle voice, where it is -­e.
2 All of the middle voice forms end in -­e, which replaces the final vowel
in active voice forms ending in a vowel (e.g. bhavati → bhavate;
bhavanti → bhavante). If the active voice form ends in visarga,
the ending -­e is simply added and the visarga changes to the aspirate
130 The Sanskrit verb

(bhavāmaḥ → bhavāmahe). There are two exceptions: bhave and

bhavadhve.
3 The stem *bhū + a would violate the sound system of Sanskrit, which
does not permit any diphthong other than ai or au. Vowel mutation
takes place, resulting in av + a (i.e. bhū → bhav + a = bhava).

Thematic verbs (Classes IV, VI, X): the present tense

√ nr̥t (to dance) Class IV
Strategy: -­ya is added to the root. There is no vowel mutation.
Stem: nr̥tya

Active voice
Person Singular Plural
First nr̥tyāmi nr̥tyāmaḥ
Second nr̥tyasi nr̥tyatha
Third nr̥tyati nr̥tyanti

√ likh (to write) Class VI

Strategy: -­a is added to the root. There is no vowel mutation.
Stem: likha

Active voice
Person Singular Plural
First likhāmi likhāmaḥ
Second likhasi likhatha
Third likhati likhanti

√ cur (to steal) Class X

Strategy: -­aya is added to the root, which undergoes vowel mutation.
Stem: coraya

Active voice
Person Singular Plural
First corayāmi corayāmaḥ
 The Sanskrit verb 131

Active voice
Second corayasi corayatha
Third corayati corayanti

(Note that cur → cor merely takes the guṇa-­form of vowel mutation.)

7.4 The imperfect

The imperfect is one of three tenses which Sanskrit employs to express past
event: I was; you were; he/she/it was; etc. The other two tenses which do so
are the perfect and the aorist. Unlike the perfect and the aorist, the imperfect
uses the present stem to generate its forms. The same is true of two moods
within the present tense, along with the indicative mood which has provided
the examples so far (These are the imperative and the optative, where the
former indicates an order or instruction, such as go and look!, and the latter
has a wide range of uses including expressing a desire or wish for something
to happen: you ought to go and look). The imperfect is extremely prevalent in
Sanskrit, so it is useful for the learner to be familiar with it at an early stage.
As with the present tense, the imperfect requires a stem to be generated
from a verbal root before personal endings can be attached and a finite verb
form created (A ‘finite verb’ is one in which a person is indicated – first,
second or third – in the singular, dual or plural number. A verb form that is
not grammatically associated with a person is referred to as ‘non-­finite’ or
‘infinitive’). Unlike the present tense, the various forms of the imperfect all
have a prefixed short a, known as the ‘augment’. It is not helpful to think
of the augment as a prefix, since Sanskrit has a considerable number of
prefixes which are attached to the verbal stem and, in many cases, modify
the meaning of the verb.
The augment signifies only one of a number of differences between the
formation of the imperfect and that of the present. The singular endings of
the present, in the active voice (-­mi, -­si, -­ti), appear in the active voice of the
imperfect as -­m, -­h, -­t. In the plural forms of the active voice, the endings of
the imperfect are not -­maḥ, -­tha and -­nti but, rather, -­ma, -­ta and -­an – not
neglecting the augment. Because of the many differences existing between
the personal endings of the present and the imperfect, these endings must be
learnt separately. A simple ‘transformation rule’ is not at hand. The learner
is advised to concentrate on the present tense, in both voices, and to look at
the imperfect once the present tense endings have become second nature.
132 The Sanskrit verb

Sanskrit conjugation is not easy. The trick is to learn something thoroughly

and to dovetail the knowledge of that thing into the acquisition of new infor-
mation. Knowledge of the present tense should precede familiarity with the
imperfect.

The imperfect (√ bhū)

√ bhū (to be) Class I
Stem: bhava

Active voice (parasmaipada)

Person Singular Plural
First abhavam abhavāma
Second abhavaḥ abhavata
Third abhavat abhavan

Middle voice (ātmanepada)

Person Singular Plural
First abhave abhavāmahi
Second abhavathāḥ abhavadhvam
Third abhavata abhavanta

Notes
1 Visarga is very often synonymous with the phoneme s. Bearing this
is mind, there is a strong similarity between the singular, active voice
endings of the present (-­mi, -­si, -­ti) and the corresponding imperfect
forms (-­m, -­ḥ, -­t). A final short -­i is absent in the imperfect endings.
2 The stem in the first person plural, in both voices, has a long -­ā.
This is precisely the case with the present tense endings. A long -­ā is
not, however, found in the first personal singular form of the active
voice.
3 The first person singular of the middle voice patterns perfectly with the
present tense (bhave with the augment: abhave). The patterning with
the first person plural of the middle voice, whilst not exact, is not a far
cry from the present tense form. Aside from the augment, only the final
vowel differs: bhavāmahe (present); abhavāmahi (imperfect).
 The Sanskrit verb 133

7.5 The infinitive

As mentioned in 7.2, Sanskrit does not use the infinitive to identify the
base form of a verb prior to conjugation. The infinitive is not the simplest
form of the verb, in Sanskrit, in that a suffix is necessary for its creation.
The suffix in question is -­tum or -­itum added to the verbal root. From that
point, there is no modification. The infinitive does not refer to any particu-
lar grammatical person, so personal endings are not applied. Its translation
into English is straightforward. It is consistent with the English infinitive:
the base form of the verb preceded by the preposition to (e.g. √ han →
hantum: to strike; slay; kill). A grammatically well-­formed sentence can-
not just contain a verb root with an infinitive ending, and Sanskrit employs
other strategies to indicate that an action is being, has been or is to be
undertaken by the subject of the sentence. The most evident of these is the
use of a finite verb.
Deciding whether a given verb root adds -­tum or -­itum in the forma-
tion of the infinitive is not always straightforward. The learner might
suppose that a verbal root ending in a consonant would take -­itum and
that is indeed the case with √ likh → likhitum (to write). Sadly, that
is not a rule which can in any sense be said to be widespread. With the
verbs to renounce; abandon and to go, for example, the verbal root ends
in a consonant but the infinitive suffix is -­tum, not -­itum. This causes
a sound change to take place, avoiding a cluster which violates Sanskrit
phonology:

√ tyaj → tyaktum (to renounce; abandon)

√ gam → gantum (to go)

With √ tyaj, the root ends in a sound that Sanskrit does not permit in
word-­final position. This is not a contradiction, since Sanskrit does not con-
ceptually identify a root as a complete word. It is an element waiting to be
shaped into a word. In the formation of a word, the element must accord
with the rules of the language. Since neither -­j nor -­m can come together
with a morpheme beginning with t-­, a suitable adjustment must be made.
In the case of -­j, the palatal becomes a velar: -­k. As regards -­m, which is
the labial nasal, it changes to the nasal of the class to which the following
t-­ belongs: this is -­n. Practically speaking, the student of Sanskrit learns
the infinitives in their fixed form as and when they are encountered. There
seems little point, under the circumstances, aiming for a set of rules which
would be more of a burden than a benefit.
134 The Sanskrit verb

7.6 The gerund

The gerund is frequently used in Sanskrit and merits investigation. The
term ‘gerund’ may not be too enlightening, even to those who are famil-
iar with grammatical terminology. In English, it signifies a verb form
which functions as a noun (e.g. my coming to this party was a mistake;
rage against the dying of the light). It can be outrightly adjectival in
nature, as can be seen in the phrases driving instructor and singing les-
son. In Sanskrit, it is formed from a verbal root, as with the infinitive,
and by the addition of a suffix. Often, the Sanskrit gerund is translated
by the English present participle having (e.g. having written; having
renounced; having gone). Unlike the finite verb, and in common with
the infinitive, the gerund is not conjugated. This means that no personal
endings are applied to the gerund. Once the appropriate suffix is added,
the gerund is complete.
The gerund takes the form of the suffix -­tvā or -­itvā. As with the infini-
tive -­tum and -­itum, there may be an assumption that, where the verbal
root ends in a vowel, one predicts the use of -­tvā and, where the root ends
in a consonant, -­itvā is used. That is a useful rule of thumb, although there
are too many exceptions for a more accurate set of rules to be postulated. If
one takes the verbs explored in the infinitive (√ likh; √ tyaj; √ gam), one
predicts the following gerunds:

√ likh → likhitvā (having written)

This is a correct deduction. The -­itvā suffix is used. Note also that San-
skrit permits lekhitvā, where the vowel in the root appears to have under-
gone mutation to guṇa grade, although this is not a feature of the formation
of the gerund. So too, lekhitum is also attested in the infinitive.

√ tyaj → tyaktvā (having renounced; abandoned)

By analogy to the infinitive, this is also a correct deduction. Note that the
-­j of the verbal root is velarized to -­k, avoiding the cluster -­jt.

√ gam → gatvā (having gone. Not *gantvā)

Here, the -­m in the verbal root is dropped, resulting in ga-­. The suffix
-­tvā is then added to this truncated root, yielding gatvā. Note that the -­m is
not dropped in the infinitive but, rather, subject to a sound change:
 The Sanskrit verb 135

√ gam + tum → gantum (to go)

Irregularities aside, the gerund is an easy form to spot in Sanskrit. The

ending – (i)tvā is distinctive. This is, however, complicated by the fact that
the gerund takes the form of the suffix -­ya when the verbal root contains
a prefix. As noted in 7.4, a prefix can modify the meaning of a verb quite
considerably:

Prefix abhi-­+ √ likh → abhilikhya (having engraved; inscribed)

The sense of writing has been maintained but refined to indicate a spe-
cific type of writing. The suffix -­ya replaces -­itvā because of the prefix.

Prefix pari-­+ √ tyaj → parityajya (having left; quit)

There is no change in the essential meaning of the verb because of pre-

fixation. The appropriate -­ya suffix is employed.

Prefix ā-­+ √ gam → āgamya (having come)

The verb has undergone quite a transformation as a result of prefixa-

tion. Note that the prefix ā-­is not to be assumed as one which reverses the
action of the verb. Prefixes are very often verb specific in terms of how they
modify meaning. A proper understanding of prefixes must come with the
reading of texts in Sanskrit. Attempting to give a bare meaning to individual
prefixes would be a red herring.
The gerund can neither begin a sentence nor finish it. Its use is con-
siderable with respect to giving information relating to the actions of the
subject of the sentence prior to a finite verb making the sentence gram-
matically complete. Gerunds are best thought of as ‘place holders’ build-
ing up a body of information whilst remaining grammatically free of the
subject or the object(s) of the sentence. Consider the following, stilted
sentence in English, which mimics how the sentence would be structured
in Sanskrit:

[To the orchard having gone,] [apples having seen,] [these having eaten,]
the mule was happy.

The first line in the example contains three gerunds, underlined. In all
cases, the gerund is not at the start of the phrase (indicated by brackets) but
follows either a noun or a pronoun. This is a requirement in Sanskrit. None
136 The Sanskrit verb

of the phrases equate to a complete grammatical sentence. The second line

of the example, containing the subject and the finite verb, is a grammati-
cally complete sentence.

7.7 Exercise: sentences for translation

In preparation for tackling sentences in Sanskrit (Chapter 9) and to consoli-
date the current chapter, the reader is asked to attempt the following eight
sentences, where sandhi has not been applied. New vocabulary is given
under the relevant sentence. Translations, together with analyses of the sen-
tences and a discussion of sandhi (where applicable), appear overleaf.

Sentences

(i) अहं पत्रं लिखामि ।

ahaṃ pattraṃ likhāmi
(pattra noun (n): a letter)

(ii) सूतः कमलानि चोरयति।

sūtaḥ kamalāni corayati

(iii) कन्याः उद्याने नृत्यन्ति।

kanyāḥ udyāne nr̥tyanti
(kanyā noun (f): girl)

(iv) तम् अश्वं हन्तुं न धर्म्यं भवति।

tam aśvaṃ hantuṃ na dharmyaṃ bhavati
(dharmya adjective: righteous)

(v) पाण्डवाः कु रुक्षेत्रं गन्तुम् इच्छन्ति।

pāṇḍavāḥ kurukṣetraṃ gantum icchanti
(√ iṣ, Class IV → Stem iccha-­: to wish; desire)

(vi) नगरं गत्वा कृ ष्णः पुत्रान् अवदत्।

nagaraṃ gatvā kr̥ṣṇaḥ putrān avadat
(√ vad, Class I → Stem vada-­: to speak; address)

(vii) तद् उक्त्वा रामः अगच्छत्।

tad uktvā rāmaḥ agacchat
 The Sanskrit verb 137

(viii) ग्रामं त्यक्त्वा/परित्यज्य जनाः सं तुष्टाः अभवन्।

grāmaṃ tyaktvā/parityajya janāḥ saṃtuṣṭāḥ abhavan
(saṃtuṣṭa past passive participle: pleased; contented)

Translations
(i) ahaṃ pattraṃ likhāmi
(No sandhi changes)
I am the one who is writing a/the letter.
It is clear, from the personal ending on the verb, who is doing the writ-
ing. The personal pronoun aham is not strictly speaking required.
Sanskrit allows a pronoun to be dropped, unlike English, where the
meaning of a sentence is clear. If the pronoun occurs, when it not
required for grammatical purposes, this is usually indicative of empha-
sis. The point should be made, all the same, that Sanskrit verse often
uses pronouns to ensure that there are the requisite number of syllables
for the metre in question.
(ii) sūtaḥ kamalāni corayati
(No sandhi changes)
The charioteer steals (the) lotuses.
A straightforward sentence which has the preferred word order of an
active construction in Sanskrit: subject – object – verb. Being heav-
ily inflected, Sanskrit permits considerable freedom in word order.
Nowhere is this more evident than in poetry, where word movement
allows for metrical requirements to be met.
(iii) kanyāḥ udyāne nr̥tyanti
(With sandhi: kanyā udyāne nr̥tyanti)
The girls dance in the garden.
As with sentence (ii), this sentence is straightforward. Note that the
word for garden (udyāna) is in the locative case, indicating where the
action is taking place.
There is one sandhi change (kanyāḥ → kanyā). This will be investi-
gated further in 8.2. Effectively, the ending -­āḥ, when followed by a
voiced phoneme (remembering that all vowels are voiced) results in
the dropping of the visarga.
(iv) tam aśvaṃ hantuṃ na dharmyaṃ bhavati
(No sandhi changes)
It is not righteous to kill that horse.
138 The Sanskrit verb

This sentence contains an infinitival phrase (to kill that horse). The
subject of the sentence is it (i.e. it is the thing which is not righteous)
but a word for it is not present. Otherwise said, the subject is not
overtly marked. The adjective, dharmya, qualifies the empty subject
(it is not righteous) and accordingly takes a neuter ending. A pronoun
can be dropped in Sanskrit whereas English does not permit such a
thing. If there is an empty subject in English, a dummy subject is pro-
vided (e.g. It is raining). In this type of construction, Sanskrit deems
the empty subject to be neuter.
(v) pāṇḍavāḥ kurukṣetraṃ gantum icchanti
(No sandhi changes)
The Pāṇḍavas wish to go to Kurukṣetra.
Note the position of the infinitive (gantum). It follows the place to
which the Pāṇḍavas wish to go. The finite verb (they wish) is at the
end of the sentence. The object (kurukṣetra) is between the subject
and the verb. One does not use the dative case to indicate movement
towards something or someone. That is a property of the accusative.
The English preposition to is potentially misleading.
(vi) nagaraṃ gatvā kr̥ṣṇaḥ putrān avadat
(No sandhi changes)
Having gone to the town, Kr̥ṣṇa spoke to the/his sons.
As with the infinitive, the gerund does not come at the start of a sen-
tence or phrase. The main sentence is kr̥ ṣṇaḥ putrān avadat, and the
finite verb is the third person singular imperfect, active voice. The
sons – who may be Kr̥ ṣṇa’s, in the right context – are three or more in
number, not just two (That would require the dual).
(vii) tad uktvā rāmaḥ agacchat
(With sandhi: tad uktvā rāmo ’gacchat)
Having said this, Rāma went.
  As with sentence (vi), the gerund is not the first word in either the
sentence or the gerundival phrase (tad uktvā). A tat-­form is deployed
to indicate the thing which has been uttered by Rāma. The thing is
an it, rather than a him or a her, so the relevant tat-­form is the neuter
accusative singular.
  A paradigm for the verb to go is given in 7.2, but in the present
tense. Here, the verb form is in the imperfect (third person singular,
active voice). If the augment (a-­) is removed, together with the per-
sonal ending for the third person singular (-­t), what remains is the
stem: gaccha-­. The stem is visible in all forms of the present tense and
can be deduced by removing the personal endings.
 The Sanskrit verb 139

  The combination of a final -­aḥ followed by an initial a-­ represents

the most complex example of visarga sandhi. Sandhification takes two
steps. First, the -­aḥ does not simply drop the visarga but changes to -­o;
second, the initial a-­ of the following word is not pronounced. It dis-
appears and its disappearance is marked with a sign called avagraha
(see 8.2).
(viii) grāmaṃ tyaktvā/parityajya janāḥ saṃtuṣṭāḥ abhavan
(With sandhi: grāmaṃ tyaktvā/parityajya janāḥ saṃtuṣṭā abhavan)
Having forsaken the village, the people were happy.
  There is a choice of gerund, with an unprefixed verb (in which
case, the -­(i)tvā suffix is used) or a prefixed verb (where the gerund
takes the form of the suffix -­ya). As with sentence (iii), the ending -­āḥ
in saṃtuṣṭāḥ drops the visarga when followed by a vowel, which is
voiced. There is no dropping of the visarga in janāḥ, since it is fol-
lowed by a sibilant – and all sibilants are voiceless.
  A separate word needs to be said about the past passive participle
which, whilst outside of the scope of the present work (along with
a number of other participles that the language possesses), is very
widely used in Sanskrit. It is formed from a verbal root and has the
suffix – (i)ta. Quite unlike the gerund, it is subject to endings. These
are not conjugational endings but declensional ones. In this respect,
the past passive participle agrees with the noun with which it is asso-
ciated. The past passive participle is most often adjectival in nature,
as in the sentence in question: saṃtuṣṭāḥ qualifies the word janāḥ.
The -­ta suffix, as a result, agrees with respect to case, number and
gender. It takes the nominative plural masculine ending, -­āḥ: -­tāḥ.
This is subject to cerebralization (-­tāḥ → -­ṭāḥ) because it is preceded
by the cerebral sibilant, ṣ. The past passive participle can also substi-
tute for a finite verb, whilst a gerund and an infinitive cannot. Given
its importance, the past passive participle will be explored in greater
detail in the reading passages (9.3.1).
8
SOUND CHANGES
Sandhi

8.1 Preliminaries
Sanskrit places great emphasis on the sound system, as one might expect
of a language in which recitation and the oral tradition have prevailed for
millennia. The Devanāgarī aims to reproduce as faithfully as possible the
sounds of the language, but all languages change over time and there are
instances where the true nature of certain sounds are subject to an interpre-
tation which, in due course, becomes ‘traditional’. This is the case with r̥ , r̥ ̄
and l̥ as well as the articulation of visarga and anusvāra. Given the impor-
tance accorded to sound, a series of rules were posited in antiquity. These
are collectively known as sandhi: an account of the adjustments which take
place in sounds when they come together. The word itself is an example of
what it addresses: prefix sam-­(with; together with) + √ dhā (to place; bring
together).
Sandhi is not unique to Sanskrit. Everyday English bristles with what are
nothing other than sandhi features. The plural morpheme -­s, for example,
may either sound like a voiceless sibilant (the ‘s’ in snake) or its voiced
equivalent (the ‘z’ in zebra). The words cats and dogs are examples of these
sounds, respectively. From a phonological point of view, the distinction is
an easy one: the voiceless s is heard after a voiceless sound (cat) whereas
the voiced sibilant occurs when it follows a voiced sound (dog). English
spelling does not reflect this difference in sound, placing the emphasis
on writing and leaving sandhi features to arise as they will. Even though
DOI: 10.4324/9780429325434-­9
 Sound changes 141

handbag contains a consonant cluster which reduces in everyday speech

to *hanbag and, from there, to *hambag, English preserves the spelling.
Consider, also, how unlikely one is likely to hear all the phonemes in the
word sixth (i.e. s + i + k + s + th); yet the more representative spelling *sikth
would be deemed a spelling error.
In Sanskrit, sandhi features have an impact on spelling. They are per-
vasive and must be grasped by the learner, since no authentic example of
Sanskrit is ‘unsandhified’. Reducing or withholding sandhification for the
learner is the natural and arguably the more sensible approach. With sandhi,
words can (and do) change to the point where endings suddenly become
invisible, leaving the learner confused and disorientated. An investigation
of sandhi ought not, however, to be deferred for very long.

8.2 Visarga sandhi

Sandhi is best approached as a phenomenon which affects particular catego-
ries of sound. With regard to this, the methodical arrangement of the sound
system is of considerable assistance. An example of a sound change was
seen with the word Buddha (5.2), where voice was shown to be a trigger
for the sound change in question. This is important to bear in mind, since the
notion of voice is central to the phenomenon of sandhi.
A good place to start exploring sandhi is not as it affects consonants
but as it affects visarga. Many declensional and conjugation forms end
in visarga, and the first mention of sandhi was made as and when it was
encountered in the sentence: Rāmaḥ aśvaṃ paśyati (6.2). This was subject
to a sandhi change which resulted in Rāmo ’śvaṃ paśyati. Once the basis
of visarga sandhi is understood, it becomes relatively easy to see why con-
sonants change. Visarga sandhi is, therefore, the first port of call.
Visarga cannot occur at the start of a word, so it follows that visarga
sandhi relates to the changes that take place at the end of words, as they
encounter a following word. In essence, visarga sandhi states that a visarga
is subject to change when it is followed by a voiced sound. It is crucial that
the reader is now fully able to identify the sounds explored and classified in
Chapter 3, since an understanding of which phonemes are voiced and which
are voiceless goes to the heart of understanding sandhi. Consonants may be
voiced or voiceless, but nasals are always voiced, as are semivowels. As for
the sibilants, these are all voiceless. Vowels are voiced, without exception.
If this much is understood, the majority of visarga sandhi outcomes can be
rationalized. Returning to the sentence which appears in 6.2, one may now
apply a visarga sandhi rule.
142 Sound changes

Before sandhi: Rāmaḥ aśvaṃ paśyati

The ending -­aḥ is followed by a vowel (which is voiced). The visarga

is subject to change. In this case, -­aḥ is reduced to -­o but there is a second-
ary effect, also: the initial a-­of aśvaṃ disappears and is replaced by a sign
called avagraha, resembling a handwritten capital S, which has no pronun-
ciation. In the IAST, avagraha is indicated by an apostrophe.

रामो ऽश्वं पश्यति

Rāmo ’śvaṃ paśyati
Rāma sees the horse.

What if Rāmaḥ had been followed by a word starting with a voiceless

sound? The following sentence shows the outcome in this event, using
khara (mule) rather than aśva (horse):

रामः खरं पश्यति

Rāmaḥ kharaṃ paśyati
Rāma sees the mule.

There is no change to the ending -­aḥ, since voice is absent in the follow-
ing phoneme (kh) and the trigger for visarga sandhi is missing. There are
ten voiceless consonants in Sanskrit, not counting the three sibilants and the
aspirate, all of which are voiceless. One might assume that, when followed
by any of these phonemes, visarga sandhi would not come into operation.
Whilst predicated on good logic, that is not the case.
Words ending in a visarga are subject to visarga sandhi changes when
followed by voiced phonemes or by the following three pairs of consonants:
c-­, ch; ṭ-­, ṭh-­; t-­, th-­. Here, the following changes take place:

-­ḥ + c-­, ch-­= ś-­

-­ḥ + ṭ-­, ṭh-­ = ṣ-­
-­ḥ + t-­, th-­ = s-­

This is the case irrespective of whether the visarga is preceded by a short

a, a long ā or a vowel which is neither of those (or indeed a diphthong). It is
worth noting that visarga may optionally assimilate to one of the sibilants.
 Sound changes 143

For example, in Rāmaḥ sīdati (Rāma sits/is sitting), the visarga may remain
as it is or be assimilated to the following sibilant: Rāmas sīdati. The learner
is more likely to see the former option being taken in textbooks aimed at the
student of Sanskrit. The latter option is, nevertheless, to be borne in mind
when Sanskrit texts are consulted in the original. Adjustment to the strategy
of assimilation is not problematic.
If one now consults a table of visarga sandhi changes, the regularity of
the application of visarga sandhi can easily be seen. Overleaf is just such
a table, where visarga is grouped according to whether it is preceded by a
short a, a long ā or any other vowel, for which a smiley symbol is used. The
patterning between -­aḥ, -­āḥ, and -­ḥ is identical in all cases where the fol-
lowing phoneme is voiceless. It is only when visarga is followed by voiced
phonemes (i.e. the last two rows) that differences exist.

Visarga sandhi
-­aḥ -­āḥ -­ḥ Following phoneme

-­aḥ -­āḥ -­ḥ k-­, kh-­, p-­, ph-­

-­aś -­āś -­ś c-­, ch-­

-­aṣ -­āṣ -­ṣ ṭ-­, ṭh-­

-­as -­ās -­s t-­, th-­

-­aḥ or -­āḥ or -­ḥ or

ś-­, ṣ-­, s-­
– aś/-­aṣ/-­as – āś/-­āṣ/-­ās – ś/-­ṣ/-­s
Voiced consonants,
-­o -­ā -­r1
semivowels and h-­2

-­a3 -­ā -­r vowels and diphthongs

Notes
1 When followed by r-­, -­iḥ and -­uḥ drop the visarga and lengthen to -­ī
and -­ū, respectively.
2 Although technically voiceless, the phoneme h acts as if it were voiced.
3 The rule applies to every vowel except a following short a-­. Before
short a-­, -­aḥ goes to -­o and avagraha replaces the following a-­.
144 Sound changes

रामः अवदत् → रामोऽवदत्

Rāmaḥ avadat → Rāmo ’vadat
Rāma spoke/said.
As seen with the pre-­sandhified sentence, Rāmaḥ avadat, the visarga is
dropped and the exposed short -­a at the end of Rāma changed to -­o. As the
visarga sandhi table indicates, that is a strategy which is not adopted in other
instances of visarga sandhi. Words ending in -­āḥ have a simpler strategy
in which the visarga is dropped, there is no mutation of the long -­ā and no
deployment of avagraha:

कन्याः अवदन् → कन्या अवदन्

kanyāḥ avadan → kanyā avadan
(The) girls spoke/said.

Note that the verb has changed, due to number (It is the third person plu-
ral of the imperfect, active voice, since kanyāḥ is in the nominative plural).
With -­āḥ, it does not matter if the phoneme which follows is a voiced con-
sonant, a semivowel or a vowel (Remember that h acts as if it were voiced).
In all cases where the following phoneme is voiced (or is h), the visarga is
dropped and that is the end of it.
When the visarga is preceded by a vowel other than a or ā, the strategy is
different again. Here, the rule is that the visarga changes to r.

अग्निः अवदत् → अग्निर् अवदत्

Agniḥ avadat → Agnir avadat
Agni spoke/said.

(The present work has not looked at the declension of short i-­stems. Suf-
fice it to say that the nominative singular ending for Agni is Agniḥ.)
There is an exception to this rule when the word which follows -­ḥ
begins with r-­ and this is subject to a note on the accompanying visarga
sandhi table. This exception aside, -­ḥ acts the same way in all cases where
the following phoneme is voiced (or is h). The visarga becomes r.
It is important to note that sandhi is a mechanical operation in Sanskrit.
Once visarga has been applied according to the rules specified, the output is
fixed. There is no recombination after sandhi, however much it may appear
 Sound changes 145

that further modification is required. This point will become particularly

relevant when vowel sandhi is investigated (8.4).

8.3 Consonant sandhi

As with visarga sandhi, the overarching criterion with consonant sandhi is
the distinction between voiced and voiceless phonemes. There are remark-
ably few non-­vocalic phonemes which may occur at the end of a word in
Sanskrit. These are -­k, -­ṭ, -­t, -­p, -­ṅ, -­n, -­m. Except for the nasals, they are
voiceless sounds (Visarga may, of course, also occur at the end of a word.
One may assume that visarga, although not usually subject to analysis in
terms of voice, is a voiceless sound according to the changes made – and
not made – with visarga sandhi). The final phoneme in a word, if voiceless,
acquires voice through the application of consonant sandhi if it is followed
by a word beginning with a voiced phoneme. A good example of this –
pertinent, given the contents of Chapter 9 – is the word Bhagavadgītā, from
bhagavat (divine; holy) + gītā (song). The t is voiceless and is in contact
with the voiced g of the following word. This causes the voiceless word-­
final phoneme (t) to acquire voice (d).
Unlike the changes that take place with visarga sandhi, it is not suggested
that the learner should memorize all the possible outcomes of consonant
sandhi. Sandhi tables (or grids) exist for reference purposes. One ought,
nevertheless, to know how to use a sandhi table and how to make sense of
the data it contains. With bhagavat + gītā, the relevant information is found
by identifying the word-­final phoneme from the list of those which appear
horizontally at the top of the table. The initial phoneme of the following
word appears in the column to the right of the table. Where the entry in the
top row intersects with the entry to the right of the table, the output is in the
body of the table. The output for the -­d-­in Bhagavadgītā is stated under t-­,
on the third row (corresponding with g-­, gh-­).
A glance at the table reveals that, where a word ends in -­k, -­ṭ, -­t, or -­p (i.e.
all the voiceless permitted final phonemes), these stay the same when followed
by a voiceless phoneme, with only three exceptions: -­t followed by c-­/ch-­is -­c;
-­t followed by ṭ-­/ṭh-­is -­ṭ; -­t followed by – ś is -­c. With word-­final -­ṅ and -­m,
the situation is even more straightforward. There are, in effect, no changes at
all, beyond the first row (when they are followed by vowels). Anusvāra is pre-
sent in all cases of word-­final -­m unless it is followed by a vowel. The situation
is more complex with word-­final -­n and with word-­final phonemes followed
by n-­/m-­, l, ś and h. Practice in reading Sanskrit is the only way to become
familiar with consonant sandhi forms which do not conform to a pattern.
146 Sound changes

Consonant sandhi
-­k -­ṭ -­t -­p -­ṅ -­n -­m
1 1
-­g-­ -­ḍ-­ -­d-­ -­b-­ -­ṅ(ṅ)-­ -­n(n)-­ -­m-­ Vowels
-­k-­ -­ṭ-­ -­t-­ -­p-­ -­ṅ-­ -­n-­ -­ṃ-­ k-­/kh-­
-­g-­ -­ḍ-­ -­d-­ -­b-­ -­ṅ-­ -­n-­ -­ṃ-­ g-­/gh-­
-­k-­ -­ṭ-­ -­c-­ -­p-­ -­ṅ-­ -­ṃś-­ -­ṃ-­ c-­/ch-­
-­g-­ -­ḍ-­ -­j-­ -­b-­ -­ṅ-­ -­ñ-­ -­ṃ-­ j-­/jh-­
-­k-­ -­ṭ-­ -­ṭ-­ -­p-­ -­ṅ-­ -­ṃṣ-­ -­ṃ-­ ṭ-­/ṭh-­
-­g-­ -­ḍ-­ -­ḍ-­ -­b-­ -­ṅ-­ -­ṇ-­ -­ṃ-­ ḍ-­/ḍh-­
-­k-­ -­ṭ-­ -­t-­ -­p-­ -­ṅ-­ -­ṃs-­ -­ṃ-­ t-­/th
-­g-­ -­ḍ-­ -­d-­ -­b-­ -­ṅ-­ -­n-­ -­ṃ-­ d-­/dh-­
-­k-­ -­ṭ-­ -­t-­ -­p-­ -­ṅ-­ -­n-­ -­ṃ-­ p-­/ph-­
-­g-­ -­ḍ-­ -­d-­ -­b-­ -­ṅ-­ -­n-­ -­ṃ-­ b-­/bh-­
-­ṅ-­ -­ṇ-­ -­n-­ -­m-­ -­ṅ-­ -­n-­ -­ṃ-­ n-­/m-­
-­g-­ -­ḍ-­ -­d-­ -­b-­ -­ṅ-­ -­n-­ -­ṃ-­ y-­/r-­/v-­
2
-­g-­ -­ḍ-­ -­l-­ -­b-­ -­ṅ-­ -­ṃ-­ -­ṃ-­ l-­
-­k-­ -­ṭ-­ -­c-­3 -­p-­ -­ṅ-­ -­ñś-­4 -­ṃ-­ ś-­
-­k-­ -­ṭ-­ -­t-­ -­p-­ -­ṅ-­ -­n-­ -­ṃ-­ ṣ-­/s-­
5 5 5 5
-­g-­ -­ḍ-­ -­d-­ -­b-­ -­ṅ-­ -­n-­ -­ṃ-­ h-­

Notes
1 If the vowel before -­ṅ or -­n is short, the nasal is doubled.
2 There is an optional change of -­ṃ to a nasalized -­l (-­l ̃). This appears in
the Devanāgarī as the semivowel l with a symbol above it: लँ
3 The final -­t changes to -­c, and the following ś-­changes to ch-­.
4 There is an optional change to -­ñch-­(i.e. ś-­changes to ch-­).
5 The h-­ changes to gh-­, ḍh-­, dh-­ and bh-­, which results in the clusters
– ggh-­, -­ḍḍh-­, -­ddh-­and -­bbh-­, respectively.
 Sound changes 147

8.4 Vowel sandhi

Since all vowels are voiced, the issue of voice is not pertinent in vowel san-
dhi. Given that they are far less numerous than the consonants, the vowels are
more straightforward as regards sandhi. Only two points need to be borne in
mind since, between them, they provide the rationale for the greater number
of vowel sandhi outcomes. First, vowels that have the same quality, irrespec-
tive of length, produce a long vowel of the same quality. For example, -­a/-­ā at
the end of a word, followed by a word beginning with a-­/ā-­produces -­ā-­; and
so on and so forth for all vowels (The diphthongs have a different strategy).
Second, Sanskrit does not permit two vowels to be adjacent to each other. The
second of these points contains a caveat, however. The operation of sandhi
can result in a word ending in a vowel being followed by a word beginning
in a vowel. This was seen with visarga sandhi in the example (8.2): kanyāḥ
avadan → kanyā avadan. It is important to remember that once sandhifica-
tion has taken place, whatever the outcome, there can be no recombination.
The first four columns of the vowel sandhi table are straightforward and
predictable. Columns two, three and four contain a single stipulation: a
vowel appears in its long form when followed by a vowel of the same qual-
ity and, where this is not the case, the relevant semivowel appears:

– i/-­ī → -­y-­
– u/-­ū → -­v-­
– r̥ → -­r-­

The first column greatly resembles the rules of vowel mutation discussed in
7.2. When -­a or -­ā is followed by the simple vowels i-­/ī-­, u-­/ū-­and r̥ -­, one sees
what are effectively the guṇa forms -­e-­, -­o-­and -­ar-­, respectively, with vr̥ ddhi
forms occurring when -­a or -­ā are followed by the complex vowels e-­and o-­,
and the diphthongs. The fifth, sixth, seventh and eight columns contain the san-
dhi forms most likely to challenge the learner. There is, nevertheless, a pattern
worth pointing out. As regards word-­final -­e and -­o, except for where these are
followed by a short a-­(where avagraha is deployed), the vowel sandhi change
is to a short final -­a, which is then followed by the unmodified vowel of the fol-
lowing word. The Devanāgarī leaves a space between the words, as reflected in
the IAST. Word final -­ai copies this strategy but has a long final -­ā in all cases.
Word-­final -­au adds the phoneme v to the end of a final long -­ā (-­āv) although
no space is left between the words on this occasion.
148 Sound changes

Vowel sandhi
-­a/-­ā -­i/-­ī -­u-­/-­ū -­r̥ -­e -­o -­ai -­au
-­ā-­ -­ya-­ -­va-­ -­ra-­ -­e ’-­ -­o ’-­ -­ā a-­ -­āva-­ a-­
-­ā-­ -­yā-­ -­vā-­ -­rā-­ -­a ā-­ -­a ā-­ -­ā ā-­ -­āvā-­ ā-­
-­e-­ -­ī-­ -­vi-­ -­ri-­ -­a i-­ -­a i-­ -­ā i-­ -­āvi-­ i-­
-­e-­ -­ī-­ -­vī-­ -­rī-­ -­a ī-­ -­a ī-­ -­ā ī-­ -­āvī-­ ī-­
-­o-­ -­yu-­ -­ū-­ -­ru-­ -­a u-­ -­a u-­ -­ā u-­ -­āvu-­ u-­
-­o-­ -­yū-­ -­ū-­ -­rū-­ -­a ū-­ -­a ū-­ -­ā ū-­ -­āvū-­ ū-­
-­ar-­ -­yr̥-­ -­vr̥-­ -­r̥̄-­ -­a r̥-­ -­a r̥-­ -­ā r̥-­ -­āvr̥-­ r̥-­
-­ai-­ -­ye-­ -­ve-­ -­re-­ -­a e-­ -­a e-­ -­ā e-­ -­āve-­ e-­
-­au-­ -­yo-­ -­vo-­ -­ro-­ -­a o-­ -­a o-­ -­ā o-­ -­āvo-­ o-­
-­ai-­ -­yai-­ -­vai-­ -­rai-­ -­a ai-­ -­a ai-­ -­ā ai-­ -­āvai-­ ai-­
-­au-­ -­yau-­ -­vau-­ -­rau-­ -­a au-­ -­a au-­ -­ā au-­ -­āvau-­ au-­

Notes
Avagraha (indicated in the IAST by an apostrophe) only ever replaces a
word-­initial, short a-­ when the previous word ended in either -­e or -­o. It
does not occur elsewhere. If it is encountered in a Sanskrit text outside of
this context, this is to be treated as manuscript or typing error. In the case of
Rāmo ’śvaṃ (6.2; 8.2), the use of avagraha is triggered at the same time as
the visarga change Rāmaḥ → Rāmo. It is not a change which takes place
after sandhification but, rather, during it.
It is difficult to say which vowel sandhi changes are the most frequent.
Certainly, it is not unusual for a word to end in -­a/-­ā (a gerund, for example,
in the case of -­ā) and to be followed by a word beginning with a-­/ā-­, i-­/ī-­,
u-­/ū-­or e-­. It is worth the learner spending a little time on the first column
of vowels sandhi forms as a result.

8.5 Cerebralization
The exploration of sandhi so far has looked at the sound changes which
take place when words come into contact with each other within a
 Sound changes 149

phrase or sentence. Sanskrit textbooks routinely refer to this process

as ‘external sandhi, in that these are complete words and sandhi affects
word endings (and occasionally the start of the words which follow).
There is also the phenomenon of ‘internal sandhi’, in which sandhifi-
cation takes place within a word rather than at its edges. For a learner
who is only starting his or her journey into the Sanskrit, an investigation
of internal sandhi features can be deferred until external sandhi has, at
least conceptually, been grasped. There are, nevertheless, two instances
of internal sandhi which merit a mention at an early stage. These take
the form of the cerebralization of the phonemes n and s. The phoneme n
is the nasal of the dental series and s is, similarly, classified as a dental
sibilant (That is technically not the case, since to pronounce s with the
tongue touching the teeth would be to produce the English th-­sound, as
is the case in Spanish – at least in Castilian pronunciation – when c is
followed by i or e).
If dental n and s are transformed by internal sandhi into cerebral
sounds, that is to say that they change to ṇ and ṣ, respectively, without
influence from the following word. When the declension of niśā and
rātrī were introduced (6.4), the genitive and locative plural forms were
cited: niśānām, niśāsu; rātrīṇām, rātrīṣu, respectively. The word
rātrī contains both examples of cerebralization whereas niśā exhibits
neither. The rules of cerebralization as regards the internal sandhi s →
ṣ are not overly complicated; the same can certainly not be said of n →
ṇ. For this reason, the learner is urged simply to acknowledge that they
exist and can be referred to, as and when it becomes necessary to do so.
On the following page, a set of rules is given for both instances of cere-
bralization, together with examples of nouns containing the instrumen-
tal singular ending -­ena to test the rules relating to the cerebralization
of n. This is aimed at pre-­empting a question which arises quite early
in the study of Sanskrit, the moment that a student is presented with an
instrumental singular declension exhibiting the cerebral ṇ. Textbooks
tend not to avoid giving sample paradigms which contain cerebraliza-
tion. That said, it is good to know from the outset why cerebralization
takes place. It suffices, by way of consolidation, to subject the declen-
sions for niśā and rātrī (6.4) to the following sets of rules and to see
why rātrī, in the plural, contains both instances of cerebralization and
niśā does not.
150 Sound changes

Cerebralization: n → ṇ

The phoneme -­n-­changes to -­ṇ-­when three conditions are met:

1 -­n-­is preceded by ṛ, ṝ, r or ṣ at any point in the word.

2 There is no intervening phoneme which can inhibit the change.*
3 -­n-­is followed by any vowel, n, m, y or v.

* The intervening phonemes which block cerebralization are any of the

palatal, cerebral or dental consonants, l, ś or s.
Example 1: mārgeṇa
The instrumental singular ending -­ena is subject to cerebralization
because all three conditions are met: n is preceded by r; n is followed by a
vowel; g is not an intervening phoneme able to block cerebralization.

Example 2: prabhāveṇa (prabhāva: strength; power)

The instrumental singular ending -­ena is subject to cerebralization
because all three conditions are met: n is preceded by r; n is followed by a
vowel; bh and v are not intervening phonemes.

Example 3: aśvena
The instrumental singular ending -­ena is not subject to cerebralization.
The first condition is not met by ś. Thus, even though v is not an interven-
ing phoneme and -­n-­is followed by a vowel, the process of cerebralization
does not take place.

Cerebralization: s → ṣ

The phoneme -­s-­changes to -­ṣ-­when two conditions are met:

1 -­s-­is immediately preceded in the word by k, r or any vowel other than

a or ā.
2 -­s-­is followed by any phoneme other than ṛ, ṝ or r.

Example: upaniṣad
Although this word is derived from √ sad (to sit), the phoneme s
meets both conditions the moment that the prefix ni-­ is added to it,
hence it is cerebralized in the word upaniṣad. It is not subject to cer-
ebralization in sadana (seat; dwelling place) since the first condition
is not met.
 Sound changes 151

8.6 Writing conventions

In the Devanāgarī, words may come together in five ways:

1 A word may end in a vowel and be followed by a word beginning with

a consonant.
2 A word may end in a visarga and be followed by a word which begins
with any phoneme (The following word cannot begin with a visarga).
3 A word may end in a consonant and be followed by a word beginning
with a vowel.
4 A word may end in a consonant and be followed by a word beginning
with a consonant.
5 A word may end in a vowel and be followed by a word beginning with
a vowel.

In the first such case, the words are written separately. Also, when a word
ends in a visarga, the rules of visarga sandhi apply. Consider the following:

नगरस्य जनाः अर्जुनं वदन्ति

nagarasya janāḥ Arjunaṃ vadanti
The people of the village address Arjuna.

The words nagarasya and janāḥ are written separately (A vowel is fol-
lowed by a consonant). Furthermore, according to the rules of visarga san-
dhi, the ending -­āḥ drops the visarga before a voiced phoneme.

नगरस्य जना अर्जुनं वदन्ति

nagarasya janā Arjunaṃ vadanti

Even though janā and Arjunaṃ put two vowels in adjacency, san-
dhi does not permit any recombination. Vowel sandhi is not subsequently
applied to produce *janārjunaṃ. If one substitutes the word jana with
guru (teacher), adjusting the verb to reflect a noun in the nominative sin-
gular instead of the nominative plural, one obtains the following, before
application of sandhi:
152 Sound changes

नगरस्य गुरुः अर्जुनं वदति

nagarasya guruḥ Arjunaṃ vadati
The village teacher addresses Arjuna.

Applying the rules of visarga sandhi, -­ḥ before a vowel becomes -­r.
This results in a semivowel at the end of the word guru-­, which takes a virāma:

नगरस्य गुरुर् अर्जुनं वदति

nagarasya gurur Arjunaṃ vadati

Virāma does not, however, appear in authentic Sanskrit examples unless

it is at the end of a sentence. Elsewhere, the writing conventions of the
Devanāgari call for its removal. With -­r as the word-­final phoneme, the
word combination is a semivowel followed by a vowel (The third possi-
ble writing combination, as identified on the previous page). The writing
conventions are different to the rules of sandhi in that, if a combination
is required, it is effected. Accordingly, the fully sandhified version of the
sentence appears as follows:

नगरस्य गुरुरर्जुनं वदति

nagarasya gurur Arjunaṃ vadati

Note that the IAST does not indicate the writing together of words except
in the case of vowel sandhi, where vowels become linked to the point where
the sandhification affects both words rather than just one.
In the third combination, the Devanāgarī requires words to be written
together. Consonant sandhi rules apply so, if a word ends in a voiceless
consonant and is followed by a word beginning with a vowel, the consonant
in question is voiced (The following example is from 6.7.2):

Before sandhi: बुद्धः ग्रामात् आगच्छति

Buddhaḥ grāmāt āgacchati
(The) Buddha comes from (the) village.

After sandhi: बुद्धो ग्रामादागच्छति

 Sound changes 153

Before sandhi: बुद्धः ग्रामात् आगच्छति

Buddho grāmād āgacchati

So far so good, but the situation is more complex in the fourth and fifth
combinations. Where a word ends in a consonant and is followed by a word
beginning with a consonant, a conjunct is created. Additionally, a consonant
sandhi may be required, if the word-­final consonant is voiceless and it is
followed by a word-­initial voiced consonant:

Before sandhi: बुद्धः ग्रामात् वनं गच्छति

Buddhaḥ grāmāt vanaṃ gacchati
(The) Buddha goes from (the) village to (the) wood (vana).

After sandhi: बुद्धो ग्रामाद्वनं गच्छति

Buddho grāmād vanaṃ gacchati

Vowels do not create conjuncts. They are simply subject to vowel sandhi
and to whatever adjustments are subsequently needed in the Devanāgarī.
Take the following example, containing the emphatic (and occasionally
restrictive) enclitic word eva preceded by a gerund:

Before sandhi: उक्त्वा एव

uktvā eva
having indeed been said/having just been said

After sandhi: उक्त्वैव

uktvaiva

The best way to accustom oneself to the conventions of the Devanāgarī

is to see them in operation. With this is mind, the present chapter concludes
with a mantra by way of a consolidating exercise. It is one of several peace
mantras and arguably the best known of these. With any mantra, one ought
to control one’s breathing and to aim to pronounce each line with a sin-
gle breath. At the end (śāntiḥ śāntiḥ śāntiḥ), the lungs are emptied, hence
the drawing out of the final syllable – a thing that is not possible unless
one articulates the echo vowel following the third and final śāntiḥ until the
breath is exhausted. As with the sentences, the first version is unsandhified
154 Sound changes

and the second one applies all the necessary sandhi features. The sandhified
version is the one which is chanted.

8.7 Exercise: reading a mantra

Before sandhi

ॐ असतः मा सत् गमय।

तमसः मा ज्योतिः गमय।
मृत्योः मा अमृतं गमय।
ॐ शान्तिः शान्तिः शान्तिः॥
OM asataḥ mā sat gamaya
tamasaḥ mā jyotiḥ gamaya
mr̥tyoḥ mā amr̥taṃ gamaya
OM śāntiḥ śāntiḥ śāntiḥ

OM. From non-­being to reality, make me go.

From darkness to light, make me go.
From death to immortality, make me go.
OM. Peace, peace, peace.

Vocabulary
asat noun (m) the non-­existent (ablative sg. = asataḥ).
mā pers. pronoun me (an alternative form for mām).
sat noun (m) the existent (accusative sg. = sat).
gamaya verb (finite) cause to go (second pers. sg. imperative).
tamas noun (n) darkness (ablative sg. = tamasaḥ).
jyotis noun (n) light (accusative sg. = jyotiḥ).
mr̥ tyu noun (m) death (ablative sg. = mr̥ tyoḥ).
amr̥ ta noun (m) immortality (accusative sg. = amr̥ tam).
śānti noun (f) peace (nominative sg. = śāntiḥ).
 Sound changes 155

After sandhi

ॐ असतो मा सद्गमय । 1 2

तमसो मा ज्योतिर्गमय ।
1 3

मृत्योर्मा अमृतं गमय।

3 4

ॐ शान्तिः शान्तिः शान्तिः॥

OM asato mā sad gamaya
tamaso mā jyotir gamaya
mr̥tyor mā amr̥taṃ gamaya
OM śāntiḥ śāntiḥ śāntiḥ

Notes
1 The words in question (asataḥ and tamasaḥ) both end in a visarga pre-
ceded by short a and are followed by a voiced phoneme (m). The rules
of visarga sandhi relating to -­aḥ apply: -­aḥ → -­o.
2 The word sat ends in an unvoiced consonant (t) and is followed by a
voiced consonant (g). The phoneme t is sandhified, according to the
rules of consonant sandhi, to its voiced equivalent (d). Since the result-
ing combination is -­dg-­, orthographic rules apply, generating a con-
junct consonant which has the effect of joining the words together.
3 Both jyotiḥ and mr̥ tyoḥ end in a visarga preceded by a vowel other
than a or ā. The rules of ‘smiley visarga’ apply, transforming the visarga
to r. Since r is a semivowel and is followed by a consonant, a conjunct
is produced and, as with sad + gamaya, the words are written together.
4 This is an interesting case of where both the rules of sandhi and writing
conventions are compromised. Technically, one ought to see the long
ā of mā and the following short a of amr̥ taṃ being fused, according
to vowel sandhi rules, to produce a single, long -­ā-­. This would yield
*māmr̥taṃ gamaya – highly ambiguous in that it is not distinguish-
able in sound to *mā mr̥taṃ gamaya (to mortality, make me go). To
avoid this, the Devanāgarī keeps the words mā and amr̥ taṃ separate,
thereby promoting a short pause between these words.
9
FROM SENTENCE TO TEXT

9.1 Preliminaries
Translating Sanskrit requires patience and great attention to detail. One must
be prepared to acquire a good working vocabulary on top of a sound grasp
of grammar whilst bearing in mind that, if reading the Devanāgarī rather
than the IAST, writing conventions will challenge the eye. A knowledge of
conjuncts is necessary, too, even if the text presented in the Devanāgarī uses
virāma to keep words separate to avoid conjuncts arising between them.
The great advantage of reading a text in the IAST is that, whilst it reflects
sound changes brought about by sandhi, there are no conjunct forms. Pho-
nemes which generate a conjunct in the Devanāgarī are kept separate in the
IAST, allowing the learner to see more clearly how a sentence is consti-
tuted (This is not, however, possible with vowel sandhi). The readings in
this chapter are verses taken from the first book of the Bhagavadgītā. The
verses in question appear in the Devanāgarī and are fully sandhified. The
IAST version is given immediately below the Devanāgarī, followed by a
close translation. The vocabulary for the individual verses lists the words in
their citation form, except for the verbs, since conjugation forms part of the
analysis following each verse.
With respect to the IAST accompanying the Devanāgarī, capital letters
are now avoided, reflecting the practice adopted by most Sanskrit scholars
who use the IAST. The reader can identify, from the translation, the proper

DOI: 10.4324/9780429325434-­10
 From sentence to text 157

nouns that would require a capital letter in English. There are a number of
nouns which may be masculine, neuter or feminine. These have been identi-
fied as masculine (unless they are neuter only), largely because a feminine
stem would require the final -­a to be modified to reflect an ending suitable
for a feminine stem. It is a minor point, in any event, and aimed at sim-
plifying matters for the learner without violating any grammatical rules.
Nouns are indicated with their gender in brackets: masculine (m), neuter (n)
or feminine (f). Finally, personal pronouns are indicated as such but there
are other pronouns: kim (Reading One, 9.3.1) and yat/yad (Reading Four,
9.3.4). These are discussed in the accompanying analysis. Being a work of
poetry rather than of prose, the Bhagavadgītā does not always adhere to all
the conventions discussed in the earlier chapters of the present work. Where
that is the case, it is commented upon.

9.2 Reading Sanskrit

There are various steps to take when attempting the reading of a text in the
Devanāgarī. One must first consult the appropriate vocabulary, assuming
that one has been presented (This tends to be the case with most textbooks
for Sanskrit, although by no means all). If vocabulary is not given and there
is only a dictionary or lexicon to which the student is expected to refer, an
alternative strategy suggests itself. For the learner, the text is best transliter-
ated into the IAST. This allows for maximum transparency and allows con-
juncts to be decrypted and checked in advance of investigating grammatical
endings. Once the words comprising the sentence have been transliterated
and separated out, the work of grammatical analysis can be undertaken.
It is often the case that a sentence in Sanskrit is ‘back to front’ when
compared to the English equivalent. Sanskrit tends to place additional
information relating to the sentence at the start, deferring the phrase which
contains the subject and the verb to the end of the sentence. It is not advis-
able to adopt the habit of translating a Sanskrit sentence word-­for-­word
in the sequence in which the words appear, expecting the result to make
sense. The student should attempt to find out who is doing what to whom
or where. This approach requires the identification of the subject of the
sentence and the verb. A finite verb may not be in evidence, having been
substituted by a participle (This is frequently the case with poetry). The
subject, however, is not so easily omitted. Once the subject can be iden-
tified, the endings of the other nouns can then be checked to see what
158 From sentence to text

case functions are indicated. The best approach to translating Sanskrit is

to attempt to group words into their respective phrases and to connect the
phrases in question to each other in order to build up a coherent and logical
sequence of units of meaning.
Certain words are useful in indicating the phrases which constitute a
sentence. Pronouns are an obvious example of this. If the personal pronoun
your or my/mine appear, one has then to establish the thing to which they are
referring. The longer the sentence, the more likely it is that there will be a
pronoun – either a personal pronoun or the demonstrative pronoun tat/etat.
The sentence can be envisaged as a jigsaw of phrases and, to extend the
analogy, the more one is able to identify the pieces, the easier they are to fit
together to make the picture emerge. Attempting to connect words in strict
sequence obscures the picture.
The readings which follow are challenging but intended to introduce the
reader to authentic Sanskrit. The reader should not, as a result, feel dispir-
ited if the sentences appear to be difficult. The Bhagavadgītā is not a text
which would ordinarily be presented to the beginner with any expectation
that he or she should be able to make sense of the grammar or the vocabu-
lary. The verses presented serve as a good introduction to the richness and
complexity of Sanskrit poetry. It should be borne in mind that the greater
part of the texts which have survived in Sanskrit are works of poetry, not
of prose. Early exposure to how Sanskrit patterns the sentence in poetry is,
therefore, desirable.
The opening verse of the Bhagavadgītā (Reading One) is the most com-
plex in terms of its structure and, accordingly, more space has been devoted
to its analysis than is the case with the three subsequent readings. Being the
first reading, certain features are identified and explained, such as the past
passive participle, which appear elsewhere. The reader is presented with a
surfeit of grammatical information with the first reading. Once the analysis
of Reading One has been undertaken, the second and third readings fall eas-
ily into place. Reading Four is not much more difficult than Reading Two
or Three, although it contains a compound, which is a phenomenon that has
not so far been discussed.
It is recommended that the reader look at the text in the Devanāgarī
to begin with, checking to see if his or her reading of this is accurate.
If many mistakes are made in the reading of the Devanāgarī, this is an
indication that a review of Chapter 4 is required. The IAST is used for all
 From sentence to text 159

grammatical explanations, since little point is served by presenting the

learner with grammatical concepts that are, in themselves, also exercises
in reading the Devanāgarī. One task at a time is more than sufficient. If the
reading of the Devanāgarī is accurate (albeit with the occasional hiccup)
but declensional endings are the cause of confusion, the reader is directed
back to Chapter 6. The visarga sandhi table in 8.2 is worth bookmarking,
for ease of reference.
Reading Sanskrit is a rewarding experience, and it pays dividends for
the learner to know something about a text prior to engaging with it in the
original. With that in mind, Chapter 10 offers some suggestions as regards
accessible publications relating to the Bhagavadgītā and, indeed, other liter-
ary works much cherished in the Indian cultural tradition.

9.3 Selected readings

9.3.1 Reading One

As seen in 4.4.3, this is the opening verse of the Bhagavadgītā, in which
Dhr̥ tarāṣṭra, king of the powerful Kaurava dynasty, asks his minister about
the preparations for a battle that is about to commence in Kurukṣetra – liter-
ally, in a field or stretch of land (kṣetra) belonging to the descendants of
King Kuru. Exact dates for King Kuru have not been established but, in the
Indic cultural tradition, he is the remote ancestor of both the Kaurava and
Pāṇḍava clans. It is likely that Kuru was an important chieftain in the Iron
Age, between around 1200 and 800 bce. There is no doubt that the story of
the great war between the Kauravas and the Pāṇḍavas relates to an ancient
conflict, although it has become a metaphor for the struggle to restore the
rule of dharma to the world.

धर्मक्षेत्रे कु रुक्षेत्रे समवेता युयुत्सवः।

मामकाः पाण्डवाश्चै व किमकुर्वत सं जय॥ १॥
dharmakṣetre kurukṣetre samavetā yuyutsavaḥ
māmakāḥ pāṇḍavāścaiva kim akurvata saṃjaya (Verse 1)
In the field of dharma, in Kurukṣetra, assembled and willing to fight,
What did mine (= my army) and the Pāṇḍavas do, o Saṃjaya?
160 From sentence to text

Vocabulary
dharma noun (m) dharma (see 5.8, viii).
kṣetra noun (n) field.
kuru noun (m) the name of a king/dynasty.
samaveta participle assembled.
yuyutsu adjective willing to fight.
māmaka noun (m) my/mine (noun form of mama).
pāṇḍava noun (m) descendant of King Paṇḍu.
ca enclitic and.
eva enclitic truly; indeed; really; only.
kim pronoun what; why.
akurvata verb (finite) they did.
saṃjaya noun (m) the name of Dhr̥ tarāṣṭra’s minister.

At the start of text, it is customary for Sanskrit to set the scene by locating
the place in which an event occurs or when such an event took place. In this,
the opening verse, the location is established with a locative singular ending
(kurukṣetra → kurukṣetre). The first word of the verse does not so much
establish a geographical location but a place in which justice is about to be
done and dharma restored. Although it has the same grammatical structure
as kurukṣetre, dharmakṣetre is figurative rather than literal. The rest of
the half line of the verse, up to the single daṇḍa, is where one expects a
grammatical subject to appear. Both samavetā and yuyutsavaḥ relate to the
grammatical subject although the first of these is a participle and the second
one is adjectival in nature.
Samavetā is a complex word. It is a participle composed of a verbal root
with two prefixes and one suffix, as follows: sam (prefix) + ava (prefix) + √
i (to go; come) + ta-­ (suffix indicating a past passive participle). The result-
ing form, samaveta, is then subject to declension as if it were a short -­a stem.
Visarga sandhi obscures the fact that the ending, before sandhification takes
place, is -­āḥ (samavetāḥ). This indicates the nominative masculine plural end-
ing of a short -­a stem (see 6.3). As for yuyutsavaḥ, the student needs to be able
to identify the word, before any grammar is applied, as one which ends in -­u
(The present work has not investigated -­u stems, since that is not one of the first
declensions to be studied by beginners). The ending -­avaḥ also indicates the
nominative masculine plural ending, albeit of an -­u stem rather than an -­a stem.
Both samavetā(ḥ) and yuyutsavaḥ encode the grammatical subject,
which is masculine and plural, even if ‘they’ have yet to be identified.
Before moving to the second half of the verse, however, it is essential to
explore how a participle (samaveta) and what is effectively an adjectival
 From sentence to text 161

word (yuyutsu) can function as if they were nouns, thereby capable of rep-
resenting the grammatical subject. The past passive participle has already
been seen (5.2, x and 7.7, viii). On the first occasion, it formed the noun
buddha, albeit that the word in question contained a complex sandhi fea-
ture in which the suffixal ending -­ta was voiced as a result of being adja-
cent to the voiced -­dh in the verbal root budh and acquired the aspirate of
the verbal root (i.e. -­ta changed to -­dha). As a noun, the word buddha is
then subject to all the declensional changes associated with a short -­a stem.
This can be captured in the English translation of buddha as the Awak-
ened One, which, as a noun, is subject to English morphology: the Awak-
ened Ones (with the plural -­s morpheme); the Awakened One’s (with the
s preceded by the genitive apostrophe). In this verse of the Bhagavadgītā,
samavetā(ḥ) are those who have come together, the meaning being deter-
mined by the verbal root, modified by two prefixes. It is case-­marked in
the nominative plural, just as buddha could be made nominative and plural
with the appropriate case ending: buddhāḥ, the Buddhas/Awakened Ones
(three or more of them). It is important to point out, at this juncture, that
two past passive participles can be placed together. Does this result in two
grammatical subjects? No, since the first past passive participle functions
to describe the second and thereby acts as an adjective. Consider the fol-
lowing sentence, which contains samaveta and buddha, both passive past
participles in structure:

Before sandhi: samavetāḥ buddhāḥ vadanti

After sandhi: samavetā buddhā vadanti
The assembled Buddhas speak/are speaking.

Importantly, the past passive participle can function as a verb as well as

a noun and an adjective. It is not uncommon to come across a sentence in
Sanskrit which has a past passive participle instead of a finite verb:

Before sandhi: buddhaḥ agacchat (Imperfect, third person sg.)

After sandhi: buddho ’gacchat
The Buddha went.
Before sandhi: buddhaḥ gataḥ (Past passive participle)
After sandhi: buddho gataḥ
The Buddha went.

The flexibility of the past passive participle is due to the fact that it is
derived from a verbal root. It can generate a noun (as with Buddha) and sub-
stituting for a finite verb generated from the relevant verb root (gata from √
gam, with the appropriate declensional ending).
162 From sentence to text

With respect to yuyutsu, acting as an adjective within the verse, that

too can be identified as a noun. Nouns and adjectives are words which may
relate to things or to descriptions of things. The word yuyutsavaḥ here
describes the assembled men (samavetāḥ) but, on its own, would mean
those who are willing to fight. It is the function within the sentence which
distinguishes a noun from an adjective.
In the second half verse, the grammatical subject is specified. There are
indeed two subjects: the army belonging to King Dhr̥ tarāṣṭra, referred to by
a noun form, appropriately in the plural (māmakāḥ: mine; those belonging
to me) and the opposing forces, described by Dhr̥ tarāṣṭra as the descendants
of King Paṇḍu, the pāṇḍavāḥ (In point of fact, there are many allies on
both sides also assembled on the battlefield). The ending of māmakāḥ is
absolutely clear as it has not been subject to sandhi, given that the follow-
ing phoneme is the voiceless p-­. As for pāṇḍavāḥ, that undergoes a visarga
sandhi change, as described in 8.2: -­āḥ + c-­→ -­ś. The enclitic ca (and) is a
conjunction uniting both Dhr̥ tarāṣṭra’s forces and those of the Pāṇḍavas (i.e.
X Y and), with ca forming a vowel sandhi with eva to produce caiva (Those
of mine and, indeed, the Pāṇḍavas).
The concluding part of the verse is a question: what have they (the
rival armies) done? Dhr̥ tarāṣṭra, being blind, is reliant on the eyewitness
testimony of someone capable of seeing how the armies have positioned
themselves on the battlefield. Dhr̥ tarāṣṭra uses the customary way of ask-
ing a question: kim. It is a pronoun, as is tat, but more specifically the
interrogative pronoun. Whilst it is the case that the interrogative pronoun
is subject to declension across three genders, three numbers and seven
cases (the vocative not being used), the learner can take heart that help
is at hand. The declension of kim differs to tat in one respect: whilst
tat is, strictly speaking, representing the neuter and accusative singular
declensional forms, kim has the same restriction (i.e. it is the form for the
neuter nominative and accusative singular). If one then substitutes k-­for
the initial t-­of every tat-­form (and, indeed, the initial s-­of the masculine
nominative singular saḥ and feminine nominative singular sā), the end-
ings of tat and kim are identical. The demonstrative pronoun tat is, there-
fore, the key to knowing the declension of the interrogative pronoun kim.
The finite verb form (akurvata) belongs to a conjugation that the present
work has not explored. It is the third person plural of the imperfect for
√ kr̥ (make; fashion; do; create (see 5.2, vii)). It is the conjugation of a
verb which does not belong to the thematic group of verb classes (√ kr̥
belongs to Class VIII).
The final word in the verse is the most straightforward of all. It is the
name of Dhr̥ tarāṣṭra’s minister, saṃjaya, a name which has survived in the
 From sentence to text 163

form Sanjay/Sañjay. Although the word does not appear to be case-­marked,

it is in the vocative singular which, for short -­a masculine and neuter stems,
is identical to the citation form.

9.3.2 Reading Two

The core of the Bhagavadgītā is the dialogue which takes place between the
god Kr̥ ṣṇa and Arjuna, one of the Pāṇḍavas and an archer without equal.
Arjuna has been looking around at the combatants, Kaurava and Pāṇḍava
alike, and is struck by the fact that he recognizes kinsmen and acquaintances
on both sides. In his despair, he addresses Kr̥ ṣṇa to say that his willingness
to fight is deserting him. It will be down to Kr̥ ṣṇa to advise Arjuna as to why
he must take part in the fighting and what will be put into jeopardy if he fails
to do so. Arjuna is not a coward. His courage is undoubted; but he is filled
with pity, faced with the inevitability of participating in what he knows will
result in the death of many.

सीदन्ति मम गात्राणि मुखं च परिशुष्यति।

वेपथुश्च शरीरे मे रोमहर्षश्च जायते॥ २९॥
sīdanti mama gātrāṇi mukhaṃ ca pariśuṣyati
vepathuś ca śarīre me romaharṣaś ca jāyate (Verse 29)
My limbs fail me and my mouth dries up
And my body quakes and my hair bristles.

Vocabulary
sīdanti verb (finite) they sit; fall down.
mama pers. pronoun my/mine.
gātra noun (n) leg; limb.
mukha noun (n) face; mouth.
ca enclitic and.
pariśuṣyati verb (finite) he/she/it dries up.
vepathu noun (m) a quivering; trembling.
śarīra noun (n) body.
me pers. pronoun my/mine (enclitic form).
roman noun (n) body hair.
harṣa noun (m) a bristling; standing on end.
jāyate† verb (finite) he/she/it is born; brought forth.
164 From sentence to text

† The verb form is derived from √ jan (Class IV), stem: jāya. The stem is
slightly irregular, in that one anticipates *janya. The ending is that of the
third person singular of the present tense, middle voice (i.e. -­te, not -­ti).
There are two finite verbs in the first half verse (sīdanti and pariśuṣyati).
The first of these has the third person plural ending of the present tense (-­anti);
the second has the ending of the third person singular ending of the present
tense (-­ati). The reader has only to figure out the meaning of the verb forms
in question and to identify their subjects. With sīdanti, the root is sad (to sit:
Class I, also occasionally, Class VI), the stem is sīda. The subject appears in
the following nominal phrase, containing a personal pronoun (mama gātrāṇi:
my legs/limbs). The noun is in the plural, suggesting limbs rather than two legs
(which would require the dual), with a cerebralized ṇ, due to the presence of
a preceding r in the word. The ending -­āni /-­āṇi indicates a plural form of a
neuter short -­a stem but can be either nominative or accusative. Since it identi-
fies the subject of the verb to sit, the declension is in the nominative.
The remaining part of the first half verse contains the enclitic ca, which
can be put to one side in the first instance, since it is the element which
combines two things. This leaves the reader with mukhaṃ pariśuṣyati
(the mouth dries up), which has the subject preceding the verb, as was
the case with the sentences in 7.7. The noun mukhaṃ is, again, a neuter
short -­a stem, meaning that the ending can indicate either the nominative
or accusative singular. Since it governs the verb, it is the nominative. As for
pariśuṣyati, it is a prefixed verb: pari + √ śuṣ. It belongs to Class IV, which
adopts the strategy of adding -­ya to an unmodified root to form the stem. For
a reminder as to the strategies for stem formation with thematic verbs, see
7.3: The Thematic Verbs (IV, VI, X).
The second half verse begins with an -­u stem in the nominative singular
(vepathuḥ) which is subject to visarga sandhi: -­uḥ → -­uś, as a result of the
following c-­ in ca. Despite being a subject, it does not have an associated
verb. One could argue that the verb to be has been dropped (i.e. there is a
quaking); Sanskrit regularly drops the verb to be if the meaning of the sen-
tence is clear enough without it. Equally, one could posit that it shares the
finite verb at the end of the verse. As for phrase śarīre me (in the body of
me), this is made up of a short -­a stem (again, neuter) in the locative case,
followed by the enclitic form of mama. Two words then come together to
give a compound (of which more will be said in the analysis of Reading
Four). This compound is treated as a masculine short -­a stem in the nomi-
native singular: romaharṣaḥ, which is sandhified for the same reason as
vepathuḥ. This agrees with the finite verb jāyate, for which a note is pro-
vided in the vocabulary.
 From sentence to text 165

9.3.3 Reading Three

Gāṇḍīva is Arjuna’s bow. It is a celestial weapon, given to him by the god
Varuṇa, along with two quivers containing an inexhaustible supply of
arrows. Arjuna is an archer whose arrows never miss their mark. Arjuna and
Gāṇḍīva are, consequently, much feared by the Kauravas, even though the
Kaurava army is greater in size than the Pāṇḍava forces arrayed against it.
Arjuna has a divine origin, having been born because of a mantra uttered by
his mother, Kuntī, calling on the gods to give her sons when her husband
(Pāṇḍu) could not. Arjuna is the son of Indra, one of the most important dei-
ties of Vedic times, but he is human and is subject to human feelings.

गाण्डीवं स्रं सते हस्तात्त्वक्चैव परिदह्यते।

न च शक्नोम्यवस्थातुं भ्रमतीव च मे मनः॥ ३०॥
gāṇḍīvaṃ sraṃsate hastāt tvak caiva paridahyate
na ca śaknomy+avasthātuṃ bhramatīva ca me manaḥ (Verse 30)
Gāṇḍīva slips from my hand and my skin burns
And I cannot stand and my mind wanders.

Vocabulary
gāṇḍīva noun (m) the name of Arjuna’s bow.
sraṃsate verb (finite) he/she/it slips; falls.
hasta noun (m) hand.
tvak noun (f) skin.
ca enclitic and.
eva enclitic truly; indeed; really; only.
paridahyate verb (finite) he/she/it is burned.
na negative part. not; no.
śaknomi verb (finite) I can.
avasthātum verb (infinitive) to stand; remain.
bhramati verb (finite) he/she/it wanders.
iva enclitic as if; like; in this manner.
me pers. pronoun my/mine (enclitic form).
manas noun (n) mind; intellect.

The previous verse began with a finite verb which appeared to con-
tradict what had been presented in 7.7. Sanskrit word order is flexible
166 From sentence to text

because of the detail contained in word endings. In addition, poetry per-

mits word-­ordering that is different to prose. In this verse, the finite verb
is not at the very start of the verse but nevertheless follows the subject
and precedes the object. In this respect, it accords with the word order
of English: subject – verb – object. The enclitic ca helps to break the
first half verse into two parts, as follows: gāṇḍīvaṃ sraṃsate hastāt/
tvak caiva paridahyate. Something happens and then something else
happens. Sanskrit does not tend to state the obvious. The bow, gāṇḍīva,
is a neuter short -­a stem and therefore nominative or accusative singular.
It falls (sraṃsate) – which makes it the subject and thereby nomina-
tive – from a hand (hastāt) which accordingly takes the ablative case.
Arjuna does not say my hand, nor does he need to. The context makes it
clear whose hand is involved. The finite verb sraṃsate (Class I), as with
jāyate in the previous verse, is the third person singular, middle voice
(present tense). The second part (tvak caiva paridahyate) again omits
the personal pronoun my and uses a middle voice form (prefix pari-­+ √
dah: to burn). It is a Class I verb, for which -­ya is not a stem formation
strategy. This is a passive construction, created by adding -­ya before the
appropriate personal ending in the middle voice: and truly (my) skin is
burned.
Using the enclitic ca, one is able, in the second half verse, to iden-
tify two parts: na śaknomy+avasthātuṃ/bhramatīva me manaḥ. The
first of these has the finite verb form I can (√ śak, Class V: to be able),
negated by na (I cannot). A noun or pronoun is not required, since the
ending makes it evident that the subject is the first person. This is fol-
lowed by a telltale infinitive, ending in -­tum (prefix ava-­+ √ sthā, Class
I: to be; stand). The prefix has the effect of refining the meaning of the
verb to indicate staying put or remaining in place. The remaining part of
the verse is straightforward. √ bhram (Class I: to wander; ramble) gen-
erates the stem bhrama to which the personal ending of the third person
singular, active voice (present tense) is applied: bhramati. Because this
ends in -­i and is followed by a word beginning with i-­, vowel sandhi
takes place. The subject, in the nominative singular, is manaḥ: mind (an
example of a consonant stem ending, which the present work does not
investigate). Strangely, one does not find either mama manaḥ (my mind)
or mano me, which would be the sandhified outcome of manaḥ me.
The enclitic me appears before the noun to which it relates, which is not
in accordance with grammatical rules. Here, poetic considerations take
precedence.
 From sentence to text 167

9.3.4 Reading Four

Arjuna reflects on the matter of killing those who belong to the same clan.
It is not far-­fetched to suggest that this may reflect the ethical code of early
Aryan society with respect to whom one may legitimately fight and slay.
The Bhagavadgītā has strong echoes of an earlier age, which lends support
to a suggestion that it derives from an earlier composition dating back to
Vedic times. Arjuna obtains Gāṇḍīva from Varuṇa, Vedic god of the sky
and of the oceans, having first approached the fire god Agni for assistance.
Arjuna’s birth is the result of an invocation to Indra, chief amongst the gods
in the Vedic pantheon. With so many references to the Vedic Age, the world
of the Bhagavadgītā is not that of the Common Era. It is an old tale recast
in a later linguistic form.

अहो बत महत्पापं कर्तुं व्यवसिता वयम्।

यद्राज्यसुखलोभेन हन्तुं स्वजनमुद्यताः॥ ४५॥
aho bata mahat pāpaṃ kartuṃ vyavasitā vayam
yad rājya-­sukha-­lobhena hantuṃ svajanam udyatāḥ (Verse 45)
Oh, alas! We are resolved to do great evil
Which, by the desire for kingship, (we are) ready to kill our own people.

Vocabulary
aho interjection o(h).
bata interjection alas.
mahat adjective great.
pāpa noun (m) evil; sin.
kartum verb (infinitive) to do.
vyavasita participle resolved; decided.
vayam pers. pronoun we.
yat/yad pronoun which.
rājya noun (n) royalty; kingship.
sukha noun (m) comfort; pleasure.
lobha noun (m) covetousness; desire for.
hantum verb (infinitive) to strike down; slay; kill.
svajana noun (m) own people; kindred.
udyata participle intent on; prepared.
168 From sentence to text

The verse begins with a frequently used interjection. Other common inter-
jections are he and bhoḥ, which tend to be reserved for when the speaker
is attempting to catch the attention of another. An additional interjection
follows (bata) which, like aho, is not addressing someone in particular but
simply lamenting a situation. The reader can put these to one side to con-
centrate on finding the subject in the first half verse. The subject cannot be
pāpaṃ, since this is a masculine short -­a stem and the declensional end-
ing indicates the accusative singular. It is the object, preceded by a word
which qualifies it (mahat). It is important to note that mahat can be both
adjective (great) and noun (a great person or thing). It is a consonant stem,
as is the case with tvak and manas in Reading Three. The declined form
mahat represents the nominative, accusative and vocative singular. Here,
one selects the accusative singular since, being adjectival, it agrees in terms
of case with the thing it qualifies. This is a good example of how nouns and
adjectives belong to the same category of word. It is their function within
the sentence that determines whether they are, in a specific context, to be
considered as nouns or adjectives. From this example, it is evident that the
adjective precedes the noun which it qualifies.
The noun, adjectivally qualified (mahat pāpaṃ), is an object and cannot,
therefore, govern a finite verb. The verb form which follows (kartuṃ) is
an infinitive formed from √ kr̥ (do; make; fashion; create). This still leaves
the subject to find and, with few contenders now remaining, one focuses
on vyavasitā vayam. If one’s knowledge of personal pronouns is in place,
vayam is instantly recognizable as the nominative plural form of the first
person pronoun: we (see 6.6). Now that a subject is visible, all that remains
is to identify its relationship with the word to the left of it. As with mahat
pāpaṃ (adjective + noun), position establishes the relationship. The word
vyavasitā acts adjectivally vis-­à-­vis the pronoun, but it is technically a past
passive participle (Note that the form is vyavasitāḥ prior to sandhification).
The word is complex, being composed of two prefixes, a verbal root with an
irregular mutation and a suffix which is subject to declension: prefix 1 vi-­+
prefix 2 ava-­ + √ śri (Class I: to fix on; turn towards) → irregular form si
+ past passive participle -­ta (with the relevant nominative plural ending for
a short -­a stem). The prefixes vi-­ + ava-­ are subject to vowel sandhi (-­i +
a-­→ -­ya-­). Complexity aside, the function of the past passive participle is
straightforward. It describes a group of people whom Arjuna refers to as we,
including himself within the group: resolved/determined (are) we.
The second half verse is significantly more resistant to translation than
the first half. It contains a pronoun at the beginning (yat/yad) which is
declined like tat (or, for that matter, kim), and it is the relative pronoun,
 From sentence to text 169

expressing which who(m), whose, etc. The reader is urged to put it to one
side whilst scanning for the subject and – if one is indeed present – the finite
verb. Regrettably, there is neither an overtly marked subject nor is there a
finite verb. In such cases, it is suggested that one looks for a past passive
participle, capable of substituting for the verb. One finds such a thing, in
the form of udyatāḥ: prefix ud-­+ √ yam (Class I: to give; support) + -­past
passive participle -­ta (Note that the m in the root is lost, as is the nasal in √
jan, when forming the stem jāya, in Reading Two). Since something verbal
is required at the end of the verse, the past passive participle now adopts this
role. Courtesy of the prefix ud-­, the meaning becomes to be intent on; be
prepared. As to the identity of those who are intent or prepared, that cannot
be svajanam (own people; kindred), since that is a masculine singular noun
marked in the accusative case. The subject is, in fact, recycled from the
previous half verse. This is vayam.
Once svajanam udyatāḥ (vayam) are in place, one is left with the sense
of a body of people prepared to do something to their kindred folk. If one
then adds the infinitive hantum, from √ han (Class II: to strike down), the
action is specified: we are prepared to kill (our) own people. This leaves
only the long word rājya-­sukha-­lobhena to accommodate. The word is a
compound containing three elements, all of which contribute to its overall
meaning. Consider the English phrase upper respiratory tract. There are
three elements encoding one concept which English expresses using three
words. Sanskrit would combine all three words into a compound. In English,
any morphology is applied to the final word in the sequence: several upper
respiratory tracts; the upper respiratory tract’s resistance to antibiotics.
This is also true of the Sanskrit compound. Of the three words comprising
the compound in this verse, only the final element contains any grammar:
lobhena (the instrumental singular ending for a short -­a stem). The other
two words appear in their citation form. Taken as a whole, the compound
yields the meaning by/through royalty pleasure greed. Sanskrit compounds
are notoriously difficult to translate and very often ambiguous, in that the
elements comprising a compound lack the grammatical endings that would
serve to inform the reader as to their relationship to each other. The longer
the compound, the greater the difficulty. As with all things related to the
learning of Sanskrit, however, the more one reads, the more familiar things
start to become.
10
TEXTS FOR THE STUDY
OF SANSKRIT

The student of Sanskrit is strongly advised to read widely. Despite its strong
and evident connections with other Indo-­European languages, not to men-
tion its pervasiveness in the languages and living cultures of India, San-
skrit is an ancient language and therefore separated from the modern era
by millennia. To read Sanskrit is to read a language that was kept fixed
by its custodians so as to reflect the unchangeable values of the tradition
which it represented. To fail to read around the language is to divorce it
from its historical and cultural base, with the result that it becomes little
other than a linguistic curio. There is an abundant amount of scholarship
on ancient India, on the linguistics of the Indo-­European language family
and on Sanskrit itself. There is also a lot of somewhat less creditable writ-
ing on such topics. Sanskrit is particularly prone to doubtful scholarship,
given the alacrity with which it has been embraced by those who seek to
elevate it to the status of the supernatural. Care must be taken to distinguish
between sources which are well researched, balanced and reliable, on the
one hand and, on the other, those which are impassioned but lacking critical
judgement.
For those whose interests incline towards archaeology but who are not
averse to dipping a toe into linguistics, both Mallory (1991) and Renfrew
(1990) are to be recommended. These are solid and accessible texts, not
aimed purely at archaeologists or linguists. They are balanced in their opin-
ions, presenting the data relating to the thorny issue of Indo-­European ori-
gins and allowing the data to speak for itself. For those interested in the
DOI: 10.4324/9780429325434-­11
 Texts for the study of Sanskrit 171

debate about the Aryans, Bryant and Patton (2005) is an excellent text, as
also Trautmann (2005). For students of ancient history and those who wish
to investigate the Indus Valley Civilization, a balanced and readable intro-
duction to the subject is provided by Robinson (2015). The scholar who
is the most authoritative, when it comes to the Indus Valley language and
writing system, is Parpola (1994), with whose work those proposing to try
their hand at deciphering the Indus Valley script should be familiar. Parpola
(2015) also investigates possible cultural connections between the Indus
Valley Civilization and early Hinduism. Works by authors claiming to have
deciphered the Indus Valley script should be read with caution and only
after Parpola has been explored. Such works are not mentioned amongst the
following texts, since they cannot be recommended.
There are a number of textbooks currently in print for the student keen
to learn Sanskrit. Some of these are accessible; others, less so. If the learner
is not studying a course in Sanskrit for which a textbook has already been
specified, the decision as which textbook to use is one that merits careful
consideration. The learner has first to decide whether he or she prefers a
solidly grammatical approach, where the language is presented purely as a
linguistic system, or whether a more culturally oriented approach is likely to
achieve the desired results. Sanskrit can be daunting for those not familiar
with grammatical terminology. For such intending learners, the principal
learning tool ought to be one in which cultural points are abundant and
where there are frequent comparisons made between what happens in San-
skrit and in English. In this respect, the pedagogical tools of yesteryear are
best kept as reference materials, with a more accessible textbook as the
primary source.
Macdonell (1927) is excellent for reference. The approach is ‘tradi-
tional’ but none the worse for it. One of the greatest drawbacks to Mac-
donell is that legible copies are hard to find. The print quality tends to be
poor, the font demanding on the eyesight. No such problems exist with
Bucknell (1994), recommended for use as a resource alongside a course-
book. All explanations are given in the IAST, and there is no use what-
soever of the Devanāgarī. Bucknell will appeal to those who wish to see
the grammatical bones of Sanskrit laid bare. For those who prefer more
flesh on the bones, Burrow (1955), despite its age, has not lost any of
its appeal. It is a beautifully written walk around the Sanskrit language.
For a more contemporary approach, and a solidly linguistic one, Cardona
and Jain (2003) is a work of great importance and distinctly ‘guru’. It is
knowledgeable and weighty, as one might expect for a text which exceeds
a thousand pages.
172 Texts for the study of Sanskrit

Slimmer than Cardona and Jain (albeit not by much) is Maurer (2009).
Not only is Maurer’s introduction worth reading but his appendices con-
tain much that will appeal to those whose primary interests in Sanskrit lie
in linguistics, grammar and literature. The reading passages are annotated,
allowing them to be accessible to the reader, and become increasingly more
complex as the reader’s knowledge of grammar, syntax and vocabulary
develops. As an all-­round coursebook, and despite being brisk in the ear-
lier chapters, Maurer represents a solid choice for the intending Sanskritist.
Maurer’s appendices are a mine of information on such matters as the Indo-­
European languages, as well as grammar.
For students who are interested in pursuing a literary exploration of San-
skrit, a number of texts can be recommended. Vedic literature is not acces-
sible to the beginner, since the language is quite far removed from classical
Sanskrit. A good starting point is the Bhagavadgītā, in that it is compact
and can be tackled in easy stages. For those who would like to approach
the Bhagavadgītā in English, to make themselves familiar with its contents
prior to attempting the text in Sanskrit, Mascaró (1962/2003) can profita-
bly be read. The 2003 edition, containing an excellent introduction by Dr.
Simon Brodbeck, Reader in Religious Studies at Cardiff University, is worth
obtaining. For a translation containing the original Devanāgarī text, Feuer-
stein (2011) is heartily recommended. Feuerstein has the additional benefit of
containing nine introductory essays, firmly establishing the Bhagavadgītā in
its literary and cultural contexts. Sanskrit students tend to gravitate towards
Sargeant (2009) which, in terms of permitting a close reading, is unmatched.
The Devanāgarī text of the Bhagavadgītā appears on the left-­hand page,
together with both a literal and more polished translation. The right-­hand
page contains an explanation of all the words which appear on the facing
page. For the reader who is also studying Sanskrit, this is ideal. An explora-
tion of the Bhagavadgītā in its wider literary context requires familiarity with
the Mahābhārata, for which the reader is directed to van Buitenan (1980).
As regards literature other than the Bhagavadgītā, Lanman (1884),
despite its age, continues to be useful. Most reading passages are taken from
wisdom literature, akin to Aesop’s fables in tone (the Hitopadeśa), and the
much early Ṛgveda, even though the latter represents Vedic rather than
classical Sanskrit. Copious notes accompany the reading passages, and Lan-
man contains a substantial Sanskrit to English vocabulary so that the reader
is not required to refer constantly to a dictionary. The Hitopadeśa represents
an accessible text, in that the stories are engaging and of manageable length.
They offer an insight into the folk wisdom and moral precepts of ancient
 Texts for the study of Sanskrit 173

India and have remained popular in contemporary India, where they are
often cited. An excellent introduction to the Hitopadeśa is available through
the Clay Sanskrit Library series (Törzsök, 2007), a series which, at the time
of writing, is out of print. Any number of copies of the various books in the
series continue to be available secondhand, and they are a useful resource
for the Sanskrit learner. Texts in the Clay Library Series are bilingual edi-
tions, with the Sanskrit (in the IAST) on the left-­hand page accompanied by
a close translation on the right-­hand page.
A more comprehensive collection of wisdom literature than the
Hitopadeśa is the Pañcatantra, older in terms of composition and widely
known in antiquity wherever Indic culture and a knowledge of Sanskrit
were spread. The reader interested in exploring in more depth this important
genre within Sanskrit literature is directed to the original text (again, in the
IAST) and scholarly analysis by Egerton (1924). This is not a translation of
the work but a detailed investigation of the text based on extent versions.
For a readable and accurate translation, Egerton (1965) is recommended.
Egerton (1924) is a work best approached with both a translation and a
dictionary to hand.
On the subject of dictionaries, there is only one that fits the bill, as far
as the student of Sanskrit is concerned. This is Monier-­Williams (1899), the
constant companion of every Sanskritist, teacher and student alike. Care
should be taken to inspect a copy of the dictionary prior to purchase, if pos-
sible, since the quality of the print can often be extremely poor in places. It
is a marvel of European scholarship in Sanskrit, and it is difficult to envis-
age a point where Monier-­Williams would become obsolete. The dictionary
is available online and the learner is strongly encouraged to take the time to
find out how its contents may be accessed. The Harvard-­Kyoto Convention,
as discussed in 2.3, is extremely useful in this respect.
A final word is needed regarding a work which is both immensely
readable and able to put the study of Sanskrit into a context where it
can fully be appreciated as one of the world’s great cultural languages.
The work in question is Ostler (2005). Ostler explores the spread of vari-
ous languages, such as Akkadian, Aramaic, Greek and Latin. A chapter
is devoted to the ‘cultured career of Sanskrit’ (Chapter 5), seeing how it
was taken to Southeast Asia and, with Buddhism, to central Asia and to
East Asia along the Silk Road. By its very nature, Sanskrit contained the
seeds of its eventual decline, despite its spread. It was the language of
the few; a language acquired through years of careful study; the language
of Hinduism and Buddhism and, ultimately, linked to the progress (or
174 Texts for the study of Sanskrit

decline) of Hindu and Buddhist thought. As other languages flourished,

either through trade, conquest or the spread of new ideas, Sanskrit fell
away. Ostler’s ‘charming creeper’ was cut back, retreating to South Asia,
where it continues to face the challenges of the modern age. The fate of
Sanskrit, as indeed of all languages, depends on the relevance it has for
learners and users. Despite the march of globalization, Sanskrit has not
yet reached the end of its life cycle.

Indo-­European studies
Mallory, J. P. (1991) In Search of the Indo-­Europeans: Language, Archaeology and
Myth. London: Thames and Hudson.
Renfrew, C. (1990) Archaeology and Language: The Puzzle of Indo-­European Ori-
gins (New Ed.). Cambridge: Cambridge University Press.

Indo-­Aryan studies
Bryant, E. F. and Patton, L. L. (Eds.) (2005) The Indo-­Aryan Controversy: Evidence
and Inference in Indian History. Abingdon, Oxon: Routledge.
Trautmann, T. R. (2005) The Aryan Debate (Debates in Indian History and Society).
New Delhi, India: Oxford University Press.

Indus Valley Civilization

Parpola, A. (1994) Deciphering the Indus Script. Cambridge: Cambridge University
Press.
Parpola, A. (2015) The Roots of Hinduism: The Early Aryans and the Indus Civiliza-
tion. New York: Oxford University Press.
Robinson, A. (2015) The Indus (Lost Civilizations). London: Reaktion Books.

Sanskrit language
Bucknell, R. S. (1994) Sanskrit Manual: A Quick-­reference Guide to the Phonology
and Grammar of Classical Sanskrit. New Delhi, India: Motilal Banarsidass.
Burrow, T. (1955) The Sanskrit Language. London: Faber and Faber.
Cardona, G. and Jain, D. (Eds.) (2003) The Indo-­Aryan Languages. Abingdon,
Oxon: Routledge.
Macdonell, A. A. (1927) A Sanskrit Grammar for Students (3rd Ed.). Oxford:
Oxford University Press.
Maurer, W. H. (2009) The Sanskrit Language: An Introductory Grammar and
Reader (Rev. Ed.). Abingdon, Oxon: Routledge.
Ostler, N. (2005) Empires of the Word: A Language History of the World. New York:
HarperCollins Publishers Inc.
 Texts for the study of Sanskrit 175

Sanskrit literature
Egerton, F. (1924) The Panchatantra Reconstructed (Volume 1: Text and Critical
Apparatus). New Haven, CT: American Oriental Society.
Egerton, F. (1965) The Panchatantra: Translated from the Sanskrit. London: George
Allen and Unwin Ltd.
Feuerstein, G. (2011) The Bhagavad-­Gītā: A New Translation. Boston, MA: Sham-
bala Publications, Inc.
Lanman, C. R. (1884) A Sanskrit Reader: Text, Vocabulary and Notes. Cambridge,
MA: Harvard University Press.
Mascaró, J. (1962/2003) The Bhagavad Gita. London: Penguin Books.
Sargeant, W. (2009) The Bhagavad Gītā (25th Anniversary Ed.). Albany, NY: Excel-
sior Editions, State University of New York Press.
Törzsök, J. (2007) Friendly Advice by Nārāyaṇa & King Vikrama’s Adventures
(Clay Sanskrit Library. General Eds. R. Gombrich and S. Pollock). New York:
New York University Press and the JJC Foundation.
van Buitenan, J. A. B. (1980) The Mahābhārata. 1: The Book of the Beginning (New
Ed.). Chicago: University of Chicago Press.

Sanskrit dictionary
Monier-­Williams, M. (1899) A Sanskrit-­English Dictionary. Oxford: Oxford Uni-
versity Press.

The University of Cologne (Universität zu Köln) has created a digital

archive (Cologne Digital Sanskrit Dictionaries) which contains all the
major dictionaries for Sanskrit as well as Monier-­Williams. The vocabu-
lary for Lanman’s Sanskrit Reader is included within this archive. It can be
accessed by typing ‘IITS Koeln’ into one’s preferred search engine.
GLOSSARY

Ablative The fifth of eight cases in Sanskrit, equating to the English prep-
osition from (e.g. The mouse came out from its lair).
Accusative The second of eight cases in Sanskrit, representing the object
of the sentence. The object does not control the verb but is the recipient
of it (Sītā addresses the guru). The accusative also indicates movement
towards a place (You go to the village).
Adjective A word which describes a noun (e.g. dark horse; small house).
In Sanskrit, an adjective agrees with the noun, so that it is case-­marked
in accordance with the case-­marking on the noun. If a noun is masculine
nominative and plural, the adjective has the same ending. Adjectives
are often expressed using compounds or the past passive participle.
Adverb An adverb describes the way a verb is taking place (e.g. to go
swiftly; to speak angrily). Most adverbs in Sanskrit are not declined,
meaning that they have a fixed form. They do not have the same ending
as the finite verb form, since finite verb forms have a personal ending.
Affricate The description of a phoneme in terms of its manner of articula-
tion. An affricate contains features of both a stop, where the air builds
up in vocal tract but is allowed to hiss out. In Sanskrit, there are four
affricates: c, ch, j and jh (These are palatal consonants). In all cases,
the tongue touches the top of the mouth and moves to allow a sh-­sound
to follow.
AMT (See Aryan Migration Theory.)
 Glossary 177

Anusvāra This is the dot, in the Devanāgarī, placed above a consonant,

semivowel or sibilant to indicate that the immediately preceding vowel
is nasalized. In the IAST, it is indicated by a dotted m (e.g. kiṃ).
Anusvāra cannot occur before a vowel or right at the end of a sentence.
It is common practice to pronounce anusvāra as the nasal belonging to
the same place of articulation as the following consonant, semivowel
or sibilant.
Apabhraṃśa The phase of an Indic language, transitional between
the Old and Middle Indic stages, as represented by Sanskrit and the
Prakrits, and New Indic, which is the term applied to the modern lan-
guages of India. In Sanskrit, the word apabhraṃśa means falling
away – a name which indicates a language variety in the process of
losing endings associated with declension and conjugation in the Old
and Middle Indic languages.
Articulation The production of a phoneme. This is often described as
pronunciation, although articulation refers to the actual creation of a
sound rather than to anything relating to its quality.
Articulators The parts of the oral tract (i.e. the soft palate, the roof of the
mouth, the hard palate, the teeth, the lips, the tongue).
Aryan A controverted term, given its political misuse, but prevalent in
writing concerned with the language and settlement of speakers of the
early Indic languages. In Sanskrit, the term ārya means honourable
or esteemed person. To what extent the term was used in antiquity to
distinguish between ethnic groups is debatable. It is more likely that
it indicated adherence to a particular cultural tradition associated with
the Vedas. The terms ‘Aryan’, ‘Indo-­Aryan’ and ‘Indic’ all relate to the
non-­Iranian languages of the Indo-­Iranian branch of the Indo-­European
family.
Aryan Migration Theory (AMT) Often conflated with the earlier Aryan
Invasion Theory, this is the term used to refer to a hypothesis which
posits that the original speakers of Vedic Sanskrit entered South Asia
bringing the language with them. The greatest objection to the AMT is
that it is claimed to represent an extension of outmoded and colonial
thinking. The opponents of the AMT promote the Out of India Theory
(OIT).
Aspirate A sound in Sanskrit which is represented by an individual pho-
neme (h) and is present in ten consonantal phonemes: kh, gh, ch, jh,
ṭh, ḍh, th, dh, ph and bh. It is a forceful exhalation which needs to
be audible in Sanskrit, since it distinguishes aspirated phonemes from
178 Glossary

unaspirated ones. The IAST represents the aspirate with the letter h in
all cases.
Avagraha The sign which resembles the handwritten, uppercase English
letter S, restricted to indicating the absence of the initial short a-­when
preceded by a word that ends with either -­e or -­o.
Avestan The oldest known Indo-­Iranian language on the Iranian side. It
is the language in which the earliest Zoroastrian scriptures were com-
posed and is the closest in structure to Sanskrit, outside of South Asia.
BJP The Bharatiya Janata Party, currently in power in India. It supports
Hindu nationalist views and holds Sanskrit in great esteem due to its
central role in the preservation of Hindu culture.
Borrowing Linguistic borrowing refers to the adoption of a word from
another language, which then undergoes a process of ‘nativization’ so
that the loan word is consistent with the sound system of the borrow-
ing language (Chinese loan words are borrowed into English without
their tone, for example). A borrowed word may potentially come from
any language. The relatedness of languages is not a key factor in this
respect.
Brāhmī The parent writing system of all Indian languages (both Indic and
Dravidian) except for those employing the Perso-­Arabic script, such
as Urdu. The Brāhmī is the first attested writing system of India, with
the exception of the Indus Valley script, and is the script used in the
pillar inscriptions of Aśoka (c. 250 bce). A direct connection with the
Indus Valley script has not been established. Although the Brāhmī is
the oldest-­known writing system known for the Indic languages, there
is currently no evidence to suggest that it was used for Sanskrit. The
origin of the Brāhmī remains contested.
Case In Sanskrit, case is marked as an ending to a noun, indicating the
relationship of the noun in question to the rest of the sentence. There
are eight cases, traditionally numbered in Sanskrit (i.e. first, second,
etc.) but for which a Latin-­based terminology is usually employed in
textbooks. The eight cases, in order, are as follows: nominative, accu-
sative, instrumental, dative, ablative, genitive, locative and vocative. It
should be noted that the Indic grammatical system did not acknowledge
the vocative case.
Cerebral A place of articulation represented by a series of consonant
phonemes (ṭ, ṭh, ḍ, ḍh, ṇ) and allocated to the semivowel r and the
sibilant ṣ. Cerebrals are often referred to as ‘retroflex’, although that is
not so much a place of articulation as a description of what happens to
the tongue tip in articulating the phoneme in question.
 Glossary 179

Citation form The form in which a word appears before any declension
or conjugation has been applied. With nouns, it is the stem form; with
verbs, it is the root. The term is synonymous with ‘dictionary form’.
Cognates (See also Borrowing.) Cognates are words which show evident
similarities in form across languages, leading to a supposition that the
similarity is due to the fact that the languages in question are related.
This is distinct from borrowing, which does not indicate any necessary
genetic relatedness between languages. Distinguishing between cog-
nates and loan words is not always easy. Borrowing can (and does) take
place between languages that are related, as evidenced by the massive
influx into English of words from Latin and French – the latter being a
descendant of Latin.
Compound Not to be confused with conjunct, the term relates to a fusion
of words to produce a composite word, often with a sense which is not
the sum of its parts. In English, for example, a bigwig is not a person
who necessarily has to wear a false hairpiece. Compounds occur fre-
quently in Sanskrit.
Conjugation This applies to verbs, not to consonants, and refers to the
process of adding endings to a verb stem to indicate who is performing
the verb.
Conjunct Conjuncts represent the coming together of any combination of
consonants without an intervening vowel. There are various strategies
used in the formation of conjuncts and they represent the greatest chal-
lenge in learning to read the Devanāgarī.
Consonant In Sanskrit, the consonants are represented by a group of
thirty-­three phonemes (thirty-­five, if anusvāra and visarga were to be
included), arranged according to five places of articulation. Consonants
are sounds which require articulators to be in contact.
Daṇḍa The only punctuation mark used in Sanskrit. It has the form of
a single vertical stroke, not connected to any letter and not bearing a
horizontal stroke at the top. A single daṇḍa indicates a full stop in Eng-
lish. A double daṇḍa (dvadaṇḍa) indicates either the end of a text or
the end of an individual verse of poetry. The word means stick or staff.
Dative The fourth of eight cases in Sanskrit, indicating the recipient or
beneficiary of an action. The English prepositions that correspond to
the dative are to and for (e.g. The teacher reads the book to his students;
the army fights for the king).
Declension This relates to the process of adding endings to nouns and to
adjectives to show their function within the sentence. Declension indi-
cates gender, case and number.
180 Glossary

Definite article In English, the definite article is the word the. Sanskrit
does not possess a definite article but may indicate that something is
being referred to by using the relevant declension of the word tat (a
demonstrative pronoun).
Demonstrative pronoun Words such as tat, etat and idam are used
in Sanskrit to indicate that one; this one; those; these, etc. They are
demonstrative pronouns and refer to something or someone mentioned
elsewhere in the text.
Dental A place of articulation represented by a series of consonant pho-
nemes (t, th, d, dh, n) and allocated to the semivowel l and the sibilant s.
Devanāgarī The writing system associated with Sanskrit. The Devanāgarī
is also used to write Hindi and Marathi. It is the most widespread of
all Indian writing systems within India and Nepal. The meaning of the
word Devanāgarī is (of/from) the city of the god(s). The Devanāgarī is
descended from the Brāhmī script.
Diacritic/diacritical mark Any sign added to a letter in the Roman
alphabet to indicate its pronunciation. Typically, this is a dot or a slash
but there are several diacritical marks, and these vary from one lan-
guage to another.
Diphthong A fusion of two audibly distinct vowel sounds. A diphthong
may not necessarily be represented by two letters in all languages.
In Sanskrit, there are two diphthongs and these have their own letter
within the Devanāgarī. Importantly, diphthongs are deemed to be sin-
gle phonemes.
Dravidian The name of a language family of southern India which
includes four languages with long literary traditions and which are offi-
cial state languages of India: Tamil, Telugu, Kannada and Malayalam.
Although heavily influenced by Sanskrit, especially with respect to
borrowing, these languages are not Indo-­European or Indic.
Dual The term which relates to grammatical number and refers to
two – and only two – participants. Dual forms exist in the noun
(e.g. two hands; two horses) and in the verb (e.g. They [the two of
them] go).
Enclitic A word which follows another and indicates something about
the preceding word. In Sanskrit, an enclitic can indicate a connection
between two words (ca), emphasize a word (eva; iva) or represent a
shortened form of a pronoun (me; te, etc.). A common enclitic is tu
(but; however).
Finite verb This is a verb which appears in a conjugated form, which
means that someone is identified as performing it (e.g. I speak, you
 Glossary 181

speak, he/she/it speaks). When no person is identified in the verb end-

ing, it is in the infinitive (to speak).
Fricative This describes the manner of articulation of a phoneme, where
air is hissed out rather than being held. The tongue is positioned in a
way in which the airflow can be sustained over a number of seconds;
the sound does not need to be released immediately. The fricatives in
Sanskrit are ś, ṣ, s and h. These are the three sibilants and the aspirate.
Gender In Sanskrit, all nouns are either masculine, neuter or feminine.
Often, the gender of a noun is determined by its biological character-
istics. Horse is masculine, mare is feminine; cow is feminine, bull is
masculine, etc. With things that do not possess obvious gender or are
abstract (knowledge; fear; thought), the gender must be learnt as it can-
not be assumed.
Genitive The sixth of eight cases in Sanskrit, indicating possession (e.g.
The horse’s mane). Although English possesses a genitive apostrophe
(Mark’s book), it does not always need to use it. In the sentence: this is
the house of your sister, the preposition of indicates the genitive func-
tion. Often the genitive function is indicated adjectivally in English
(e.g. The town road – meaning the road of the town).
Germanic A branch within the Indo-­European language family repre-
senting languages such as English, Dutch, German and the Nordic lan-
guages (Icelandic, Norwegian, Swedish, Danish).
Gerund In Sanskrit, a gerund is a word which gives verbal information
but does not involve conjugation. It is composed of a verbal root, which
often undergoes some form of modification, to which -­(i)tvā is added,
or -­ya, if the verb is prefixed (In a handful of cases, -­na is added rather
than the predicted -­(i)tvā). A possible way of translating a gerund is
to use the English formula: having + past participle, such as tyaktvā
(having forsaken; renounced). The gerund tyaktvā contains the verbal
root tyaj, with a mutation affecting the final consonant (-­j to -­k), plus
-­tvā.
Guṇa A form of vowel mutation in which a primary vowel other than a
undergoes a change in quality (i/ī → e; u/ū → o; r̥ /r̥ ̄ → ar; l̥ → al).
Guttural (See Velar.)
Harvard-­Kyoto Convention A transliteration system for the Devanāgarī
which is keyboard-­friendly and widely used with computer applications.
Indefinite article In English, the indefinite article is the word a (or an,
before a vowel). Sanskrit does not possess an indefinite article but may
indicate that a singular something is being referred to by using the word
eka (one).
182 Glossary

Indic This word is not to be conflated with ‘Indian’. ‘Indian’ indicates

something as originating from India, whereas ‘Indic’ specifically draws
attention to a group of related languages. The Indic languages are those
belonging to the Indo-­Iranian branch of the Indo-­European family. The
term is often used in preference to ‘Aryan’, to avoid unwanted histori-
cal overtones.
Indo-­European The term for the language family to which both English
and Sanskrit belong. It is the most geographically widespread of all the
language families of the world.
Indo-­Iranian A branch within the Indo-­European language family which
some linguists consider to be too broad, preferring to split the branch
into two component parts: Iranian in the west and Indic in the east.
Other branches of Indo-­European include Romance (or Italic) and Sla-
vonic (or Slavic). English is from the Germanic branch.
Infinitive A verb form which does not have a personal ending (e.g. to
speak; to go; to live). In Sanskrit, infinitives end in -­(i)tum. In the sen-
tence: he wants to speak to her, the verb form wants is conjugated, and
therefore finite, and the verb form to speak is in the infinitive. Note
that infinitives in English are preceded by to. Dictionaries of English
list verbs in their infinitive forms, whereas dictionaries of Sanskrit list
verbs according to their root.
Instrumental The third of eight cases in Sanskrit, representing the means
by which something is taking place. In English, the prepositions by
and with reflect the Sanskrit instrumental (e.g. she travels by train; she
always visits with her brother).
Interjection A word which represents an exclamation, such as oh! or ah!
Labial A place of articulation represented by a series of consonant pho-
nemes (p, ph, b, bh, m) and allocated to the semivowel v. Labials are
strictly speaking ‘bilabials’, since the term ‘labial’ can also indicate
phonemes that are not attested in Sanskrit (e.g. *f).
Language branch A division within a language family which identifies
a group of language more closely related to each other than to other
members within the family in question. For example, Italian and Span-
ish are closer to each other than either of them are to Polish or Russian.
Language family A concept which groups languages on the basis that are
deemed to have descended from the same ancestral form. With Indo-­
European, the ancestral language form is the hypothesized (and recon-
structed) Proto-­Indo-­European. Other language families of Europe and
Asia include Finno-­Ugric, Altaic, Semitic, Dravidian, Sino-­Tibetan and
Austroasiatic. Languages belonging to these families may have been
 Glossary 183

profoundly influenced by languages from other families, making the

process of determining family affiliation a problematic one.
Locative The seventh of eight cases in Sanskrit, giving information on the
location of something within the sentence (e.g. on the ground; through
the street; in the shade).
Macron This is a diacritical mark which takes the form of a horizontal
line written over a vowel to indicate that the vowel is long (e.g. ā; ī).
The diacritical mark for indicating a short vowel is known as a ‘breve’
(e.g. ĕ; ŏ). The breve is not used in Sanskrit to indicate a short vowel.
Vowels are short unless indicated to the contrary. There are two excep-
tions to this: e and o. Both are long in Sanskrit and do not have short
equivalents.
Manner of articulation Distinct from place of articulation in that it
describes how a phoneme is pronounced rather than where. Voice and
aspiration are manners of articulation. The term ‘sibilant’ is another
example of manner of articulation, since it indicates airflow. In San-
skrit, a sibilant is palatal (ś), cerebral (ṣ) or dental (s), according to its
place of articulation.
Mantra A short composition aimed at concentrating the mind by placing
the focus on sound rather than meaning.
Mātrā The horizontal topline which is found over most letters in the
Devanāgarī. This is the last stroke to be written in the formation of a
letter. The word mātrā means measure.
Mood Often conflated with tense, mood indicates the way a verb is being
undertaken rather than if the action is taking place, has taken place or
will take place.
Morpheme A unit of meaning which can be a word itself or a small part
of it. The word cat is a morpheme, since it cannot be split into anything
smaller and still retain its meaning. The words cats, cat’s and cattish,
however, contain additional morphemes which add to the meaning. In
the first of these, the noun is put into the plural with the addition of the
plural morpheme -­s; in the second, the genitive apostrophe is used (’s),
indicating that something belongs to the cat; in the third, the morpheme
-­ish is added, indicating similarity or equivalence.
Morphology The identification of a word according to the units of mean-
ing contained within the word in question. Morphemes can be indi-
vidual words (e.g. cat) or things added to a word to add to its overall
meaning (uncatlike).
Nasal In Sanskrit, five consonantal phonemes are identified as nasal: ṅ,
ñ, ṇ, n, m. With these consonants, the air must be allowed to travel
184 Glossary

through the nose as well as the mouth when they are pronounced. This
passage of air through the nose is what makes them nasals (see also
Anusvāra and Manner of articulation).
Nominative The first of eight cases in Sanskrit, representing the subject
of the sentence. The subject controls the verb (Sītā speaks to the guru).
Noun A thing as opposed to an action. Nouns may be proper nouns (as
with the names of people, or places), concrete nouns, representing
things which may be touched or detected by the senses (e.g. soil; a
garden; scent) or abstract nouns (wisdom; time, etc.). In Sanskrit, a
noun can function as an adjective. It is not so much the form of a noun
but its position within a phrase or sentence which indicates whether it
is functioning as a noun or as an adjective.
Number With respect to grammar, number indicates whether a noun or a
verb involves one person or more than one. English has two numbers:
singular and plural. Sanskrit has three: singular, dual and plural (see
also Dual). Number needs to be considered alongside person, where
verbs are concerned. In the phrase third person singular, person and
number are indicated and identify a specific conjugational form.
Object The person or thing who/which is the recipient of the action of a
verb, assuming that the sentence is an active construction (e.g. The dog
chases the cat). In a passive construction, the situation in English is
more complicated, in that the recipient of the verb is placed in subject
position, before the verb, but is still the recipient of the action: The
cat is chased by the dog. Since word order is more flexible in Sanskrit
than in English, Sanskrit uses case endings to distinguish the object
from the subject. The subject must be indicated by the nominative
case, whereas the object may appear in any of the other cases (except
the vocative).
OIT (See Out of India Theory.)
Old Indic A term used to designate the earliest attested Indic languages:
Vedic, Sanskrit and classical Sanskrit. Prakrits tend to be classified as
Middle Indic languages but have a long history and contain features
which connect them closely to Vedic in terms of antiquity.
Out of India Theory (OIT) A hypothesis suggesting that Sanskrit origi-
nated in South Asia, along with its ancestral form. This posits the
Indo-­European homeland as South Asia, given that Sanskrit is an Indo-­
European language. Opponents of this hypothesis point to the fact that
is it is not supported by linguistic and archaeological data.
Palatal A place of articulation represented by a series of consonant pho-
nemes (c, ch, j, jh, ñ) and allocated to the semivowel y and the sibilant ś.
 Glossary 185

Paradigm A pattern or model exemplifying a particular phenomenon.

Typically, it is used in language learning to indicate a table of declen-
sion or conjugation.
Participle In Sanskrit, a participle is a word formed from a verbal root
which serves a number of functions. The past passive participle is an
example of how broad such uses can be: it can substitute for a finite
verb or can be purely adjectival in nature. It is better for the learner
to put participles to one side until a firm grasp has been achieved on
declension and conjugation.
Person As with English, Sanskrit grammar acknowledges three persons:
the first (I/we); the second (you [singular]/you [plural]); the third (he,
she, it/they). It is important to remember that, whilst English and San-
skrit agree as to this grammatical concept, person is to be seen within
the context of number. With respect to number, English and Sanskrit
differ.
Personal ending The addition to a verbal stem which creates a finite verb.
For example, the verbal root bhū (to be) generates the stem bhava-­, to
which a personal ending is added to create a finite verb form (e.g. bha-
vati, he/she/it is). The knowledge of personal endings is what allows
the student of Sanskrit to make sense of conjugation.
Phoneme A sound which is recognized by the sound system of a language.
The word ‘sound’ has its uses when attempting to capture similarities
between languages but, within the context of describing Sanskrit, is too
vague. The classification of phonemes in the Indic tradition is precise
and comprehensive. A phoneme has only one outcome, unlike the let-
ters of the Roman alphabet as it is used for English. For example, the
plural morpheme -­s in the word cats is the same as the initial phoneme
in the word snake. It is an unvoiced sibilant. In the word dogs, however,
the sibilant is voiced and sounds like the English z in zebra.
Phonology The description of the sound system of a language from the
point of view of how the sounds in question interact with each other.
Investigating sounds is, technically speaking, the science of phonetics;
but sounds can undergo change due to a number of factors. Note the
difference in sound in the plural morpheme -­s in the words cats and
dogs. The difference is due to a phonological process.
Phrase A phrase is a component within a sentence, not a grammatically
complete sentence. My hamster is a noun phrase whereas my hamster is
cute is a complete sentence (It contains both a subject and a finite verb).
Phrases account for a large number of sayings and expressions which do
not qualify as complete sentences, such as my country, right or wrong.
186 Glossary

PIE (See Proto-­Indo-­European.)

Place of articulation Sanskrit identifies five positions within the oral (or
vocal) tract that account for the production of its consonants, semivow-
els and sibilants: the velum (or soft palate), producing velar sounds; the
hard palate, which accounts for the palatals; the cerebrum (the highest
part of the hard palate, presumed to be closest to the brain – hence its
name), producing the cerebral sounds; the top teeth, with which the
tongue is in contact to produce the dentals; the lips, which must be in
contact with each other to create the labials. These five positions are the
places of articulation.
Plural The term which relates to grammatical number and, in Sanskrit,
refers to three or more participants. Plural forms exist in the noun (e.g.
several hands; several horses) and the verb (e.g. They [three or more
of them] go).
Prakrit(s) There is ample linguistic evidence to show that the Prakrits
are not the descendants of Sanskrit but, rather, ancient vernacular
languages of India, used for everyday purposes whilst Sanskrit was
reserved for religious and scholarly purposes and literary composi-
tion. In due course, some Prakrits became literary languages, such as
Pāḷi. The Prakrits precede the Apabhraṃśa phase, which represents
the transition between Middle Indic and New Indic languages. There
is no distinct cut-­off point between phases; these represent a conveni-
ent linguistic fiction, albeit justified in terms of representing language
change.
Prefix A morpheme added at the start of a word to modify its meaning.
Nouns may be prefixed in Sanskrit (e.g. vidyā, knowledge; avidyā,
ignorance), but it is with verbs that prefixes are at their most complex.
Before verbs, they may drastically modify the root meaning of the verb
(e.g. √ gam, to go; with prefix ava-­[ava-­+ √ gam], to understand).
Pronoun A word that relates to something or something which occurs
in the same sentence or wider text (either spoken or written). One
may refer to him, her or it; but without a clearer indication as to who
or what is being discussed, the pronoun does not shed any light. It is a
place-­holder for a noun and relies on the speaker or reader to deduce
the noun in question. Consider the sentence: That dog is overweight,
so it’s slow. The pronoun it refers to that dog – not to another dog or
to its owner.
Proto-­Indo-­European (PIE) The hypothesized ancestor of the Indo-­
European languages. The term ‘proto’ is used to refer to any language
variety for which there are no literary records, as such Proto-­Germanic,
 Glossary 187

the putative ancestral form of the Germanic languages. Proto forms

must be reconstructed.
Punctuation (See Daṇḍa.)
Retroflex (See Cerebral.)
Romance A branch within the Indo-­European language family repre-
senting languages such as Italian, Spanish, French and Romanian. All
members of the Romance branch are descendants of Latin.
Root The core form of a verb which carries the meaning. A root, in
Sanskrit, is not a complete word but requires an ending. Roots usu-
ally undergo modification to generate nouns or verbal stems (see also
Stem).
Sandhi A Sanskrit word adopted by linguistics to indicate a sound change
which takes place in the process of articulation (It is triggered by pho-
nological phenomena). In Sanskrit, the rationale behind sandhi is that
it is deemed to reflect what happens in natural pronunciation and its
effects have an impact in spelling. This tends not to be the case in Eng-
lish, where spelling is preserved irrespective of actual pronunciation.
For example, in rapid speech, one might hear the phoneme r in the
phrase India (r)and its neighbours, but it would not be written.
Schwa This is another term for the indeterminate vowel, frequently heard
in English, which occurs in an unstressed syllable (e.g. insider). San-
skrit does not possess an indeterminate vowel. The inherent vowel in
Sanskrit (a) is not to be confused with the indeterminate vowel.
Semivowel In Sanskrit, these are the phonemes y, r, l and v. They are a
group of sounds contained within the larger category of consonants but
do not have the feature of aspiration. All semivowels are voiced.
Sibilant Literally, a whistling sound. Sanskrit possesses three sibilants: ś,
ṣ and s. The aspirate h can be convenient placed alongside the sibilants,
for the purposes of analysis, since it is technically a fricative – a cat-
egory which includes all the sibilants (see also Fricative).
Singular The term which relates to grammatical number and refers to
one – and only one – participant. Singular forms exist in the noun (e.g.
a hand; a horse) and in the verb (e.g. She goes; I go).
Stem A form which is derived from a verbal root. As with the root, the
stem is not a word in itself but requires an ending to be complete. With
nouns, the ending is the relevant case ending; with verbs, it is a per-
sonal ending, resulting in a finite verb form.
Subject The person or thing who/which controls a verb, assuming that
the sentence is an active construction (e.g. The dog chases the cat). The
subject of an active construction is indicated by the nominative case.
188 Glossary

Suffix A morpheme added to the end of a word to modify its meaning.

Participles are examples of suffixes. Infinitives and gerunds are also
created by the addition of a suffix.
Syllable A unit of sound which may contain a vowel by itself or in com-
bination with a consonant, semivowel or sibilant (or more than one).
In all cases, a syllable requires a vowel. Syllables are best thought of
as beats. Consider the sentence: Cry ‘Havoc!’ and let slip the dogs of
war. There are ten beats to that sentence. Otherwise said, the sentence
contains ten syllables.
Tense Often conflated with mood, tense indicates the time in which a verb
is taking place, has taken place or will take place.
Transliteration A transliteration system aims at a one-­to-­one correspond-
ence from one writing system to another, in an attempt to capture the
pronunciation of an unfamiliar script. Transliteration systems for San-
skrit are many but the IAST is the system of choice for Sanskritists.
Velar A place of articulation represented by a series of consonant pho-
nemes (k, kh, g, gh, ṅ) and allocated to the aspirate h. Velars are often
referred to as ‘gutturals’, a term which is wider than it needs to be,
given that it includes sounds which are not attested in Sanskrit.
Verb Often defined as a ‘doing word’, which is a pithy and accurate
description. In Sanskrit, one needs to distinguish between roots, stems
and finite verbs. When giving a full description of a finite verb, one cites
the following: root, stem, verb class, person, number, tense/mood, voice.
Vocative The eighth of eight cases in Sanskrit, used to indicate a calling
out to someone or something (e.g. Hey, Sid!; oh, God!; o cruel fate!).
In Sanskrit, the vocative case ending is applied to the person or thing
being addressed. Often, the word marked in the vocative is preceded by
aho, bhoḥ or he.
Voice Voice has a twofold meaning in Sanskrit. When investigating a pho-
neme or a sound change, voice relates to when the vocal cords vibrate
during the articulation of the phoneme in question. As regards conjuga-
tion, voice indicates if a finite verb form is active, middle or passive.
In English, you eat the cookie is active, whereas the cookie is eaten
by you is passive. The consensus is that there is no real difference in
meaning between the active and middle voice in Sanskrit and that it is
a purely grammatical phenomenon: some verbs specify for the active
voice; others, the middle.
Vowel In Sanskrit, vowels may be short or long. Long vowels are indi-
cated with a macron, except for e and o. Unlike consonants, semivowels
and sibilants, vowels do not involve any touching between articulators.
 Glossary 189

Vowel mutation A process by which the vowel of a verbal root undergoes

a change prior to forming a stem. This process is also known as ‘vowel
gradation’, ‘vowel strengthening’ and ‘ablaut’. English employs vowel
mutation in man → men and mouse → mice in the creation of irregular
plural forms.
Vr̥ddhi A form of vowel mutation in which a vowel undergoes a change
with respect to length or diphthongization (a → ā; e → ai; o → au).
See guṇa.
INDEX AND LIST OF VERB ROOTS

adjectives 176; dharmya 136, 138; diphthongs 26 – 30, 50 – 51, 54 – 59, 78,
mahat 167 – 168; yuyutsu 75, 129 – 130, 180
159 – 161 dual 101, 115, 119, 131, 138, 164, 180
anusvāra 31, 44 – 46, 57, 78, 103,
140, 177 enclitics 180; ca 78, 106, 119,
Apabhraṃśa 8 – 9, 18, 186 122 – 123, 159 – 160, 162 – 163,
Aryan Migration Theory (AMT) 165 – 166; eva 153, 160, 162,
15, 177 165 – 166; iva 165; mā 118, 121; me
Aśoka 52 118 – 119, 121, 163 – 166; naḥ/vaḥ
aspirate 33 – 40, 43, 46 – 47, 142, 118, 121; te 118, 121; tvā 118, 121
177 – 178
avagraha 139, 142 – 143, 148, 178 gerunds 134 – 135, 181; abhilikya 135;
āgamya 135; gatvā 134, 136, 138;
Brāhmī (script) 52 – 53, 64, 80, 178 likhitvā 134; parityajya 135, 137,
139; tyaktvā 134, 137, 139; uktvā
cerebralization 148 – 150 136, 138, 153
cognates 11 – 12, 88, 179 Gurmukhī script 65
compounds 169, 179
consonants: cerebrals 24, 31, 37 – 38, Harvard-Kyoto Convention 23 – 25, 51,
40 – 41, 43 – 44, 178; dentals 31, 173, 181
38 – 40, 44, 180; labials 31, 40, 44,
182; palatals 31, 36 – 37, 40 – 41, Indus Valley 15 – 17, 52
43 – 44, 184; velars 31, 33 – 35, 40, inherent vowel 44, 48, 55, 57, 59, 62,
188; see also semivowels; 64 – 65
sibilants interjection 167 – 168, 182

daṇḍa/dvadaṇḍa 24, 51, 160, 179 Jones, Sir William 9 – 10, 13 – 14, 88, 98
Dhātupāṭha 92
diacritic/diacritical mark 79, 180 Kālidāsa 10
 Index and list of verbal roots 191

language/language family 182 – 183; (n) 159 – 160; kuru (m) 159 – 160;
Avestan 8, 17 – 18, 178; Bengali 7, lobha (m) 167, 169; māmaka (m)
9, 18, 80; Dravidian 18, 37, 52, 65, 159 – 160, 162; manas (n) 165 – 166;
180; Greek 7, 10 – 14, 18, 53, 79, mānuṣa (m) 85 – 86; marda (m)
89, 173; Gujarati 9, 15, 18, 80 – 81; 68; mārga (m) 68, 85 – 86, 150;
Hindi 7 – 9, 15, 18, 76; Indic 8, 15, mokṣa (m) 64, 71, 85 – 86, 94 – 95,
18, 52, 182; Indo-European 6 – 11, 98; mr̥ tyu (m) 154 – 155; mukha
17 – 19, 52, 170, 182; Indo-Iranian (n) 163 – 164; nagara (n) 105 – 108,
7 – 8, 18, 182; Latin 6 – 7, 10 – 14, 122 – 123, 131, 136 – 138, 151 – 152;
88 – 93, 105, 173; Lithuanian 7, nirvāṇa (n) 95 – 97; niśā (f) 91, 102,
11 – 12, 14; Oriya 9, 80; Panjabi 7, 111, 115 – 116, 120 – 121, 149; nr̥ pa
9, 15, 18, 52, 65, 80; Semitic 19, 55; (m) 85 – 86; pāṇḍava (m) 77 – 78,
Tamil 18, 65, 79; Urdu 52, 55, 80 136, 138, 159 – 160, 162; pāpa
(m) 167 – 168; pattra (n) 136 – 137;
Mahābhārata 1, 77, 94, 172 prabhāva (m) 150; putra (m) 119,
Monier-Williams 27, 29, 89, 91 – 92, 122, 136, 138; rājya (n) 167, 169;
173, 175 rāma (m) 23, 98, 103 – 108, 117,
122 – 123, 136, 138, 141 – 144, 148;
namaste/namaskāra 119 rāmāyaṇa (m) 1, 85 – 86; rātrī (f) 91,
negation 165 – 166 95, 102, 111 – 112, 116, 120 – 122,
nouns: agni (m) 91 – 92, 144; amr̥ ta (m) 149; roman (n) 163 – 164; r̥ ṣi (m)
154 – 155; anna (n) 105 – 106, 122; 85 – 86; sadana (m) 150; saṃjaya (m)
arjuna (m) 123, 151 – 152, 163; ārya 77, 159 – 160, 162 – 163; saṃskr̥ta
(m) 92; asat (m) 154 – 155; aśva (m) (m) 3, 18, 85 – 86, 122 – 123; śānti (f)
47 – 48, 103 – 108, 131 – 132, 46, 154 – 155; śarīra (n) 163 – 164; sat
141 – 142, 150; ātman (m) 94; (m) 154 – 155; śiva (m) 25, 29; sukha
bhagavadgītā (f) 77, 115, 145, (m) 47, 167, 169; sūta (m) 107 – 108,
156 – 169, 172, 175; bhakti (f) 71; 136 – 137; svajana (m) 167, 169;
brāhmaṇa (m) 67; buddha (m) 67, tamas (n) 154 – 155; tvak (f)
95 – 97, 100, 122, 152 – 153, 161; 165 – 166; udyāna (n) 109, 136 – 137;
deva (m) 85 – 86, 88 – 89, 91, 101, vana (n) 153; vedānta (m) 65,
109 – 111, 114, 120 – 121; dharma (m) 85 – 86, 93; vepathu (m) 163 – 164;
94 – 95, 100, 159 – 160; dharmya (m) vidyā (f) 71, 90 – 92, 95, 100;
136 – 138; dhr̥ tarāṣṭra (m) 77, 94, yuddha (n) 122 – 123
159, 162; gāṇḍīva (m) 165 – 166;
gaṅgā (f) 67; gātra (n) 163 – 164; OM sign 57
gaya (m) 30; grāma (m) 85 – 86, 122, Out of India Theory (OIT) 15 – 17, 184
137, 139, 152 – 153; guru (m) 92 – 93,
95, 100, 152; harṣa (m) 163 – 164; Pāṇini 92
hasta (m) 165 – 166; jana (m) past passive participles 139, 160 – 161,
122 – 123, 137, 139, 151; jñāna (n) 168; buddha 67, 95 – 97, 100, 122,
71; jyotis (n) 154 – 155; kamala (n) 141, 152 – 153, 161; citta 71; gata
89 – 91, 95 – 96, 100, 102, 109 – 111, 161; khyāta 65; samaveta 159 – 161;
114 – 115, 120 – 122, 136 – 137; saṃtuṣṭa 137, 139; udyata 167, 169;
kanaka (n) 49, 85 – 86; kanyā (f) vyavasita 167 – 168
136 – 137, 144, 147; karma (m) PIE see Proto-Indo-European
93 – 95, 100; kārtsnya (n) 68, 73; place of articulation 31, 33, 39 – 40
kaurava (m) 159, 163, 165; Prakrit(s) 1, 8 – 9, 18, 52, 186
khara (m) 142; kr̥ṣṇa (m) 22 – 23, 25, prefixes 187; ā- 106, 135; abhi- 135;
28, 122 – 123, 136, 138, 163; kṣetra ava- 160, 166, 168; ni- 97; nir- 96;
192 Index and list of verbal roots

pari- 135; sam- 46, 140, 160; ud- icchanti, √ iṣ (Class IV) 136, 138;
169; upa- 97; vi- 168 jāyate, √ jan (Class IV) 163 – 164;
pronouns 187; etat (demonstrative) 115, jīvanti, √ jīv (Class I) 122; jīvati,
118, 158; kim (interrogative) 78, √ jīv (Class I) 119; kartum, √ kr̥
157, 159 – 162; personal pronouns (Class VIII) 167 – 168; likhāmi,
116 – 119; tat (demonstrative) √ likh (Class VI) 130, 136 – 137;
112 – 116, 118 – 119, 158, 162, 168; likhitum, √ likh (Class VI) 133;
yat (relative) 115, 157, 168 nr̥ tyanti, √ nr̥t (Class IV) 130,
Proto-Indo-European (PIE) 14 – 15, 136 – 137; paridahyate, pari- +
88 – 89, 91, 186 – 187 √ dah (Class I) 165 – 166; pariśuṣyati,
punctuation 19, 45 – 46 pari + √ śuṣ (Class IV) 163 – 164;
paśyati, √ dr̥ś (Class I) 103 – 104,
Ṛgveda 1, 17, 28, 92, 97, 172 122 – 123; paṭhanti, √ paṭh (Class I)
122; pūjayanti, √ pūj (Class X)
sandhi: consonant sandhi 145 – 146; 122 – 123; rohati, √ ruh (Class I)
visarga sandhi 78, 141 – 145, 148, 109; śaknomi, √ śak (Class V)
151 – 152, 162, 164; vowel sandhi 165 – 166; sīdanti, √ sad (Class I)
147 – 148, 151 163 – 164; sīdati, √ sad (Class I)
Sarasvatī (river) 17 143; sraṃsate, √ sraṃs (Class I)
schwa 187 165 – 166; tyaktum, √ tyaj (Class I)
semivowels: Devanāgarī script 54 – 55, 133; vadanti, √ vad (Class I) 151,
62 – 63; pronunciation 41 – 42 161; vadati, √ vad (Class I) 152;
sibilants: Devanāgarī script 55, 63; yacchati, √ yam (Class I) 105, 122
pronunciation 42 – 44 verb roots
√ ag (Class I) to move tortuously.
Upaniṣads 1, 97 – 98, 150 √ aj (Class I) to drive; propel.
√ bhram (Class I) to wander; ramble.
Vedic 8 – 9, 13, 125, 165, 167, 172 √ bhū (Class I) to be.
verb forms: abhavan, √ bhū (Class I) √ budh (Class I) to awaken.
137, 139; āgacchanti, ā + √ gam √ car Class I) to go; move.
(Class I) 106 – 107; agacchat, √ cur (Class X) to steal.
√ gam (Class I) 45, 136, 138, 161; √ dah (Class I) to burn.
āgacchati, ā + √ gam (Class I) √ dhr̥ (Class I) to protect preserve.
107, 122, 152 – 153; akurvata, √ div (Class IV) to shine; be bright.
√ kr̥ (Class VIII) 77 – 78, 159 – 160, √ dr̥ ś (Class I) to see.
162; avadan, √ vad (Class I) 144, √ gam (Class I) to go.
147; avadat, √ vad (Class I) 136, √ gr̥ ̄ (Class IX) to make known; teach.
138, 144; avasthātum, ava- + √ han (Class II) to strike down; kill.
√ sthā (Class I) 165 – 166; bhavati, √ iṣ (Class IV) to want; desire.
√ bhū (Class I) 108, 129, 136 – 137; √ jan (Class IV) to be born.
bhramati, √ bhram (Class I) √ jīv (Class I) to live.
165 – 166; corayati, √ cur (Class X) √ kam (Class I) to love; long for.
131, 136 – 137; gacchāmi, √ kr̥ (Class VIII) to do; make; create.
√ gam (Class I) 126; gacchati, √ likh (Class VI) to write.
√ gam (Class I) 45 – 46, 105 – 106, √ muc (Class VI) to let loose; release.
153; gamaya, √ gam (Class I) √ nr̥ t (Class IV) to dance.
154 – 155; gantum, √ gam (Class I) √ paṭh (Class I) to recite.
133, 135 – 136, 138; hantum, √ han √ pūj (Class X) to worship.
(Class II) 133, 136 – 137, 167, 169; √ ram (Class I) to rest; enjoy; rejoice.
 Index and list of verbal roots 193

√ ruh (Class I) to grow. 131, 133, 165 – 169, 182; mood

√ sad (Class I) to sit. 124, 183; present tense 126 – 127,
√ sādh (Class I) to attain a goal. 129 – 131; tense 124, 188; voice
√ śak (Class V) to be able. (active/parasmaipada) 124 – 125;
√ śam (Class IV) to exert; come to rest. voice (middle/ātmanepada)
√ sraṃs (Class I) to slip; fall. 124 – 125, 164, 166; voice (passive)
√ śri (Class I) to fix on; turn towards. 124 – 125, 166
√ sthā (Class I) to be; stand. virāma 48 – 50, 53, 65, 70, 78, 152
√ śuṣ (Class IV) to dry; parch. visarga 31, 46 – 48, 141, 155; see also
√ tyaj (Class I) to renounce; abandon. sandhi
√ vā (Class II) to be extinguished. vowel mutation 90, 94 – 95,
√ vad (Class I) to speak; address. 126 – 129, 147, 189; guṇa 95,
√ vid (Class II) to know. 127 – 128, 147, 181; vr̥ ddhi
√ yam (Class I) to give; support. 127 – 128, 147, 189
verbs: finite 131, 134 – 136, 180 – 181; vowels: Devanāgarī script 55 – 58;
imperfect tense 131 – 132; infinitive pronunciation 26 – 31