FROM:

RENÉ DIRVEN AND MARJOLIJN VERSPOOR

COGNITIVE EXPLORATIONS OF LANGUAGE AND LINGUISTICS
AMSTERDAM: JOHN BENJAMINS. 1998

CHAPTER 5

The sounds of language:

Phonetics and Phonology

5.1. Introduction: Phonetics and phonology

Human beings can make an infinite number of speech sounds. If you say
the same word several times, or ask different people to say the same
word, there will be differences between the pronunciations. In spite of
these differences, we would still want to say that the pronunciations are
in some important respects “the same.”
The same holds with the written language. The following symbols
all have a different shape. Nevertheless, we are able to regard the
different shapes as examples of the “same” entity, i.e. the “first letter of
the alphabet.”

A a A a a A a
A very general cognitive ability is involved here: categorization, i.e. the
ability to perceive different things as examples of the same category.
108 COGNITIVE EXPLORATION OF LANGUAGE AND LINGUISTICS

One aspect of knowing a language is the ability to categorize the great

variety of speech sounds heard in that language. The sound categories
that a speaker of one language recognizes will not necessarily coincide
with those that a speaker of another language will “hear.” Speakers of
Thai hear the two “p”-sounds in pie and spy as different; conversely, for
the Japanese, the “s” and “sh” sounds in sushi are the same.
Here we have the basis of the distinction between phonetics and
phonology. Phonetics studies speech sounds as sounds, in all their
complexity and diversity, independent of their role in a given language.
Phonology studies speech sounds as these are categorized by speakers of
a given language. In standard British English, there are about 45
different categories of speech sounds, called phonemes. As the
languages of the world go, English is about average. Some languages
have fewer phonemes (Japanese has about 20). Others, e.g. the Bushman
(“Khoisan”) languages of Southern Africa, have over a hundred, amongst
which a very intricate system of click sounds.

5.1.1. Spelling and pronunciation

Some languages (e.g. Spanish) have a writing system that is (almost) a

phonemic one, i.e. each phoneme is always represented by the same
letter, and vice versa. But in English the relationship between
pronunciation and spelling is, as we all know, far from perfect. There are
various reasons for this. First there are more phonemes (about 45) than
there are letters of the alphabet (26). Next, there are historical reasons:
when English spelling was standardized, many centuries ago, it was
broadly phonemic in character. Spelling has remained virtually the same,
while pronunciation has changed considerably over the centuries, and
continues to do so. The vowels of English have been especially
“unstable.” Moreover, English has borrowed from other languages.
Words of foreign origin may be spelt according to the rules of the donor
language, thereby introducing numerous “irregularities” into English
spelling. Examples include French borrowings like rouge, chateau,
champaign, quiche. Furthermore, spellings have sometimes been
influenced by speakers’ beliefs about etymology (etymological
spelling). Debt is a borrowing from Old French dette. The “b” was never
pronounced, but was inserted to show the supposed relation of the word
to Latin debitum. Finally, there is a very marked tendency for a given
morpheme always to be spelled the same way, even though its
pronunciation may vary from context to context. The is spelled the same
 PHONETICS AND PHONOLOGY 109

way in the man and the apple, although it is pronounced differently. You
can recognize the invariant spelling of the root morphemes in
photograph and photographer, clean and cleanse, sign and signature,
family and familiar, even though the morphemes are pronounced
differently in each case.
Speakers sometimes attempt to re-establish the link between spelling
and pronunciation, not by changing the spelling, but by modifying the
pronunciation. At the beginning of this century, waistcoat was
pronounced /wesköt/ or /weskit/, to rhyme with biscuit. The current
pronunciation /weistkout/ is a spelling pronunciation; the pronunciation
is based on the conventional spelling. Speakers who pronounce the “t” in
often are likewise being influenced by the spelling.

5.1.2. Phonetic symbols

Because spelling is not a faithful representation of pronunciation, it is
useful to have a set of special symbols whose values are generally agreed
upon. This is the function of the phonetic symbols of the International
Phonetic Alphabet (IPA). These symbols are in general use amongst
linguists and are employed in this book. Most modern dictionaries now
give pronunciations of words using these symbols.

5.2. Production of speech sounds

We can distinguish two main stages in the production of speech sounds:
phonation and articulation.

• Phonation stands for the airstream becoming voiced or voiceless as

explained further on. As air is expelled from the lungs, it passes
through the glottis (located behind the “Adam’s apple”). Located in
the glottis are the vocal folds - two flaps of flesh that can be brought
together or held apart. Phonation refers to the modulation of the
airstream in the glottis. If the vocal folds are brought together, they
may vibrate, to produce voice. If air passes freely through the glottis,
the air stream is minimally affected (this is the state of voicelessness).
110 COGNITIVE EXPLORATION OF LANGUAGE AND LINGUISTICS

• Articulation refers to the creation of a special resonance space for

each sound. This involves the shaping of the vocal tract (i.e. the
tubular structure above the larynx), by adjustment, in the oral cavity,
of the tongue, jaw, velum (soft palate), lips, etc. The great variety of
speech sounds that we are able to make depends very largely on the
manner in which we shape the vocal tract.

Phonation and articulation will be discussed in more detail below.

5.2.1. Phonation

If you clasp your hand tightly over your larynx while saying the word
zoo, you should be able to feel a certain vibration. The vibration is that
of the vocal folds, technically known as voice. Both [z] and [u] are
voiced sounds.
If you repeat this exercise while saying a prolonged [s], you should
feel no vibration in the larynx. [s] is a voiceless sound.
For the production of voice, the vocal folds are brought together.
When air is pushed out from the lungs, it encounters the vocal folds as an
obstacle. Air pressure builds up under the folds until the folds are
literally blown apart, and air escapes through the glottis. The folds then
return to their original position. Air pressure builds up again, and the
cycle is repeated. Each opening and closing cycle is very brief. In men,
the frequency ranges from about 80 to 150 cps (cycles per second), in
women, from about 120 up to 300 cps. For children, the rate may be
even higher.
The frequency of the opening and closing cycle determines the pitch
of the sound; the higher the frequency, the higher the perceived pitch.
The auditory sensation of pitch is produced by the pattern of regular
bursts of air passing through the glottis.
Voicelessness ensues when the vocal folds are completely brought
apart. When air from the lungs reaches the larynx, it encounters no
obstacle, and flows freely though the glottis. Whisper is characterized by
voicelessness throughout. For obvious reasons, it is not possible to
produce a voiceless sound with pitch, or with pitch variations!
Many speech sounds come in pairs of voiced and voiceless. These
sounds are collectively known as obstruents (so-called because the
airstream, in passing through the oral cavity, is “obstructed” to a marked
degree). Here is a list of the English obstruents, in their voiced/ voiceless
pairings.
 PHONETICS AND PHONOLOGY 111

VOICED VOICELESS

[b] ”big” [p] ”pig”

[d] ”do” [t] ”too”
[g] ”gum” [k] ”come”
[v] ”vine” [f] ”fine”
[th] ”them” [th] ”thin”
[z] ”zoo” [s] ”Sue”
[zs] ”measure” [sh] ”mesh”
[dzs] ”jeer” [tsh] ”cheer”

The other main class of consonants, the sonorants, are typically voiced.
(We can think of sonorants as the ”hummable” consonants.) These
include the nasals [m], [n] and [ng], the liquids [l] and [r], and the glides
[j] and [w].

[m] ”me”
[n] ”knee”
[ng] ”sing”
[l] ”love”
[r] ”ray”
[j] ”yes”
[w] ”when”

5.2.2. Articulation

The second major component of speech production is articulation, i.e.

the shaping of the vocal tract as air passes through it. Aspects of
articulation will be studied in the next sections, which deal with the
characterization of consonants and vowels.

5.3. Consonants
Consonants and vowels are distinguished mainly in terms of the degree
of constriction in the vocal tract. Consonants involve some major
constriction, which obstructs the airflow at some point. Vowels on the
112 COGNITIVE EXPLORATION OF LANGUAGE AND LINGUISTICS

other hand merely involve a distinctive shaping of the oral cavity, with
relatively little impedance of the air flow.
Consonants can be described in terms of two major parameters: the place
in the vocal tract at which constriction occurs (place of articulation),
and the nature of the constriction (manner of articulation).

5.3.1. Places of articulation

In the articulation of a consonant, a movable articulator (usually some

part of the tongue, or the lips) is moved towards a more stable articulator
(e.g. the upper teeth, or some part of the palate). The following terms
describe the more common places of articulation.

• bilabial [p, b, m]. The lower lip articulates with the upper lip

• labiodental [f, v]. The lower lip articulates with the upper teeth

• dental [T, D]. The tongue tip articulates with the top teeth

• alveolar [t, d, n, l, s, z]. The tongue tip articulates with the alveolar
ridge. Also many articulations of ”r”.

• alveopalatal [sh, zs]. The tongue front (excluding the tip) articulates
with the back part of the alveolar ridge.

• palatal [j]. The tongue blade articulates with the back part of the
alveolar ridge.

• velar [k, g]. The tongue back articulates with the velum (soft palate).
[k, g] before back vowels, as in core, gore.

5.3.2. Manner of articulation

Manner of articulation describes the kind of constriction that is made.

The following are the major categories:

• stops (plosives) [p, t, k, b, d, g]. Stops are made by completely

blocking the airstream at some point in the oral cavity.
 PHONETICS AND PHONOLOGY 113

• fricatives [f, v, th, th, s, z, sh, zs]. They are made with a very narrow
gap between the articulators. The airstream passes through this gap
under high pressure, causing friction.

• affricates [tsh, dzs]. They are complex sounds, consisting of a stop

followed immediately by a fricative at the same place of articulation.
Here, the stop is gradually released so as to form a narrow constricted
gap for the air to flow through; this is the fricative part of the affricate.

• approximants [r, l, j, w]. They are articulated with only minimal

constriction; consequently, there is virtually no friction. In most
varieties of English, word initial “r” is an approximant. It is articulated
by moving the tip of the tongue towards the alveolar ridge, deflecting
the air over the tongue without causing friction.
It is useful to distinguish several subcategories of approximants:
− laterals [l]. In laterals the air flows along the sides (or along one
side) of the tongue. For [l], the tip of the tongue forms a
complete closure centrally against the alveolar ridge (as in a
stop), but the side(s) of the tongue is/ are lowered, and air is
deflected between the side(s) and the gums.
− glides [j, w]. They are very short unstable versions of vowels,
functioning in syllable structure as consonants. The initial [j] of
yes is actually a kind of [i], whilst the [w] of we is a short
version of [u]. Notice that [w] has a prominent bilabial (“lip-
rounding”) component also.
− trills. Here, one articulator is let to vibrate in the outflowing air
stream. Scottish pronunciation of “r” is often an alveolar trill,
with the tongue tip vibrating under the alveolar ridge. The “r” in
some varieties of French and German is typically an uvular trill
[R], with the uvula in vibration.
− flaps are produced when the tongue strikes agains the alveolar
ridge once in passing. American speakers often articulate “t” and
“d” as flaps, especially when “t” and “d” occur intervocalically
(between vowels), e.g. matter, city, medal.

• A difficult sound to classify is [h]. It is a kind of fricative, and

essentially involves an open glottis, and no further significant
modification of the airstream.
114 COGNITIVE EXPLORATION OF LANGUAGE AND LINGUISTICS

• nasals [m, n, ng]. They involve a blocking of the oral airstream, by

lowering of the velum. Thus, the air is allowed to escape through the
nasal cavities.

5.4. Vowels
As will be remembered from the previous section, consonants involve
some major obstruction of the airflow at some point of the vocal tract.
Vowels differ from consonants in that there is relatively little impedance
of the air flow, but the oral cavity is shaped in many different ways and
this gives rise to the different vowels and diphthongs. Vowels are more
difficult to describe than consonants. There are three reasons for this:

(a) Because there is no constriction of the vocal tract, it is often

diffi-
cult to describe precisely the posture adopted by the oral cavity;
(b) Vowel categories tend to ”overlap” and ”merge into” each other,

much more than consonant categories;

(c) Vowels tend to vary from accent to accent. Accent, in this sense,
means the regional or social differences in pronunciation. What
makes the different varieties of English sound so different, is
mainly the vowels.

Since the tongue is the instrument par excellence to determine the

posture adapted by the oral cavity, vowel sounds are described primarily
in terms of the position of the tongue. Two parameters are important.

(a) front vs. back. The highest part of the tongue may be towards
the
front of the mouth, or towards the back;
(b) high vs. low (also called close vs. open). The degree to which
the
tongue is raised.

Independent of these two aspects are the following:

(a) lip position. The lips may be rounded, or spread;

 PHONETICS AND PHONOLOGY 115

(b) duration. A vowel can be long or short;

(c) nasalization. A vowel can be oral or nasal.

5.5. Phonemes and allophones; phonemic transcription

Just as one word may have many different senses and the exact sense of
the word does not become really clear until it is used in a context, sounds
may have many variations, too, dependent on the sounds surrounding it.
In the next sections we will look at the terms used for a “family of
sounds” and where the different “family members” may occur.

5.5.1. Definitions

The “p” sound in pin is different from the “p” sound in spin; the former
is aspirated [ph], the latter unaspirated [p]. Yet, in an important sense,
we want to say that the two “p” sounds of English, in spite of their
phonetic difference, are variants of the same sound. The term phoneme
designates the more abstract unit, of which [ph] and [p] are examples.
[ph] and [p] are allophones of the same phoneme, /p/.
By convention, phonemes are written between slashes //, while
allophones (or, more generally, sounds considered in their phonetic
aspects) are written between square brackets [ ].

Figure 3. A phoneme and its allophones

/p/ (phoneme)

[p] [ph] (allophones)

116 COGNITIVE EXPLORATION OF LANGUAGE AND LINGUISTICS

Two languages may classify their sounds in different ways. English and
Spanish both have [d] and [th]. For English speakers, the sounds are
different (they are categorized as different phonemes), and they serve to
distinguish word meanings (den vs. then). For the Spanish speaker, the
two sounds are merely variants of the same phoneme. Thus, [d] occurs
word-initially, while [th] occurs intervocalically. Compare donde
“where” [donde] and lado “side” [latho]. If you pronounce lado with a
[d], you get a variant pronunciation of lado, but you do not get a
different word.
Whereas English speakers regard aspirated and unaspirated voiceless
stops as variants of the same sound, Thai speakers do not. Compare
Thai /phaa/ “split” and /paa/ “forest.”
As this example shows, a simple way of deciding whether two
sounds in a language belong to one phoneme or to two different
phonemes is to look for minimal pairs. A minimal pair is a pair of
words that are identical in all respects except for the sounds in question.
The minimal pair pat, bat confirms that /p/ and /b/ constitute separate
phonemes in English. On the other hand the impossibility of a contrast
between [sphai] and [spai], or between [phai] and [pai] confirms that
[ph] and [p] do not belong to different phonemes in English.

5.5.2. Free variation and complementary distribution

The precise amount of aspiration in English stops (as in the initial stops
in pat, cat, tat) is not linguistically relevant. Stops with different degrees
of aspiration are in free variation. When sounds are in free variation, it
basically doesn’t matter which sound you select, and the meaning of an
utterance is not affected.
Another situation is where one allophone occurs exclusively in one
environment (context), another allophone occurs exclusively in another
environment. The sounds are then said to be in complementary
distribution.
Native speakers are usually quite unaware of the extent of allophonic
variation in their language. English speakers think of the /p/ in pie and
the /p/ in spy as “the same sound;” it is only after studying phonetics that
one realizes that they are in fact very different sounds! Speakers’
intuitions thus reflect a knowledge of the phonemic structure of their
language, rather than its phonetic reality.
PHONETICS AND PHONOLOGY 117