ALLOPHONES:

De-voicing of plosives
Consider how the voiced bilabial plosive /b/ is articulated in the word ball /bɔl/. You will probably need
to say it aloud to both here and feel the effect. Now try saying aloud the word nib /nɪb/, again focusing
on the phoneme /b/. Do they sound the same? For most speakers, the /b/ at the beginning of the
word ball /bɔl/ is fully voiced – it is a strong sound. However, the /b/ at the end of the word nib /nɪb/ is
typically less forceful – it is not fully voiced. This phonetic realization is referred to as being de-voiced
and this de-voiced allophone of /b/ is represented symbolically as [b̥ ]. We can say, therefore, that /b/
has at least two allophones, [b] and [b̥ ].
Of course, only voiced consonants can be de-voiced and the remaining two voiced plosives /d, g/ are
similarly de-voiced in word-final position, e.g.
bed /bɛd/ → [bɛd̥ ]
bag /bæg/ → [bæɡ̊ ]
De-voicing of fricatives
The voiced fricatives /v/, /ð/ and /z/ are also similarly voiced in word-final position, e.g.
have /hæv/ → [hæv̥ ]
seethe /sið/ → [siːð̥ ]
haze /heɪz/ → [heɪːz̥ ]
As we have seen, the voiced post-alveolar fricative /ʒ/ typically occurs intervocalically in English.
Exceptionally, when it does occur word-finally in loan words it may also be subject to de-voicing, e.g.
beige /beɪʒ/ → [beɪːʒ̊ ]
garage / gærɒʒ/ → [gærɒːʒ̊ ]
There are, however, only a handful of words where /ʒ/ occurs in word-final position and so the
frequency of occurrence for de-voicing of the post-alveolar fricative is extremely low.
De-voicing of affricates
As we know, there is only one voiced affricate in English, the voiced post-alveolar affricate /ʤ/. As
with plosives and fricatives, this affricate is also de-voiced in word-final position, e.g.
badge /bæʤ/→ [bæʤ̊ ]
De-voicing of approximants
Of the four approximants, only the liquids /r/ and /l/ show any appreciable de-voicing in particular
contexts. There is a de-voiced allophone of the phoneme /r/ which occurs after the voiceless
consonants /p, t, k, f, θ, ʃ/. It is represented symbolically as [ɹ̥ ] and occurs in words such as the
following.
pray /preɪ/ → [pʰɹ̥ eɪː]
tray /treɪ/ → [tʰr̥ eɪː]
cry /kraɪ/ → [kʰɹ̥ aɪː]
free /fri/ → [fɹ̥ iː]
three /θri/ → [θɹ̥ iː]
shrew /ʃru/ → [ʃɹ̥ uː]
The fully voiced allophone is represented symbolically as [ɹ] and this occurs elsewhere, e.g.
run /rʌn/ → [ɹʌn]
carry /kærɪ/ → [kʰæɹɪ]
Thus, the phoneme /r/ has at least two allophones: [ɹ] and [ɹ̥ ].
The lateral /l/ is de-voiced when it immediately follows either the voiceless bilabial plosive /p/ or the
voiceless velar plosive /k/. For example:
play /pleɪ/ → [pʰl̥ eɪː]
clay /kleɪ/ → [kʰl̥ eɪː]
The fully voiced allophone [l], however, occurs before vowels, e.g.
lay /leɪ/ → [leɪː]
early /ɜlɪ/ → [ɜːlɪ]
Consequently, the phoneme /l/ has at least two allophones: [l] and [l̥ ].

Aspiration of plosives
We have already highlighted how certain phonemes may be produced with an accompanying short
puff of air in certain contexts. We considered the production of the voiceless alveolar plosive /t/ in the
words top and stop (see What Is an Allophone?) In the word top we saw that the /t/ is aspirated, i.e.
/tɒp/ → [tʰɒp]. In the word stop, however, we saw that there is typically no aspiration, i.e. /stɒp/ →
[stɒp]. We can say, therefore, that the phoneme /t/ has at least two allophones: [t] and [tʰ].
Aspiration also affects the two other voiceless plosives: the alveolar /p/ and the velar /k/. In fact, they
are affected in the same way as /t/ when they occur in the same environments.
For example, /p/ is unaspirated when it follows /s/ (e.g. spin /spɪn/ → [spɪn]) but aspirated in a word
such as pin, where /pɪn/ → [pʰɪn]. The phoneme /p/, therefore, has at least two allophones: [p] and
[pʰ].
Similarly, /k/ is also unaspirated following a /s/ consonant (e.g. skin /skɪn/ → [skɪn]) but aspirated in a
word such as kin (/kɪn/ → [kʰɪn]). The phoneme /k/ must also, therefore, have at least two allophones:
[k] and [kʰ].
We see, then, that each of the voiceless plosives /p/, /t/ and /k/ has at least two allophones: an
aspirated allophone [pʰ], [tʰ] and [kʰ], and an unaspirated allophone [p], [t], and [k]. The unaspirated
allophones only occur after /s/.
Only plosive sounds can be aspirated and so there are no examples of nasals, fricatives, affricates, or
approximants being aspirated. Note also that only voiceless plosives are aspirated, i.e. the voiced
plosives /b/, /d/ and /g/ are not aspirated.
 Vowel nasalization
Say the word art /ɑt/ aloud. Now contrast this by saying the word mart /mɑt/. Do you notice any
difference in the quality of the /ɑ/ vowel? If you are uncertain, say the words aloud in quick
succession: art – mart – art – mart… If you listen carefully, you may be able to detect a nasal quality
to the vowel in the word mart: it is nasalized.
In mart, the vowel is immediately preceded by the nasal consonant /m/. In this context, it is almost as
if the /m/ lends its nasal feature to the neighboring vowel. The neighboring vowel then appears to
assimilate this feature, becoming nasalized.
This transformation of a speech sound owing to the influence of a neighboring speech sound also
functions when the nasal consonant follows the vowel. Try saying the words add /æd/ and Anne /æn/
in quick succession: add – Anne – add – Anne… Can you detect any difference in the nasal quality of
the vowel? If not, try saying the sequence again but, this time, pinch your nose with your thumb and
index finger to prevent air escaping out of the nose. Can you now detect that in the word Anne there
is a slight resonance in the nasal cavity that begins on the vowel /æ/? This is because when vowels
occur in words immediately before a neighboring nasal consonant, they appear to borrow the nasal
feature of the following consonant, assimilating this to become nasalized.
The nasalized vowel is transcribed as [ɑ̃]. As we have seen, nasalization of vowels typically occurs
when the vowel immediately precedes, or follows, a nasal consonant /m, n, ŋ/, as in words such as
man [mæ̃ n], now [naʊ̃ː] and wing [wɪ̃ŋ].
We can conclude that the phoneme /a/ has at least three allophones: [ɑ], [ɑː] and [ɑ̃].
Similarly, every other vowel will have a nasalized allophone when the phoneme appears either
immediately before or immediately after a nasal consonant.

Dentalization of plosives
We have seen (see Plosives) how the plosive sound /t/ is articulated with the tongue tip contacting
the alveolar ridge, i.e. it is an alveolar plosive. Whilst this is the predominant place of articulation,
consider the placement of the tongue tip in the word eighth /eɪtθ/. In this instance, the tongue is not
raised to the alveolar ridge but is brought into contact with the back of the upper incisors. In other
words, it is dentalized, being produced in the same place as the dental fricative /θ/ it precedes, i.e.
eighth /eɪtθ/ → [eɪt̪ θ]
This example amply demonstrates the dentalization of /t/, here represented as [t̪ ]. However, as the
voiceless alveolar plosive does not follow /s/ it will also be aspirated in this context. Consequently, the
full representation of this phonetic variation is as follows.
eighth /eɪtθ/ → [eɪt̪ ʰθ]
We now see that /t/ has at least three allophones: [t], [tʰ] and [t̪ ʰ].
The voiced counterpart alveolar plosive /d/ is similarly dentalized when it appears before a dental
fricative, e.g.
width /wɪdθ/ → [wɪd̪ θ]
We can conclude, therefore, that the phoneme /d/ has at least three allophones: [d], [d̥ ] and [d̪ ].
Only alveolar sounds are dentalized when they appear before a dental fricative. The remaining
alveolar sounds of English are the nasal /n/, the two fricatives /s/ and /z/, and the approximants /r/
and /l/. Of these, only /n/ and /l/ can appear before a dental fricative in British English and each may
be dentalized. [In reality, there are very few examples of English words that contain an alveolar
immediately before a dental sound. However, this principle of the dentalization of alveolars is a
common feature of connected speech, which will be discussed later.]
Dentalization of nasals
As indicated above, only alveolars are dentalized before a dental fricative, and the only alveolar nasal
is /n/. Two examples follow.
tenth /tɛnθ/ → [tʰɛn̪ θ]
plinth /plɪnθ/ → [pʰl̥ ɪn̪ θ]
The non-dentalized allophone [n] occurs in words such as no [nəʊ] and can [kæn]. Thus, the
phoneme /n/ has at least two allophones: [n] and [n̪ ].
Dentalization of approximants
As pointed out already, the only alveolar approximant to undergo dentalization before a dental
fricative is the lateral /l/, e.g.
health /hɛlθ/ → [hɛl̪ θ]
This example demonstrates the dentalization of /l/ very well. However, we must also take account of
the fact that the /l/ appears before a consonant. Consequently, the /l/ is also velarized. The complete
representation of this allophonic variation, therefore, is as follows.
health /hɛlθ/ → [hɛɫ̪ θ]
We may now conclude that the phoneme /l/ has at least four allophones, [l], [l̥ ], [ɫ] and [ɫ̪ ]. In fact, this
is the total number of allophones for this phoneme in English, no more are to be found.

Glottalization – consonants
We know that during the production of voiceless sounds the glottis is open: the vocal folds are held
gently apart (abducted) and are relaxed (see Making Speech Sounds). This is also the state of the
glottis for restful breathing. During the production of voiced sounds, however, the glottis is opened
and closed in rapid succession as the vocal folds vibrate. Two further states of the glottis can be
identified. One is when the vocal folds are held relatively close together under tension but without
vibration. This is the configuration adopted for whispering. The final state is when the vocal folds are
held firmly together, closing off the glottis all together. This configuration is adopted when lifting
weights; where it serves to prevent air escaping from filled lungs, thereby holding the chest in an
expanded, rigid position against which the arms can gain better leverage. This state can also be
momentarily adopted during speech to produce what is known as a glottal stop. It is produced by the
sudden shutting and opening of the glottis and is transcribed in IPA as /ʔ/.
Its use is restricted in English. It can occur intervocalically, where it serves to separate two vowels,
e.g.
co-owner /kəʊəʊnə/ → [kʰəʊːʔəʊːnə↓]
co-operation /kəʊɒpəreɪʃən/ → [kʰəʊːʔɒpʰəɹeɪʃən]
In certain contexts, it can also serve as glottal reinforcement for the voiceless plosives /p, t, k/. This
means that the plosive is produced near simultaneously with the glottal stop. Examples include the
occurrence of /p/ intervocalically such as in the word happy or in syllable-final position preceding a
vowel such as in the word jumper.
The glottal stop is also used widely in some accents of English. In Cockney (an accent used by
natives of East London) it famously features in words like butter and matter which are pronounced
[bʌʔə↓] and [mæ̃ ʔə↓] respectively. Glottal reinforcement is also widely used in Geordie (an accent
used by natives of Tyneside, in north-east England).
About allophonic variation in standard British English, a main variation is the glottalization of the
voiceless alveolar plosive /t/ when it occurs in syllable-final position, e.g.
hat /hæt/ → [hæʔ]
Consequently, the phoneme /t/ has at least five allophones: [t], [tʰ], [t̪ ʰ], and [ʔ].

Velarization – consonants
We have noted elsewhere (see Approximants) that the lateral /l/ is formed by the tongue tip creating a
complete closure at the alveolar ridge and the air stream being allowed to escape over the sides of
the tongue laterally. This configuration appears in words such as look /lʊk/, silly /sɪlɪ/ and like /laɪk/.
However, in certain contexts, the quality of the phoneme is quite different. Compare the production
of /l/ in the words let /lɛt/ and always /ɔlweɪz/. In the word let, the /l/ appears before a vowel and it is
articulated with tongue tip closure against the alveolar ridge and lateral air escape. However, in the
word always, the /l/ now appears before a consonant. This time it is articulated with the body of the
tongue slightly flatter, as the back of the tongue is raised towards the velum. The sound is, therefore,
said to be velarized. We represent this allophone symbolically as [ɫ]. This allophone occurs before
consonants, as we have seen, and word-finally (e.g. ball /bɔl/ → [bɔːɫ]). It is sometimes referred to as
dark l.
We can now conclude, therefore, that /l/ has at least three allophones: [l], [l̥ ] and [ɫ].

Summarizing