English Consonants

RESEARCH PAPERS | 23
How Many Consonant Sounds Are There in English?

How Many Consonant Sounds Are There in English? by David Deterding, National Institute of Education, Singapore

.......................................................................................................................................................................................................
Most analyses agree that there are 24 consonant sounds in English. However, it is valuable to consider in some detail a few issues that affect the status of these consonants. First, we can think about why the affricates /tʃ/ and /d / are treated as single consonants rather than sequences of two consonants. Second, one might discuss why it is that /w/ and /j/ are classified as consonants rather than vowels. Third, there is the possibility of a voiceless counterpart of /w/ that, for some speakers, differentiates which from witch. And finally, there is the question of whether the velar nasal /ŋ/ is actually an allophone of /n/. After considering these issues, most people will still conclude that there are 24 consonants in English. However, the discussion can help us gain a deeper understanding of English phonology.

.......................................................................................................................................................................................................

s Introduction
How many consonant sounds do you think there are in English? Of course, most of us know that there are 20 consonant letters in our alphabet (or 21 if you include ‘y’), but here we are talking about sounds, not letters. And there is a mismatch between sounds and letters: sometimes two letters combine to represent one sound, so that ‘s’ + ‘h’ combine to represent the sound /ʃ/ and ‘t’ + ‘h’ combine for /θ/, and sometimes one letter is pronounced as a sequence of two sounds, as ‘x’ is usually /ks/. So the number of consonant letters in our alphabet is irrelevant when considering the number of consonant sounds (phonemes) in English. The basic answer to the original question is that there are 24 consonant sounds in English: q 6 plosives : /p b t d k / q 9 fricatives : /f v θ ð s z ʃ h/ q 2 affricates : /tʃ d / q 3 nasals : /m n ŋ/ q 1 lateral-approximant : /l/ q 3 approximants : /w j r/ However, things are never quite as simple as that in the study of languages, and there are a number of issues that we might consider in more depth: q Why are /tʃ/ and /d / regarded as single phonemes and not as sequences of two phonemes? q Why are /w/ and /j/ regarded as consonants and not vowels? q Do those people who distinguish which from witch have one extra phoneme, / /, a voiceless equivalent of /w/? q Should /ŋ/ really be regarded as a separate phoneme? Or can it be analysed as an allophone of /n/?

s The status of /tʃ/ and /d /
The two affricates are each written as a sequence of two symbols, so why do we regard them as single consonants? Why do we not, for example, analyse cheese /tʃi z/ as having two consonants at the start, /t/ followed by /ʃ/? The answer is that /tʃ/ behaves phonologically as a single sound, even if phonetically it is rather similar to a plosive followed by a fricative. In analysing its behaviour, we need to think about the patterns of distribution of /t/ and /ʃ/ (Laver, 1994:365), so we should consider what sequences of sounds can occur together, particularly at the start of a syllable. English allows quite complex syllable onsets, such as /str/ in string and /spl/ in splash, but it does not generally permit a plosive followed by a fricative, so */pf k/, */tsɒ / and */kʃp/ are not possible words of English. (In the few cases where the spelling does suggest a plosive followed by a fricative at the start of the word, such as psychology, the plosive is actually silent.) But note that chip /tʃp/ and check /tʃek/ are perfectly good words of English. So if we treated /tʃ/ as a sequence of two phonemes, we would have to make a special exception to the rule that an English word cannot begin with a plosive followed by a fricative. Note that /tʃ/ can also occur at the end of a word, as with catch /k tʃ/ and rich /rtʃ/, and there are no other instances where /ʃ/ can occur after a plosive at the end of a word, as */k kʃ/ and */rpʃ/ are not possible words in English. The situation with /d / provides even stronger evidence. The consonant / / is rather rare in English, and apart from in some recent loan words such as genre / ɒnrə/, beige /be /, and rouge /ru / (all of which still sound rather foreign), / / can only occur in the middle of a word, mostly between two vowels, as in pleasure
Copyright © 2005 Singapore Tertiary English Teachers Society1, 2005 STETS Language & Communication Review, Vol. 4, No. (STETS)

w

24 | David Deterding

s The status of /w/ and /j/
If you say /w/ and drag it out, it sounds rather like /u /, and similarly /j/ sounds rather like /i / (Roach, 2000:64). If they sound like vowels, why do we classify /w/ and /j/ as consonants? Sometimes it is valuable to make a distinction between a contoid and a consonant: contoids are articulated with an obstruction in the vocal tract, but consonants are sounds which can occur at the edge of a syllable (Laver, 1994:147-8). In other words, contoid is a phonetic term which describes the articulation of a sound, while consonant is a phonological term which describes its behaviour within a syllable. From the phonetic perspective of articulation, we find that plosives, fricatives, nasals, and the lateral approximant /l/ are all contoids, because they all involve a constriction in the vocal tract, but /j/ and /w/ (and maybe /r/ as well) are not contoids. But now we should consider phonological behaviour and thereby determine which sounds should be classified as consonants. Let us think about what can occur before /et/ to create a monosyllabic English word. We have words such as bet /bet/, pet /pet/, set /set/, net /net/, and debt /det/, but not */ɒet/ or */ et/, so we regard /b p s n d/ as consonants because they occur at the edge of a syllable, but /ɒ / are vowels. However, note that we can also have wet /wet/ and yet /jet/. This confirms that /w/ and /j/ are consonants. In one other aspect of behaviour, we can consider the distribution of the indefinite articles a and an: a occurs before consonants, while an occurs before vowels, and this depends on the pronunciation and not the spelling, so it is an hour not *a hour because /aυə/ begins with a vowel (the ‘h’ is silent). But note that we have a waste and a year, not *an waste and *an year, and notice that once more this is based on pronunciation and not on spelling, as it is a university (which begins with /j/) and not *an university. So again we see that /w/ and /j/ behave as consonants, not as vowels (Roach, 2000:64).

s The status of /ŋ/
In standard phonemic analysis, we assume that if the occurrence of a sound can be predicted from the surrounding sounds, it is regarded as an allophone and not as a phoneme. So, for example, we treat [ ], the dark /l/ sound that occurs at the end of a word such as fill, as an allophone of /l/ because we can specify that it only occurs in the coda of a syllable (or as a syllabic consonant in words such as bottle), unlike its clear counterpart which occurs before a vowel. So what about /ŋ/? Note that /ŋ/ can also only appear in the coda of a syllable, and furthermore we can predict that /ŋ/ rather than /n/ will always occur before another velar sound, such as in bank /b ŋk/ and anger / ŋ ə/. So should /ŋ/ be regarded as an allophone of /n/ (and then be written as [ŋ] rather than /ŋ/)? The crucial test for a phoneme is the existence of a minimal pair: if there are two words which only differ with respect to one sound distinction, then we know that we have two separate phonemes. For example, we know that /f/ and /v/ are different phonemes of English because of the existence of the minimal pair fan /f n/ and van /v n/ where the only difference is in the initial consonant, and similarly the difference in the final sound of back /b k/ and bag /b / establishes /k/ and / / as separate phonemes of English. On this basis, we can be confident that /n/ and /ŋ/ are different phonemes, because we have many minimal pairs such as sin /sn/ and sing /sŋ/, and also ran /r n/ and rang /r ŋ/. This would seem to be the end of the story, but of course it is not.

s The possibility of / /
Do you make a distinction between which and witch? For most speakers, these two words are homophones as they are both pronounced as /wtʃ/, though many Americans do make a distinction (Wells, 1982:126), and most
Copyright © 2005 Singapore Tertiary English Teachers Society (STETS)

w

w

w

w

/ple ə/ and measure /me ə/. But notice that jet /d et/ and barge /bɑ d / are perfectly good words in English. So if we were to regard /d / as a sequence of two separate sounds, we would have to say that / / can only occur near the start or at the end of a word if it is preceded by /d/, which would be rather strange. So the claim that /tʃ/ and /d / are single sounds in English is well-founded, because they behave phonologically like single sounds in the structure of English words. However, one might note that Ladefoged (2001:27) does treat both these English affricates as sequences of two sounds, partly because his emphasis is rather more on phonetics than on the phonological structure of English.

Scottish speakers also do (Wells, 1982:408). Indeed, it was once normal for all speakers of English to make this distinction, but by the end of the eighteenth century even educated southern speakers no longer maintained it (Mugglestone, 2003:132). For speakers who retain this distinction, it might be necessary to include an extra phoneme, with / / representing the voiceless counterpart of /w/, so that which is / tʃ/ while witch is /wtʃ/ However, even here the analysis is not so simple. Historically, this sound was a consonant cluster /hw/, parallel to other clusters beginning with /h/, such as /hr/, /hn/ and /hl/ (Cruttenden, 2001:215). These others have now disappeared, so apart from the possibility of /hw/, the only remaining consonant cluster involving /h/ is /hj/ in words such as huge /hju d / and human /hju mən/. And even the status of this is doubtful, as one might alternatively regard /ju / as a diphthong (Deterding, 2004). So, from a historical perspective, / / might be treated as /hw/. But from a synchronic perspective, we should note that the contrast between / / and /w/ is parallel to the contrast between many pairs of consonants in English, such as /t/ and /d/, /s/ and /ʃ/, and /f/ and /v/. The fact that the voiceless/voiced contrast is wellestablished in English lends support to the treatment of / / as a phoneme in its own right. We might therefore conclude that some speakers do have this extra phoneme.

w

v

w

v

How Many Consonant Sounds Are There in English?

| 25

so it’s fun being with them

[F9-f:40]

In fact, extra velar plosives also occasionally get inserted at the end of words such as selling, studying and young in relatively informal Singapore data (Lim & Deterding, 2005), as shown in the following examples also from the NIECSSE corpus: selling um decorative stuff [iF9-c:83] that I was studying … this [iF9-c:238] when we were young … we used to erm [iF10-e:180] If a velar plosive gets inserted occasionally after /ŋ/, maybe we should analyse it as present in the underlying representation of the word, and then instead of saying that it sometimes gets inserted, we should state that it sometimes fails to get deleted. And if this is the case,

s Conclusion
It is still basically true that there are 24 consonants in English, though it may under some circumstances be possible to regard /tʃ/ and /d / as sequences of two sounds, some speakers may have an extra phoneme / /, and the status of /ŋ/ is questionable. Even though we can conclude that there are 24 consonants in English, consideration of some of the issues regarding the phonological analysis of English can give us a deeper understanding of the structure of the sound system of the language.

STETS Language & Communication Review, Vol. 4, No. 1, 2005

w

v

v

An alternative possibility is to say that words like sing have a silent / / at the end, and this silent / / gets deleted when it occurs at the end of a word (Roach, 2000:68). In fact, for some speakers of English, this deletion rule does not apply and sing is pronounced as [sŋ ] (Roach, 2000:67), so clearly for speakers such as this, we should analyse [ŋ] as an allophone of /n/. Furthermore, in careful pronunciation, some speakers insert a velar plosive at the end of words such as being, and this can occur in Singapore English (Setter & Deterding, 2003) as is evident from the following utterance from the NIECSSE corpus (Deterding & Low, 2001):

then the distribution of [ŋ] is entirely predictable, so it is an allophone and not a phoneme. Finally we might note that words such as long /lɒŋ/, strong /strɒŋ/ and young /j ŋ/ have no final / /, but there is a / / when a comparative suffix is added: longer /lɒŋ ə/, stronger /strɒŋ ə/, younger /j ŋ ə/. So this seems to lend further support to the possible existence in the base form of these words of a final / / which gets deleted in some circumstances. (But note that there is no / / with the –ing suffix or the agentive –er suffix: singing /sŋŋ/ and singer /sŋə/; not */sŋ ŋ/ and */sŋ ə/.) In conclusion, we can say that, on the basis of minimal pairs, /ŋ/ is generally regarded as a phoneme of English, but that there are some counter-arguments which raise a few questions about its status.

26 | David Deterding

REFERENCES Deterding, D. (2004). How many vowel sounds are there in English? STETS Language & Communication Review, 19(10): 19-21. Deterding, D. & Low, E.L. (2001). The NIE corpus of spoken Singapore English (NIECSSE). SAAL Quarterly, 56: 2–5. Ladefoged, P. (2001). A course in phonetics (4th edition). Fort Worth: Harcourt College Publishers. Laver, J. (1994). Principles of phonetics. Cambridge: Cambridge University Press. Lim, S.H. & Deterding, D. (2005). Added final plosives in Singapore English. In D. Deterding, A. Brown and E.L. Low (Eds.), English in Singapore: Phonetic research on a corpus, pp. 37-42. Singapore: McGraw Hill.

Mugglestone, L. (2003). ‘Talking proper’: The rise of accent as a social symbol (2nd edition). Oxford: Oxford University Press. Roach, P. (2000). English phonetics and phonology: A practical course (3rd edition). Cambridge: Cambridge University Press. Setter, J. & Deterding, D. (2003, August). Extra final consonants in the English of Hong Kong and Singapore. Paper presented at the International Conference of Phonetic Sciences, Barcelona. Wells, J. (1982). Accents of English. Cambridge: Cambridge University Press.

Copyright © 2005 Singapore Tertiary English Teachers Society (STETS)

References: Deterding, D. (2004). How many vowel sounds are there in English? STETS Language & Communication Review, 19(10): 19-21. Deterding, D. & Low, E.L. (2001). The NIE corpus of spoken Singapore English (NIECSSE). SAAL Quarterly, 56: 2–5. Ladefoged, P. (2001). A course in phonetics (4th edition). Fort Worth: Harcourt College Publishers. Laver, J. (1994). Principles of phonetics. Cambridge: Cambridge University Press. Lim, S.H. & Deterding, D. (2005). Added final plosives in Singapore English. In D. Deterding, A. Brown and E.L. Low (Eds.), English in Singapore: Phonetic research on a corpus, pp. 37-42. Singapore: McGraw Hill. Mugglestone, L. (2003). ‘Talking proper’: The rise of accent as a social symbol (2nd edition). Oxford: Oxford University Press. Roach, P. (2000). English phonetics and phonology: A practical course (3rd edition). Cambridge: Cambridge University Press. Setter, J. & Deterding, D. (2003, August). Extra final consonants in the English of Hong Kong and Singapore. Paper presented at the International Conference of Phonetic Sciences, Barcelona. Wells, J. (1982). Accents of English. Cambridge: Cambridge University Press. Copyright © 2005 Singapore Tertiary English Teachers Society (STETS)