The building blocks of language: describing speech sounds
Phonetics, the study of speech sounds, has three main branches:
Articulatory phonetics explores the way in which speech sounds are produced by the vocal organs.
Auditory phonetics examines the way in which speech sounds are identified by the hearer.
Acoustic phonetics studies the physical properties of sound waves produced by the activity of the vocal organs.
Of these, articulatory phonetics is the longest established. Because the most familiar descriptions of speech sounds are based on their articulation, we shall focus primarily in this chapter on the articulatory description of the sounds of English.
We will familiarize ourselves with the symbols of the IPA, and learn the bases for the description of vowels and consonants in any language, before considering suprasegmental phenomena such as stress and intonation, which go beyond individual speech sounds.
According to an ancient Hindu myth, the god Indra, in response to an appeal made by the other gods, attempted for the first time to segment speech into its separate elements. After he accomplished this feat, according to the myth, the sounds could be regarded as language. Indra thus may be the first phonetician.V. Fromkin & R. Rodman, An Introduction to Language (6th edition, Harcourt Brace College Publishers, 1998), p. 214
The vocal tract
Speech sounds are produced by the organs of the vocal tract, which modify the airstream from the lungs. In the larynx or ‘voice box’, situated at the top of the windpipe, are the vocal cords, the gap between which is called the glottis. When these cords are kept close enough together and made to vibrate as air passes through the glottis, the sound produced is voiced; where there is no vibration the sound is voiceless. All vowels in English are voiced; consonants may be voiceless or voiced.
Speech sounds may also be oral or nasal, depending on whether or not the velum (or soft palate) is lowered to allow air to pass through the nose as well as the mouth. English has only oral vowels, but nasal and oral consonants; French, Portuguese and Polish have nasal vowels as well as nasal consonants. Consonants are produced by full or partial obstruction of the airstream, while vowels are produced by positioning the tongue in different configurations which do not impede the flow of air. The distinction, however, is not a clear-cut one and some sounds classified as consonants have vowel-like qualities.
You might find it helpful to think in terms of a continuum or sonority hierarchy with highly vocalic or ‘vowel-like’ sounds like \[a] (as in British English rat) at the top and strongly consonantal sounds like \[p] in pip at the bottom. Stressed syllables in English must have a vowel as their head or nucleus (see Chapter 5), while Slovak, for example, allows sounds further down the sonority hierarchy, such as \[r] in the place name Brno, to occupy this position.
In general, consonantal sounds show greater constriction of the vocal tract than vocalic sounds and have less prominence.(Clark and Yallop 1995: 36)
Spotlight: The International Phonetic Alphabet (IPA)
First published in 1888, and revised several times since (most recently in 2005) in the light of new discoveries, the International Phonetic Alphabet provides a notational standard for the phonetic representation of all languages. The IPA uses symbols representing speech sounds, which are conventionally placed in square brackets. Unlike conventional written characters or graphemes, IPA symbols have a constant value, so while the letter g is pronounced differently in gap, gnaw and beige, the symbol \[g] always represents a voiced velar plosive, i.e. the sound in gap.
The founders of the International Phonetic Association were French and British linguists, so it is not surprising that IPA symbols are heavily influenced by those languages. Many of the symbols are taken from Latin script and the cardinal vowels chosen as anchor points correspond fairly well to those of standard French. Other symbols have been drawn from other writing systems, for example \[θ] (the first consonant in **th**ink) is taken from Greek, while its voiced equivalent \[ð] (as in this) is the letter eth from Icelandic. Diacritics on symbols provide further precision where required: \[
], for example, indicates a vowel more open than \[o] but closer than \[ɔ]; a straight diacritic below a consonant indicates that it is syllabic, as for example for the second syllable in listen [lis].
As the phonetics of less familiar languages have become better known, entirely new symbols have had to be created. Some Southern African languages, for example, have clicks produced by a sucking action in the mouth. A dental click \[|] (or \[ʇ]) represents the tut, tut! sound used in English to convey disapproval, while the lateral click \[‖] (or \[
]) is the sound often used by English speakers to get a horse to move. The symbol for a bilabial click \[⊙] reflects the lip-closure akin to kissing with which this sound is produced.
The consonants of English
Try pronouncing the word rapid, and then say rabid (to rhyme with rapid). Say these words together a few times, and pay particular attention to the middle sound in each case. You’ll notice that your lip and tongue positions are exactly the same for both words, but they are nonetheless quite distinct. The other sounds are not significantly different, so in what way are the consonants different? To answer this question, try saying just the p and b sounds, conveniently transcribed in IPA as \[p] and \[b] respectively, rapidly several times in succession, holding your larynx (or Adam’s apple) between thumb and forefinger as you do so. (Don’t be shy: you’re doing this for linguistic science.) Now do the same for t and d (\[t] and \[d] in IPA). You’ll feel your larynx vibrating for \[b] and \[d], but not for \[p] and \[t], because the former are voiced sounds, while \[p] and \[t] are voiceless.
We can therefore distinguish a number of pairs of speech sounds (or phones) on the basis of whether they are voiceless (like \[p], \[t] and \[k] or voiced (like \[b], \[d] and \[g]). We can now subdivide them further on the basis of their place of articulation. The \[p] and \[b] sounds you produced earlier, for example, are produced by the lips coming together and being released: the tongue is not really involved at all. These sounds are therefore bilabial. By contrast, for \[t] and \[d] the tongue touches the back of the teeth and the alveolar ridge to produce a dental-alveolar sound, while contact between tongue and soft palate produces the velar sounds \[k] and \[g]. Other English consonants are labio-dental (bringing bottom lip and top teeth together), palatal (the tongue meets the hard palate), alveolar-palatal (closure is made at the point where hard palate and alveolar ridge meet) or glottal (involving full or partial closure of the glottis). The full set of English consonants, with examples, is shown on the next page:
Figure 4.1: The consonants of English
The final parameter for the description of consonants is manner of articulation, of which five kinds are relevant to English. When you produced \[p] and \[b] earlier, you felt a build-up of air pressure behind your lips, which was suddenly released to produce the sound. Similarly, the consonants \[t, d, k, g] are produced by releasing pressure behind the tongue. Sounds like these are called plosives. Plosives are literally small explosions in the vocal tract, and therefore cannot be extended. They are very different, therefore, from the \[f] and \[v] sounds in _f_i_v_e, which you can extend for as long as you have enough breath. For these sounds, the airstream passes through the articulators to produce audible friction, hence the name fricatives. As was the case for the plosives, the fricatives come in voiceless/voiced pairs: \[f] and \[v] are voiceless and voiced labio-dental fricatives respectively; the other English fricative pairs shown in the diagram above are \[
, Ʒ] \[s, z] and \[θ, ð]. Some consonants, known as affricates, combine a plosive and a fricative at the same place of articulation, as in the case of \[t
] and \[dƷ] in English.
The plosive sounds we identified earlier involve a complete closure of the vocal tract: they are, in other words, kinds of stop consonants. But not all stops are plosives: another set of consonants involves complete closure but these consonants are nonetheless continuous sounds: these are the nasals \[m, n, ŋ], which are produced by allowing air to pass through the nasal cavity.
A final group of consonants, known as approximants, do not require closure of the vocal tract at all: the sounds are produced instead via the passage of air between articulators, which are close but not touching. Try producing the sound at the start of leap, and you’ll notice that the tip of the tongue touches the roof of the mouth around the alveolar ridge, but the passage of air which produces the sound actually passes along the sides of the tongue: for this reason this sound \[l] is known as a lateral approximant. For most English speakers, the r sound in road is also an approximant, but this time the sides of the tongue are raised and the air passes through the gap between to produce a central approximant \[ɹ] (see Spotlight below).
Spotlight: r-sounds
Cross-linguistically, there are a lot of different types of r sound. The Scots trilledr \[r], for example, is produced by repeated beating of the tongue tip against the alveolar ridge in intervocalic positions; the French r is usually a uvular fricative \[ʁ], produced by bringing the base of the tongue close to the uvula at the back of the soft palate (a similar variant is used in some Northumbrian English accents). Some Scots use a flapped or tappedr \[
], produced by rapidly striking the hard palate with the tongue tip, in intervocalic positions, e.g. in very: many Americans have the same sound in better or motor.
Many British English speakers have the labio-dental approximant pronunciation \[υ] associated notably with TV chat-show host Jonathan Ross and football manager Roy Hodgson: this sound is produced by bringing the bottom lip close to the upper teeth, as for \[f] and \[v] but not quite letting them touch. This sound does not, as is sometimes thought, result from a speech impediment, unless one believes that an epidemic of speech defects have emerged in the UK in the last 30 years for no apparent reason: this pronunciation is in fact one of the most rapidly advancing changes in British English and there are even credible claims that it’s becoming the majority usage among younger speakers in some English urban areas.
This pronunciation does not appear to have made any headway in North America, though the US singer Billy Joel clearly uses it in the original recording of his 1977 hit Just the Way You Are. It’s not clear why he uses this form, but it may be significant that Bronx-born Joel has the traditionally non-rhotic pronunciation of working-class New York (i.e. he does not pronounce the r sound in clever, colour or hair): the labio-dental variant is rare among rhotic speakers, and most Americans are rhotic. Joel is also of Jewish heritage, and the pronunciation was identified by Wells (1982: 303) as having once been closely associated with the Jewish community in London. We will have more to say about the New York r in Chapter 11.
A third approximant, the palatal approximant \[j], is formed by the passage of air between the tongue blade and the hard palate to produce the first sound in yacht, yew, unit. Lastly, the first sound in water and wish is produced by rounding the lips and raising the back of the tongue towards the velum, hence the name labial-velar approximant. As with the nasal consonants, all approximants in English are voiced, so the indication of voicing in the consonant description is redundant.
We now have a framework for describing consonants along three parameters: \[p] for example is a voiceless bilabial plosive; its voiced partner is \[b]; \[z] is a voiced alveolar fricative, and so on. The fundamental economy of the consonant system, which makes it relatively simple for a child to learn, is evident from the diagram above: the consonants are quite few in number and involve a small number of places and manners of articulation. Most consonants come in voiceless/voiced pairs and three positions have voiced, voiceless and nasal articulation.
The consonants of English, of course, represent but a small subset of those used cross-linguistically. The same places and manners of articulation can be combined in other ways: Scots, Dutch and Russian, for example, have a velar fricative \[x] (as in Scots loch), while Cockney and many other non-standard varieties of English have a glottal stop \[
] pronunciation in, for example, water. Other places of articulation are used, too: Arabic, for example, has a uvular stop \[q], which is often represented by q orthographically in English (e.g. in Qatar, Iraq).
Key idea: Speech sounds
Speech sounds are generally described in articulatory terms. Consonants are described in terms of voicing where relevant, place of articulation and manner of articulation.
Describing vowels
Whereas consonants can generally be described in terms of the nature of the closure in the vocal tract, vowels present a challenge in that there is no full contact between articulators, so we are left with the task of describing tongue position in vocal space. To address this problem, phoneticians have identified a number of anchor points in relation to which vowels may be located in phonetic space: these are known as cardinal vowels, and they correspond, as we will see, fairly well to the main vowels of standard French.
These cardinal vowels are located within an idealized version of the vocal tract known as the vowel quadrilateral (shown in the diagram on the next page). These vowels are numbered for ease of reference: phoneticians refer to the primary cardinal vowels 1–8, shown in bold, and the secondary cardinal vowels 9–16, shown in brackets.
Figure 4.2: Cardinal vowels
Cardinals 1–4 are all pronounced with the tongue raised at the front of the mouth, and for this reason are called front vowels. Try producing the vowel in ski, keeping your tongue perfectly still (imagine the French or Scottish pronunciation): you’ll notice that your tongue is raised almost as far forward and as close to the roof of your mouth as it can go without actually making contact. The vowel you’re producing, \[i] (Cardinal 1), is therefore a close (or high) front vowel. By contrast, the vowel \[a] in RP cat for which the mouth is generally wide open and the tongue raised at the bottom of the vocal tract, is an open (or low) front vowel. The two intermediate vowels, \[e] and \[ε], those of French thé and belle respectively (and similar to the vowels in Scots English cake and English English head), are known as half-close (or high-mid) and half-open (or low-mid).
Key ideas: Vowels and tongue position
• Vowels are described in terms of their backness (front/back), tongue height (close, half-close, half-open, open) and lip-rounding (unrounded/rounded).• The 16 cardinal vowels provide a set of reference points for the location of vowels in the vocal tract.• These vowels are located on the periphery of the vowel quadrilateral, an idealized model of the vocal tract used by phoneticians for expository purposes.
Vowels 5–8 are back vowels, for which the tongue is raised at the back of the vocal tract. Cardinal 5, \[α], corresponds to the conservative pronunciation of the vowel in the French word pas, and is very close to that of cart in RP (see Spotlight on next page). Again, the lips are not rounded, so this is an unrounded open back vowel. Cardinals 6–8, however, are all rounded. Cardinal 6, \[ɔ], is a half-open vowel, produced at the same tongue height as \[ε]. This is the vowel in French sotte; RP has a longer version of this sound in caught \[ɔː] (note that the diacritic \[ː] is used to indicate vowel lengthening; \[ˑ] is used for half-lengthened vowels). Cardinal 7, \[o], is the vowel in French beau or Scots English (but not RP) home, while Cardinal 8, \[u], brings us back to the same tongue height as for \[i], but with the tongue raised at the back of the vocal tract and with lips rounded. This is the vowel of French vous; some conservative RP speakers have a vowel close to this in goose, but most English varieties use a more fronted vowel. (The RP vowel in cat has also changed: the prescribed pronunciation was once \[æ], between Cardinals 3 and 4, but the vowel has lowered for almost all RP users since the 1960s).
Spotlight: Received pronunciation (RP)
RP is the socially prestigious but regionally unmarked pronunciation used predominantly by British newsreaders: while British English accents are regional, an RP speaker could come from anywhere within the UK, although RP’s features are predominantly southern English in origin. For all the reverence accorded to this pronunciation, RP is an accent like any other and is subject to change: the speech of BBC newsreaders who used RP in 1940s and 1950s newsreels sounds very different from that of today’s RP users. Where the latter would have the a vowel close to Cardinal 4 in hat \[hat], for example, the former would have insisted on a vowel raised to somewhere between Cardinals 3 and 4 \[hæt].
The secondary cardinal vowels 9–16 simply reverse the lip-rounding value of their primary counterparts; thus Cardinal 9, \[y] which corresponds to the vowel in French tu or German süss, is simply Cardinal 1, \[i], with firmly rounded lips. Likewise Cardinals 10 and 11, the vowels of standard French feu \[ø] and fleuve \[œ] respectively, are the rounded equivalents of 2 and 3. Cardinal 12, \[Œ], the rounded equivalent of \[a] is rare cross-linguistically, but Cardinal 13, \[ɒ] is used in RP body or cot (most US English varieties have an unrounded vowel here). Cardinal 14 \[Λ] is the conservative RP pronunciation of the cup vowel (most southern English speakers have a more fronted variant: see below). Cardinals 15 \[४] and 16 \[ɯ] are less familiar to speakers of the major western European languages, but both occur in Scots Gaelic and Thai, and Cardinal 16 is an important vowel in Japanese.
Great prestige is still attached to this implicitly accepted social standard of pronunciation. Often called Received Pronunciation (RP), the term suggesting that it is the result of a social judgement rather than of an official decision as to what is ‘correct’ and ‘wrong’, it has become more widely known and accepted through the advent of radio. The BBC formerly recommended this form of pronunciation for its announcers mainly because it was the type which was most widely understood and which excited least prejudice of a regional kind. Indeed, attempts to use announcers who had a mild regional accent provoked protests even in the regions whose accent was used.(Gimson 1980: 89)
Figure 4.3: Positions of RP English monophthongs
Helpful though the cardinal vowels are as reference points, they do not correspond closely with the vowel positions of English, which are shown in the diagrams below. For the RP vowels in push \[ɷ] (the older symbol \[ʊ] is also used) and kick \[I], for example, the tongue is retracted to a more central position from \[u] and \[i] respectively, and requires less muscular effort (for this reason these two vowels are sometimes called lax vowels). Confusingly, some of the phonemic symbols for RP vowels (the concept of the phoneme is explained in the next chapter) no longer correspond to their modern pronunciation. Textbooks still refer, for example, to /æ/ and /Λ/ in spite of the fact that, for most RP speakers, these vowels have moved to \[a] and \[ɐ] respectively. The modern positions of RP monophthongs can be seen in Figure 4.3.
Figure 4.4: RP English vowels
RP has one long central vowel, \[зː], the vowel in bird, which may bear stress, and another, schwa, \[ә], which cannot. Generally, vowels produced at the centre of the vocal tract require least articulatory effort. When asked a question and needing a bit of thinking time, many speakers naturally start their reply with ‘Er…’, the sound produced with the tongue in the central rest position. Although schwa has no letter of its own in English orthography (Bulgarian uses the Cyrillic character ъ, Turkish has ı), it is a very important sound in English and in many other languages. For most British English speakers it is the underlined vowel in each of the following words: potato, reader, banana, support, phonetic. In both English and French it can, in some cases, can be deleted altogether, so that in rapid speech, for example, sport and support can be homophonous.
DIPHTHONGS
The presence of a number of diphthongs, which show a change in vowel quality as the tongue travels between two points, is another reason why English does not fit the cardinal vowel schema particularly well.
Think about the vowel in right (and ignore the fact that the spelling still retains a gh sequence that is no longer pronounced). Try saying it slowly a number of times, and you’ll feel your tongue travel quite a distance. In RP the vowel begins somewhere near \[a] but moves to the position of the kick vowel \[I], and it is this trajectory that gives the sound its distinctive quality. The right vowel is therefore transcribed \[aI] the convention for diphthongs being to transcribe their beginning and end points. The trajectory for the RP vowel in house is even wider, starting from a similar point but ending near the push vowel \[ɷ]. The starting point for the blow vowel in RP is different: for most speakers the tongue starts in schwa position and moves to the \[ɷ] position of the push vowel again. Two further vowels, those of boy \[ɔI] and bay \[eI], make up the set of closing diphthongs in English, all of which have a close vowel as their end point, as can be seen in the figure below.
Figure 4.5:Closing RP diphthongs
Another set of diphthongs, all of which end in schwa \[ə], are known as centring diphthongs. These are the vowels in RP beer \[Iə] and square \[εə]. Some RP speakers also have a centring diphthong in poor \[ɷə] and in pore \[ɔə], while others do not, pronouncing both words identically to paw.
Figure 4.6: Centring RP diphthongs
Note also that where a centring diphthong occurs, an r generally is present in the spelling, which in RP and many other varieties of English has been replaced by schwa. In what are known as rhotic areas (for example, North America and the west of England), this r has never been lost, and there is consequently no diphthong.
Suprasegmentals
The descriptors and symbols introduced so far provide a good basis for analysing the sounds of any language. The IPA enables us, moreover, to divide up connected speech into individual sounds, or segments, which we can present in ordered sequence, for example:
The mouse ran up the clock
\[ðəmaɷsɹanɐpðəklɒk]
But the neat boundaries between phones that such sequences imply are something of a fiction. Speech sounds roll into one another, and one sound can significantly influence its neighbours. Take the vowel in ran in the example above, for instance, which for most British English speakers sounds slightly different from that of rat, having a slight nasal quality that the latter lacks. This is because speakers generally lower the velum in readiness for the nasal consonant \[n] well before the vowel has completed, with the result that nasality affects both segments and cannot be seen as the exclusive property of the consonant, as the broad linear transcription above suggests. A number of other phenomena can only be analysed above and beyond the level of the segment: these are known, appropriately enough, as suprasegmentals.
Key ideas: Segments of speech
• Speech is continuous and does not divide neatly into discrete sounds, in the way that written words and sentences are built from individual letters and spaces. For this reason phoneticians refer to their divisions of the speech chain as segments.• IPA symbols can be used to represent the segments of a speech chain on the page, e.g. cat \[kat].
One type of suprasegmental is stress, which refers to the relative prominence of one syllable (see Chapter 5) over another in a word. In English, for example, the sequence of segments in the noun increase and its corresponding verb increase is the same, but the two forms sound different because a different syllable (underlined here) is stressed in each case. An unstressed vowel is sometimes reduced in quality, being given less prominence and articulatory effort. The first syllable in photograph for example has the diphthong \[əɷ] but in unstressed position in photography this reduces to \[ə]. Stress is generally indicated by a raised diacritic before the stressed syllable, so for the examples above photograph \[ٰfəɷtəgɹαːf] but photography \[fəٰtɒgɹəfi]. Stress is a relative concept, referring to the prominence of one syllable with respect to another, and involves a combination of pitch (stressed syllables have a higher frequency or pitch than unstressed ones), loudness or intensity and possibly vowel length. Length itself is also a relative rather than absolute concept, or an inherent quality of a speech sound itself. The vowels \[ɔː] caught and \[αː] cart, for example, are viewed as long vowels because English speakers generally pronounce them longer than vowels such as \[ɒ] and \[a], but it is important to remember here that one speaker’s \[ɔː] may be shorter than another’s \[a].
Other important suprasegmental phenomena include intonation and tone, both of which involve changes in pitch within a word or sentence. For example, a simple English sentence like You see him every Saturday would be pronounced with a falling intonation if uttered as a statement, but with a rising intonation at the end if intended as question (You see him every Saturday?). Orthographically, question marks can provide a rough and ready indication of rising intonation, but in most cases readers have to deduce the appropriate intonation for themselves, as conventional writing lacks the resources to make intonation patterns clear. Linguists generally indicate only as much information as the context demands, either via intonation contour lines above the speech string or arrows after the relevant sequence to show the intonation pattern involved, for example a fall ↘; a rise ↗, or a rise–fall ↗↘. In tone languages word-level intonation is important for distinguishing meaning. In Thai, for example, the same sequence of segments uttered with a level, falling or rising tone will have a completely different meaning, as this example (taken from Blake 2008: 139) illustrates:
Table 4.1: Examples of tones in Thai
Fact-check
[1](answers.mdx#rfn4-1) Which of these is the IPA symbol for a voiceless velar plosive?
\[
ð]
\[
k]
\[
ø]
\[
Ʒ]
[2](answers.mdx#rfn4-2) What is the vowel \[y]?
Cardinal 7
Back, half-open, unrounded
Front, close, rounded
Back, close, rounded
[3](answers.mdx#rfn4-3) What is the consonant \[m]?
Bilabial nasal stop
Bilabial plosive
Labio-dental nasal stop
Labio-dental approximant
[4](answers.mdx#rfn4-4) What is the IPA symbol for the back, half-open rounded vowel (Cardinal 6)?
\[
o]
\[
ɤ]
\[
a]
\[
ɔ]
[5](answers.mdx#rfn4-5) What is a labio-dental approximant?
The phonetic name for the sound \[
f]
A speech impediment
The voiced sound at the start of very, vat and vole in English
A variant pronunciation of the r sound in words like rat, road and very.
[6](answers.mdx#rfn4-6) Which of the following is a closing diphthong?
\[
Iə]
\[
aɷ]
\[
εə]
\[
i]
[7](answers.mdx#rfn4-7) Which of the following statements is NOT true of RP?
It is unchanging
It is the majority pronunciation in England
It is different in northern and southern England
All of the above
[8](answers.mdx#rfn4-8) Which of the following is untrue?
\[
n] is a stop consonant
All plosives are stops
\[
p] is a stop consonant
All stops are plosives
[9](answers.mdx#rfn4-9) What do affricates combine?
A plosive and a fricative from different places of articulation
A stop and an approximant
A plosive and a fricative from the same place of articulation
A fricative and an approximant from the same place of articulation
[10](answers.mdx#rfn4-10) The distinction between \[θ] and \[ð] is one of:
Voicing
Tongue position
Manner of articulation
Place of articulation
Dig deeper
A. Cruttenden, Gimson’s Pronunciation of English (7th edition, Routledge, 2013), Chapters 1, 4, 5, 8 & 9
V. Fromkin, R. Rodman & N. Hyams, An Introduction to Language (10th edition, Wadsworth, 2013), Chapter 5, ‘Phonetics: the Sounds of Language’ (Chapter 3 or 4 in some earlier editions)
P. Ladefoged (& S.F. Disner), Vowels and Consonants (3rd edition, Blackwell, 2012), Chapters 4, 6, 11, 12 & 14
A. McMahon, An Introduction to English Phonology (Edinburgh University Press, 2002), Chapters 3, 6 & 7
G. Yule, The Study of Language (4th edition, Cambridge University Press, 2010), Chapter 3