Phonology
Agenda
• Narrow transcription (using diacritics to represent
other features of phones)
• Articulatory processes
• phones vs. phonemes
• Phonological analysis
2
Narrow
transcription
Transcribing phonetic variations
in more detail
English vs. Spanish
English Spanish
pin [ph] poco [p]
pool [ph] peso [p]
pace [ph] papa [p]
tool [th] todo [t]
tea [th] te [t]
4
Aspiration
Period of silence between the release of the closure
(e.g. stop) and the beginning of the following sound
(e.g. vowel)
[th] = aspirated [t]
5
Aspirated vs. unaspirated
consonants in English
pin spin
[phɪn] [spɪn] Unaspirated
consonants occur
after [s] in English
tall stall
[thɑl] [stɑl]
6
Aspiration
7
Aspiration in English
[ph], [th], [kh]
…anything in common?
(hint: think about the articulatory features)
8
All voiceless stops (natural class)
9
Articulatory processes
as rules in the mind
(Narrow) Transcription
time [thaIm] aspirated
tree [tʃɹi] palatalized
pot [phat̚] unreleased
stop [stap] unaspirated
water [waɾɚ] flap (voiced)
glottal stop
button [bʌʔṇ]
11
But we have the intuition that they are
all “t” sounds…
(in fact, they are all spelled with ‘t’ in English)
12
How do we store the pronunciations of
“water” and “button”?
[waɾɚ] or [watɚ]
[bʌʔṇ] or [bʌtṇ]
13
Hypothesis A: [waɾɚ] must be stored as a
special pronunciation of “water” (i.e.
simple memorization)
Hypothesis B: the pronunciation of “t” as a
flap [ɾ] is derived by a rule
14
Is the pronunciation
predictable?
[th] [ɾ] [ʔ]
time water button
top little kitten
table butter carton
notable
15
Patterns (Generalizations)
[th] occurs at the beginning of stressed
syllables
[ɾ] occurs between two vowels
[ʔ] occurs in the middle of words before a
syllabic nasal ([ən] or [ṇ]).
16
Articulatory Processes
• Sounds are pronounced differently in
different environments
These are RULES in mental grammar
17
Phonological Rule
The sound /t/ changes to [ʔ] before a syllabic
nasal
/t/ → [ʔ] / __ [ṇ]
18
Phonological Rule
The sound /t/ changes to [ɾ] between
vowels
/t/ → [ɾ] / ’V __ V
19
Different pronunciations of /t/
in English
[ɾ]
Phonological rules MAP
[tʃ] mental representations
/t/ to surface forms
[t]
Why do we all think that
[th] these surface forms are “the
same sound” ?
What’s stored in
the mind Surface forms
(i.e., actual articulation) 20
The way words are stored in the
Mental Lexicon
time /taım/
tree /tɹi/
pot /pat/
stop /stap/
water /watɚ/
button /bʌtən/
21
[waɾɚ] and [wɑɾəɹ] are both acceptable transcriptions
and refer to the same pronunciation
Lexicon Rules Output
/wɑtɚ/ /t/ → [ɾ] / ‘V_ V [wɑɾɚ]
Phoneme Allophone
Note: // (phonemic) vs. [ ] (phonetic) 22
Phoneme
• the abstract representation of a sound
• the way the sound is stored in word in the
mental lexicon
23
Allophone
• how the sound is actually produced in a given
environment
• an instance of a phoneme
24
Allophone
Environment =
the phonological context of a sound
position in a syllable/word
neighboring sounds
25
Summarizing so far…
There are RULES in the mental grammar
that determine the pronunciations of
sounds
26
Building a Grammar
What is stored What we hear
Lexicon Output
/tɑım/ RULES [thaIm]
/tɹi/ [tʃɹi]
/pɑt/ [phat̚]
/stɑp/ [stap]
/wɑtəɹ/ [waɾɚ]
/bʌtən/ [bʌʔṇ] 27
Different pronunciations of /t/
in English
[ɾ]
[tʃ]
/t/
[t]
[th]
Phoneme
Allophones
28
Articulatory processes are
• Rules in the mind
• Not automatic reflexes of the vocal tract
Different processes in different languages and
dialects
29
Note for this class
• Focus on understanding
• WHY we think there are rules
• HOW we find and state those rules
• It is less important to remember
• WHAT those rules are, or what the relevant language data
looks like
30
How to find
phonemes and rules
across languages
Inventory of Phonemes
• What are the phonemes and allophones in a
language?
• We need a way to identify/distinguish
phonemes and allophones
• Steps in finding out the phonemic status of
sounds (Phonemic Analysis)
32
Phonemic analysis steps
1. Look for minimal pairs (contrastive
distribution?)
2. If no minimal pairs, then is there a
complementary distribution for particular
sounds? → Search for generalizations
3. Generalizations → Rules
33
Phonemic Analysis - 1
First step: look for minimal pairs
• if difference between sounds causes a difference in
meaning
=> contrastive
=> different phonemes
34
Phonemic Analysis - 1
• Are [p] and [b] different phonemes in English?
pin vs. bin [pɪn] vs. [bɪn]
tap vs. tab [tæp] vs. [tæb]
→ contrastive distribution
Conclusion:
/p/ and /b/ are separate phonemes 35
Phonemic Analysis - 1
[wɑɾɚ]
[wɑtɚ]
... no difference in meaning!
→ free variation; optional rules
[ɾ] and [t] are allophones of /t/
36
Phonemic Analysis - 1
• Lexicon – collection of Sound + Meaning pairs
• Sound information in lexical items is encoded
in terms of phonemes
• That’s why only phonemes affect meanings
37
Phonemic Analysis - 2
Second step:
If no minimal pairs, look for a pattern
(distribution of the two sounds)
• What environments does sound X occur in?
• What environments does sound Y occur in?
38
Phonemic Analysis - 2
can /k(æ)n/
I can ask [aɪ kn æsk]
I can see [aɪ kn si]
I can bake [aɪ km beɪk]
I can play [aɪ km pleɪ]
I can go [aɪ kŋ goʊ]
I can gather [aɪ kŋ gæðəɹ]
39
Phonemic Analysis - 2
I can ask [aɪ kn æsk] k_æ
I can see [aɪ kn si] k_s
I can bake [aɪ km beɪk] k_b
I can play [aɪ km pleɪ] k_p
I can go [aɪ kŋ goʊ] k_g
I can gather [aɪ kŋ gæðəɹ] k_g
40
Phonemic Analysis – 3
Generalizations
[m] occurs before a bilabial consonant
[ŋ] occurs before a velar consonant
[n] occurs everywhere else (elsewhere)
→ Complementary distribution (they do not
occur in the same environment)
41
Phonemic Analysis – 3
Generalizations => Rules
/n/ becomes [m] before a bilabial consonant
/n/ → [m] / ___ bilabial stop (nasal & oral)
/n/ becomes [ŋ] before a velar consonant
/n/ → [ŋ] / ___ velar stop
elsewhere /n/ is pronounced [n]
/n/ → [n] / elsewhere 42
Important notes
• Generalizations ≠ Rules!!!
• Generalizations: patterns you see in the data (e.g. x
occurs in y environment)
• Rules: statement about the mental operations that
convert phonemes to allophones (e.g. x becomes y
in z environment)
43
Question 1: Which is the
underlying phoneme?
[m] occurs before a bilabial stop
[ŋ] occurs before a velar stop
[n] occurs everywhere else
→ [n] has a wider distribution; likely to be the
default for Tuesday)
44
Question 2:
But don’t [m] and [n] belong to
separate phonemes in English?
• How can you show that [m] and [n] are
separate phonemes?
45
[m] and [n]
/n/ /m/
[n] [m] [ŋ] [m]
Some phonemes have overlapping surface
realizations
46
Allophones and distributions
/p/
[ph] [p]
a) Superman and Clark Kent
How do we know that they are actually the same person?
→ They can NEVER occur in the same environment at the
same time.
b) Ice, water, steam:
Are all H2O but have a different manifestation depending on
the environment they occur in:
<0C → occurs as ice
> 0 C and < 100 C → occurs as water
> 100 C → occurs as steam
47
Phonemic analysis steps
1. Look for minimal pairs (contrastive
distribution?)
2. If no minimal pairs, then is there a
complementary distribution for particular
sounds? → Search for generalizations
3. Generalizations → Rules
48
Natural class
Earlier example: /n/
can /k(æ)n/
I can ask [aɪ kn æsk]
I can see [aɪ kn si]
I can bake [aɪ km beɪk]
I can play [aɪ km pleɪ]
I can go [aɪ kŋ goʊ]
I can gather [aɪ kŋ gæðəɹ]
50
Generalizations => Rules
/n/ becomes [m] before a bilabial consonant
/n/ → [m] / ___ bilabial stop
/n/ becomes [ŋ] before a velar consonant
/n/ → [ŋ] / ___ velar stop
elsewhere /n/ is pronounced [n]
/n/ → [n] / elsewhere 51
More data: /t/ and /d/
• hat trick [hæt tʃɹIk]
• hit batsman [hIp bætsmən]
• night class [nɑIk klæs]
• bad dream [bæd dʒɹim]
• head band [hɛb bænd]
• bad guy [bæg gaI]
52
More rules
• /n/ → [m] / ___ (bi)labial consonant
/n/ → [ŋ] / ___ velar consonant
/n/ → [n] / elsewhere
• /t/ → [p] / ___ (bi)labial consonant
/t/ → [k] / ___ velar consonant
/t/ → [t] / elsewhere
• /d/ → [b] / ___ (bi)labial consonant
/d/ → [g] / ___ velar consonant
/d/ → [d] / elsewhere
53
Back to features…
• /n/ = voiced, alveolar nasal (stop)
• /t/ = voiceless, alveolar stop
• /d/ = voiced, alveolar stop
54
Back to features…
• /n/ = voiced, alveolar nasal (stop)
• /t/ = voiceless, alveolar stop
• /d/ = voiced, alveolar stop
• Are there any other alveolar stops in English?
55
56
• /n/, /t/ and /d/ form a Natural Class of alveolar
stop.
• Natural class: group of sounds in a language that
share one or more articulatory property to the
exclusion of all other sounds in that language.
• Revised rule:
Alveolar stop → velar stop / __ velar stop
57
Why natural classes?
• For the following data, what’s wrong with stating a rule like
(1) instead of (2) or (3)?
I can bake [aɪ km beɪk]
I can play [aɪ km pleɪ]
(1) /n/ becomes [m] before [b]; /n/ becomes [m] before [p]
(2) /n/ becomes [m] before a bilabial stop
(3) /n/ becomes [m] before a bilabial consonant
58
Stating Rules
• Whenever possible, you MUST use natural class in
your rule statement!
• The more general the rules are, the better
• Makes strong, easily-testable predictions
e.g., /n/ becomes [m] before a bilabial consonant
Prediction: The same change in e.g. “I can wink”?
59
Stating Rules
• Whenever possible, you MUST use natural class in
your rule statement!
• The more general the rules are, the better
• Makes strong, easily-testable predictions
• Fewer rules to store in the mind (simpler, parsimonious theory)
• Fewer rules for children to learn (remember the paradox)
60
Natural Class Exercise
• Which phones belong to the following
natural classes?
• high vowels
• back vowels
• voiceless fricatives
• voiced bilabial
• labial consonants
61
Assimilation
• /n/ → [m] / ___ bilabial stop
• /n/ → [ŋ] / ___ velar stop
• /n/ → [n] / elsewhere
• What feature assimilates?
• Place of articulation
62
Dissimilation
Greek
/epta/ → [efta]
/ktizma/ → [xtizma]
What dissimilates here?
Manner of articulation
63
Insertion
Plural forms of…
fox, ditch, bush, orange, maze
• Schwa insertion (epenthesis) in English
64
Deletion
• He handed her his hat.
• /h/ - deletion
65