Syllabic writing systems (syllabaries)

There are some languages whose writing systems are based on the syllable, rather than the individual vowel and consonant sounds. The kana (hiragana and katakana) system of Japanese is the most familiar example of this (Bowring and Laurie 2004), but there are many others: Akkadian (Mesopotamia, extinct), Bopomofo (China, Taiwan), Cherokee (Southeast USA), Linear B (Greece, extinct), Mayan (Central America), Pahawh Hmong (Laos, Vietnam), and Vai (Liberia).

Ability to identify syllables

Everyone, regardless of their native language and its writing system, seems to be able to identify by and large how many syllables words contain (but see the section Problems in syllabification below). That is, it is a unit that people are consciously aware of. “[I]ndeed, explicit awareness of syllables [by children] has been shown to developmentally precede explicit awareness of phonemes” (Gnanadesikan 2008).

Importance to literacy

Literacy experts are agreed that an awareness of the syllables in a word, the sounds that make up the syllables, and of phenomena such as alliteration and rhyme (see below) are essential for efficient spellers of English (Carson, Gillon, and Boustead 2013; HearBuilder n.d.; Justice et al. 2013; Moats 2010; Moats and Tolman 2009; Wilson 2013).

Other reasons relate to the place of the syllable in linguistic analysis.

Hierarchy of phonological units

The syllable fits nicely into a hierarchy of phonological units. Features (such as [± voice], [± labial]) are present in segments (vowels and consonants). Segments make up syllables. Syllables combine into feet, units used in the analysis of speech rhythm, and tone groups, units used in intonational analysis, may be composed of one or more feet.

Stress

Stress in words is placed on syllables rather than individual vowel and consonant phonemes. For example, the noun insight and the verb incite have identical phonemes. The difference in their pronunciation is the stress placement, on the first or second syllable: insight /ˈɪnsaɪt/ˌ incite /ɪnˈsaɪt/, where /ˈ/ marks the start of the stressed syllable. Similarly, the intonational focus (tonic; see Chapters 8 and 10 on intonation) of an utterance falls on a particular syllable, rather than a phoneme or word. For instance, an utterance “It’s absolutely ridiculous!” is likely to have the tonic (probably a fall from high to low) on the second syllable of ridiculous.

Combinations of phonemes

The syllable is the largest unit that is required for accounting for the combinations of phonemes in a language. For instance, is the sequence /mftfr/ possible in English? The answer is “yes”, as in the phrase triumphed frequently. However, because of syllable constraints, there must be a syllable division between /mft/ and /fr/.

Allophonic realization rules

Many of the rules accounting for the occurrence of variants (allophones) of sounds (phonemes) can only be stated in terms of the syllable. For example, many accents of English distinguish between a “clear” /l/, with the tongue bunched upwards and forwards towards the hard palate, and a “dark” /l/, with the tongue bunched upwards and backwards towards the soft palate; in these accents, clear /l/ occurs at the beginning of syllables, as in lick, while dark /l/ occurs at the end of syllables, as in kill.

Differences between languages

Some differences between languages in the occurrence of sounds can only be stated in terms of the syllable and its structure. For example, the sounds /h/ and /ŋ/ occur in English and many other languages. However, in English /h/ can only occur at the beginning of a syllable, as in help, behave (/hɛlp, bɪ.heɪv/, where the dot marks the syllable division). However, there are languages, such as Arabic, Malay, and Urdu, where /h/ can occur at the ends of syllables, e.g., Malay basah /basah/ “wet”. Notice that, in analysing syllable structure, we are talking about sounds (phonemes); the spelling is irrelevant. Thus, while many English words end in an h letter, this letter never represents an /h/ sound. It may be silent as in messiah, cheetah, or may work in combination with another letter, as in th (path), ph (graph), sh (fish), gh (laugh) and ch (rich). Note also that, while one-syllable (monosyllabic) words may often be given here as examples, they are given as examples of syllables, not of words. Also, differences between British English (BrE) and American English (AmE) are discussed where relevant.

Likewise, while /ŋ/ can occur in syllable-final position in English, as in sang, ring, it cannot occur in syllable-initial position. However, it can in other languages (Anderson n.d. a, n.d. b) including Fijian, Malay/Indonesian, Māori, Thai, and Vietnamese; for instance, the Thai word for “snake” is /ŋu:/.

Some differences in sound combinations between languages can only be stated in terms of the syllable and its structure. For example, both English and German have the /p, f, l/ sounds. While English has the sequence /pfl/ in an example such as hipflask, this is only possible because there is a syllable (and morpheme) break between the syllable-final /p/ and the syllable-initial /fl/. In German, on the other hand, words can start with /pfl/ as in pflegen “to be accustomed (to doing something)”. Therefore, while both languages have all three sounds, and both languages have sequences of /pfl/, in German these can all be in syllable-initial position, but in English they can only be across a syllable boundary in the middle of a word. This explains why English speakers find German words like this non-English and awkward to pronounce.

Number of syllables

Speakers may differ in their opinions as to the number of syllables particular words have. These differences may arise from various factors:

• Elision of /ə/: The schwa vowel /ə/ may be lost in certain environments, e.g., comfortable /kʌmfətəbəl/ (four syllables) or /kʌmftəbəl/ (three).
• Morphology: The word evening is one unit of meaning (morpheme) in good evening (usually /i:vnɪŋ/, two syllables), but is two morphemes (even + ing) in evening out numbers (usually /i:vənɪŋ/, three syllables).
• Spelling: The letter a in the spelling may lead speakers to believe there are three syllables in pedalling (thus /pɛdəlɪŋ/), but only two in peddling (thus /pɛdlɪŋ/).
• Long vowels + dark /l/: The vocalic nature of the darkness of dark /l/ may lead to different opinions about words such as boil.
• Triphthongs: Differences exist as to the syllabification of triphthongs, such as the /aɪə/ of BrE fire /faɪə/, as constituting one or two syllables.
• Compression: Sequences involving sounds such as /iə/ may be analysed as one or two syllables. For example, a word like lenient /li:niənt/ may be considered in its fullest form to be three syllables; however, the /iə/ is often compressed (or smoothed; Wells 2008: 173–174) into one syllable, and may be reanalysed as /jə/, thus /li:njənt/.

Definition of the syllable

While the syllable may seem a clear entity, there is no universally agreed definition of the syllable. Three attempts to define the syllable (one articulatory, one acoustic, and one auditory) will be discussed here.

The prominence theory of the syllable is based on auditory judgments. Syllables correspond to peaks in prominence, usually corresponding to the number of vowels.

In the sonority theory (see, for example, Ladefoged and Johnson 2010), which is probably the most reliable and useful of the three attempts to define the syllable, syllables correspond to peaks in sonority. Sonority is the relative loudness (acoustic amplitude) of sounds compared with other sounds. This can be plotted on a scale of sonority (most sonorous first):

1. Low/open vowels, such as /æ, ɑ:/
2. Mid vowels, such as /ɛ, ɔ:/
3. High/close vowels, such as /i:, u:/ and semi-vowels /j, w/ (see below)
4. The lateral-approximant /l/
5. Nasal-stops, such as /m/
6. Voiced fricatives, such as /v/
7. Voiceless fricatives, such as /f/
8. Voiced oral-stops, such as /d/
9. Voiceless oral-stops and affricates, such as /t, tʃ/

While this works in most cases, there are exceptions. For instance, the word believe /bɪli:v/ has two vowels and two peaks of sonority. However, the word spy /spaɪ/ has one vowel, but the /s/ has greater sonority than the /p/; it also therefore has two peaks of sonority. Thus, whereas both words have two peaks of sonority, the first is clearly two syllables but the second only one. Many languages do not have initial clusters like /sp/ and they are often pronounced as two syllables by foreign learners. Similarly, instances involving syllabic consonants (see below) are counterexamples, e.g., hid names and hidden aims may involve the same sequence of phonemes /hɪdneɪmz/, but the first is two syllables while the second is three (involving a syllabic consonant; see below).

The articulatory chest pulse theory relates to the contraction of the intercostal muscles surrounding the lungs as they push air out during speech. It has been claimed (Gimson 1980: 56) that the number of chest pulses determines the number of syllables. This theory has been used most notably by Abercrombie (see, for example, 1967) in differentiating between syllable pulses and stress pulses, in order to formulate a theory of rhythm in speech.

Syllable boundaries

While speakers can usually tell how many syllables a word has, there may be confusion as to where one syllable ends and the next begins. For instance, in sequences where two vowels are separated by one or more consonants (e.g., photon, pastor, outgrow, obstruct), do these consonants belong with the first or the second syllable, or are they divided between them? Various writers (e.g., Wells 1990; Eddington, Treiman, and Elzinga 2013a, 2013b; Redford and Randall 2005) have researched this, investigating the features that correlate with syllabification preferences and proposing principles to account for them. Reasonably uncontroversial principles are the following:

• Syllable boundaries cannot divide the affricates /tʃ, dʒ/.
• Syllable divisions cannot create clusters that are otherwise impermissible. Thus, panda can be considered /pæn.də/ or / pænd.ə /, but not / pæ.ndə /, as /nd/ is an impermissible initial cluster.
• Syllable boundaries occur at morpheme boundaries. For instance, loose-tongued, regardless of whether you think it should be written as one word, hyphenated, or two words, could be analyzed as /lu:.stʌŋd/ or /lu:st.ʌŋd/ without breaking cluster constraints; however, analysts would always break it at the morpheme division: /lu:s.tʌŋd/.

However, that still leaves a number of more controversial examples, and the following principles (which are incompatible with each other) have been proposed:

• Intervocalic consonants go with the following vowel, wherever possible. This is known as the Maximum Onset Principle.
• Consonants go with whichever of the two vowels is more strongly stressed (or, if they are equally stressed, with the preceding vowel).
• Stressed syllables cannot end with a short vowel (i.e., they must be closed with a final consonant; see below) and two consonants are split between the two syllables.
• Allophonic detail may be a strong clue. A clear /l/ between two vowels is perceived as initial in the second syllable, because clear /l/s appear in the onset position. An aspirated plosive is perceived as initial in the second syllable because aspirated plosives appear in the onset position.
• Spelling has been claimed to have some effect on syllabification judgments. In a word like yellow, the /l/ is taken to belong to the first syllable because the spelling sequence ll cannot start words in English (with the possible exception of foreign words such as llama), but can end words such as yell.

Eddington, Treiman, and Elzinga (2013a) report that “80% or more of the subjects agreed on the syllabification of 45% of the items with four medial consonants, 69% of the items with three consonants, and 80% of the words with two consonants. What is surprising is that this number drops to 50% for words with a single medial consonant in spite of the fact that only two syllabification responses are possible.” This leaves a fair amount of listener variability, even for examples with only two possible responses, which are split perfectly. Perhaps for this reason, an ambisyllabicity principle has long been proposed whereby an intervocalic consonant can be analyzed as belonging to both the preceding and the following syllable (Anderson and Jones 1974; Lass 1984: 266). By this analysis, the /b/ of a word like habit is shared between the two syllables: /[hæ [b]ɪt]/, where the square brackets show syllable boundaries.

For many purposes, these are problems that do not need to be solved. As Wells (2000: xxi), who uses spaces in the Longman Pronunciation Dictionary to represent syllable boundaries, says, “any user of the dictionary who finds it difficult to accept the LPD approach to syllabification can simply ignore the syllable spaces.”

Semi-vowels

There are many one-syllable words that have the structure /*ɛt/, that is, their peak and coda (rhyme) is /ɛt/. They include pet, bet, debt, get, jet, vet, set, met, net, het (up). These all clearly have /ɛt/ preceded by a consonant sound (/pɛt, bɛt, dɛt, gɛt, dʒɛt, vɛt, sɛt, mɛt, nɛt, hɛt/).

There are also the words yet and wet, although it may be unclear whether the initial sounds are consonants or not. In answering this, we need to distinguish between phonetic form (the way these sounds are articulated) and phonological function (the way they function in syllables). In terms of function, these sounds seem to occur in the onset position and the words have the same structure /*ɛt/. However, in terms of form, they are unlike the other consonants. If you slow down the initial sounds of yet and wet, you will appreciate that they are articulated like, and sound like, the vowels /i:/ and /u:/, as in tea and two. That is, the tongue and lips do not form any substantial obstruction to the airstream, which escapes relatively freely. Therefore, in terms of function, they are nonsyllabic, in that they do not occur in the peak position, but in terms of form they are vowel-like (vocoid). As a result, /j, w/ are often referred to as semi-vowels.

Syllabic consonants

A further complication relates to the pronunciation of words such as sudden and middle. Both words are clearly two syllables, and in their fullest form would be pronounced /sʌdən, mɪdəl/, that is, with a schwa vowel after the /d/. However, this pronunciation was described as “strikingly unusual – and even childish” by Gimson (1980: 320), and it is much more usual to run from the /d/ straight into the /n, l/ without any intervening schwa vowel.

Let us analyse this from the articulatory point of view. The /d/ sound is a voiced alveolar oral-stop (plosive). This means that the tongue comes into contact with the alveolar ridge behind the upper teeth, completely stopping the airstream. The /n/ sound is a voiced alveolar nasal-stop. Thus, in going from a /d/ to an /n/, the tongue does not move, as it is already in the required position. Instead, the soft palate (velum) leading to the nose opens, so that air escapes through the nose. This is known as nasal release of the /d/.

A similar transition occurs with /l/, a voiced alveolar lateral-approximant. Therefore, in going from an /d/ to an /l/, the velum does not move, as both sounds are oral. Instead, the tongue sides lose contact, allowing air to escape over the sides; the tongue tip maintains contact. This is known as lateral release of the /d/.

In terms of syllable structure, these pronunciations mean that we have two-syllable words, but with no vowel sound in the second syllable, as the /n, l/ sounds are clearly pronounced like consonants with substantial obstruction to the airstream (contoids). We thus analyse the consonants /n, l/ as occupying the peak position in the second syllable, and label them syllabic consonants. They are shown by a subscript tick: /sʌdn̩, mɪdl̩/. This situation is summarized in Table 5.3.

/s/ + consonant initial clusters

A common pattern for two-consonant initial clusters is for the first consonant to be /s/. Although there are 24 consonants in English, fewer than half of them can follow /s/ in a CC cluster. Consonants cannot follow themselves in clusters, that is, there is no /ss/ initial cluster. Those that can follow /s/ fall into categories by manner of articulation (the kind of sound they are):

• Oral-stops (plosives): All three voiceless oral-stops of English can follow /s/, as in span, stuck, skill /spæn, stʌk, skɪl/. Perceptive readers may wonder why the clusters /sb, sd, sg/ do not occur or, more profoundly, why the clusters in span, stuck, skill are analyzed as /sp, st, sk/ rather than /sb, sd, sg/, that is, the voiced equivalents. The answer is that they could as easily be analyzed as /sb, sd, sg/ (and identical clusters in Italian are analyzed in this way). The sounds following /s/ are (i) voiceless (as the voiceless /p, t, k/ in pan, tuck, kill are) but (ii) unaspirated, that is, there is no burst of voiceless air when the sound is released (as is true of /b, d, g/ in ban, duck, gill). In other words, the sounds resemble both /p, t, k/ and /b, d, g/ and could be analyzed either way. The fact that they are represented by p, t, k in English spelling may influence the analysis here. (Similarly, for word-medial sequences, see Davidsen-Nielsen (1974), who found that listeners could not distinguish disperse from disburse.)
• Nasal-stops: /s/ can be followed by /m, n/ as in small, snow. It cannot be followed by the third nasal-stop of English because /ŋ/ never occurs in the onset position (see above).
• Fricatives: /s/ can be followed by /f/, as in sphere, and /v/, as in svelte. However, there is clearly something odd about these clusters.
  • The /sf/ cluster only occurs in a handful of words in English: sphagnum (moss), sphincter, sphinx. Furthermore, all these words (and many other technical words with /sf/) are of Greek origin; /sf/ is a regular cluster in Greek. For these and other reasons (see below), we may consider /sf/ irregular in English.
  • Similarly, the /sv/ cluster only occurs in the one word svelte (and svengali?), and even there it may be pronounced with /sf/. This word is from Latin, via Italian and French. The name Sven is Swedish and may be pronounced with /sv/ in English. However, it is often regularized to /sw/; many Swedes regularized their surname from Svensen to Swensen when they migrated to the USA, including Earle Swensen, the founder of the restaurant chain.
• Approximants: There are four approximants in English: /l, r, w, j/. Of these, two are uncontroversial: /sl/ as in sleep and /sw/ as in swim. However, /r, j/ need further discussion.
  • The cluster /sr/ only occurs in Sri Lanka (and other clearly foreign words). It may therefore be considered non-English. Indeed, many speakers pronounce this country with /ʃri:/, that is, making the initial cluster regular, as in shrimp, etc.
  • The cluster /sj/ may occur for some BrE speakers in words like suit and assume. However, this cluster seems to be getting rarer, being replaced by one of two things. Firstly, the /j/ may disappear, leaving plain /s/, this pronunciation becoming increasingly common in British accents. Wells (2008: 790) gives a graph showing that the pronunciation without /j/ is almost exclusively preferred by younger speakers, whereas older speakers often favored /sju:t/. The second possibility is for the (underlying) /sj/ to coalesce into /ʃ/ (see Chapter 9 on connected speech processes). This is quite common in some accents with words like assume (thus /əʃu:m/). Neither of these possibilities exists in AmE /su:t, əsu:m/.

In summary, only seven consonants follow initial /s/ uncontroversially, while another four are dubious.

Plosive + approximant initial clusters

If all the permutations of the six plosives /p, b, t, d, k, g/ and the four approximants /l, r, w, j/ existed, there would be 24 possible combinations. However, only 18 of the 24 possible combinations occur, e.g., play, bring, quick /pleɪ, brɪŋ, kwɪk/. The following do not occur: /pw, bw, tl, dl, gw, gj/. There are some rare words and foreign loanwords that contain these clusters (e.g., pueblo, bwana, Tlingit, guava, guano, gules), but no common native words.

Three-consonant initial clusters can be considered a combination of the two patterns just described. In such clusters in English, the first consonant can only be /s/, the second must be a voiceless plosive /p, t, k/, and the third an approximant /l, r, w, j/, e.g., spring /sprɪŋ/, split /splɪt/, squid /skwɪd/. However, again, not all 12 possible permutations occur: /spw, stl, stw/ do not exist.

Rhoticity

Some speakers of English pronounce two /r/ sounds in the phrase car park, while others pronounce none. That is, speakers either have both or neither of what is represented by the r letter in the spelling. In phonological terms, speakers either can or cannot have /r/ in the coda position in the syllable. Accents that have syllable-final /r/ are called rhotic, while the others are nonrhotic.

This difference is pervasive throughout the phonology of accents of English. In Shakespeare’s day, all speakers of English had syllable-final /r/ (were rhotic). However, a change spread from the Southeast of England and /r/ was dropped in the coda position. This nonrhoticity change spread to most areas of England and Wales; however, it did not affect Scotland and Ireland. The status of countries that England colonized or where native speakers migrated depends on the most influential part of Britain that they came from. Nonrhotic accents include Australia, New Zealand, South Africa, Trinidad, certain eastern and southern parts of the United States, and most of England and Wales. Rhotic accents include Scotland, Ireland, Canada, Barbados, certain western parts of England, and most of the United States. Because most US speakers are rhotic, rhotic speakers are in the majority in global numerical terms.

Learners of English tend to be rhotic if (i) their native language allows syllable-final /r/ and/or (ii) AmE is more influential than BrE.

The /ʒ/ phoneme

The phoneme /ʒ/ is a peculiar one in the phonology of English. It commonly occurs in the middle of words such as vision, as the result of a historical process like that described above for assume. It is a moot point whether this is final in the first syllable or initial in the second. On occasions like this, analysts often consider whether the sound(s) can occur at the beginnings or ends of words, that is, as the onset of the first syllable or the coda of the final syllable of a multisyllable word. However, in this case this is inconclusive as no native English words begin or end with /ʒ/. There are plenty of words used in English that begin or end with /ʒ/, but they all have clear foreign origins, usually French: gendarme, genre, Giselle, je ne sais quoi, joie de vivre; barrage, beige, blancmange, camouflage, collage, cortege, dressage, entourage, espionage, fuselage, liege, luge, massage, mirage, montage, prestige, rouge.

One of three things can happen with loanwords like these:

1. If English speakers know French, and can pronounce initial or final /ʒ/, then they may pronounce it with that sound.
2. If English speakers do not know French, and/or cannot pronounce initial or final /ʒ/, then they may substitute the closest native English sound, which is /dʒ/. In this way, beige (a French loanword) and page (a native English word) rhyme.
3. Words may be more fully integrated into English phonology. This is the case with garage. By the first process described above, this is /gærɑ:ʒ/. By the second, it is /gærɑ:dʒ/. In AmE, French loanwords are typically pronounced with stress on the second syllable, e.g., ballet, AmE /bæˈleɪ/, BrE /'bæleɪ/. Similarly, garage typically has stress on the second syllable in AmE, but on the first in BrE. A possible BrE pronunciation is thus /ˈgærɑ:dʒ/. The weakening of unstressed syllables, very common in BrE, then changes the /ɑ:/ vowel to /ɪ/, giving /ˈgærɪdʒ/, which rhymes with marriage.

Notice, however, that the above processes depend on how recently the word was borrowed into English and whether it has been fully integrated, like garage. Other fully integrated French loanwords include mortgage and visage.

Open syllables

One-syllable words that have no final consonant sound fall into only two categories, in terms of the vowel:

• Long monophthong vowels: /i:/ (see /si:/), /ɑ:/ (shah /ʃɑ:/), /ɔ:/ (law /lɔ:/), /u:/ (shoe /ʃu:/) (and /ɜ:/ (fur /fɜ:/) in nonrhotic BrE).
• Diphthong vowels (which are also long): /eɪ/ (way /weɪ/), /aɪ/ (fly /flaɪ/), /ɔɪ/ (boy /bɔɪ/), /aʊ/ (cow /kaʊ/), /oʊ/ (go /goʊ/), (and /ɪə/ (here /hɪə/), /eə/ (there /ðeə/), /ʊə/ (pure /pjʊə/) in nonrhotic BrE).

Such syllables, without final consonants, that is, with empty codas, are termed open.

Final /ŋ/ sound

Words that contain a final /ŋ/ sound fall into only one category, in terms of the vowel preceding the /ŋ/. The vowel preceding the /ŋ/ is a short monophthong vowel: /ɪ/ (ring /rɪŋ/), /æ/ (hang /hæŋ/), /ʌ/ (tongue /tʌŋ/). Examples with the other short vowels are rarer, e.g., length /lɛŋθ/, kung fu /kʊŋ fu:/. BrE examples with /ɒ/ (long /lɒŋ/) continue this pattern, although they are pronounced with long /ɑ:/ in AmE.

In summary, only long vowels can occur in open syllables with no final consonant. Secondly, short vowels can only occur in closed syllables with a final consonant. This is true with /ŋ/; the only exceptions are examples where assimilation (see Chapter 9 on connected speech processes) has taken place, e.g., green card /gri:n kɑ:(r)d/ > /gri:ŋ kɑ:(r)d/. It is also true with other consonants, e.g., tip, bet, sack, bomb, good /tɪp, bɛt, sæk, bɒm, gʊd/ occur, but not /tɪ, bɛ, sæ, (bɒ,) gʊ/.

Consonant + /j/ initial clusters

There are plenty of examples of consonant + /j/ initial clusters: puma /pju:mə/, cute /kju:t/, future /fju:tʃər/, music /mju:zɪk/. The generalization here is that the vowel that follows a CC initial cluster, where /j/ is the second consonant, must be /u:/ (or /ʊə/, especially before /r/ in BrE, e.g., puerile /pjʊəraɪl/, cure /kjʊə/, furious /fjʊərɪəs/, mural /mjʊərəl/). In unstressed syllables, this can weaken to /ʊ/ or /ə/, e.g., regular /rɛgjʊlə, rɛgjələ/.

In AmE, /j/ does not occur in clusters after dental and alveolar sounds, e.g., enthusiasm, tune, news AmE /ɪnθu:zɪæzəm, tu:n, nu:z/, BrE /ɪnθju:zɪæzəm, tju:n, nju:z/.

Three- and four-consonant syllable-final clusters

Syllable-final clusters can contain up to three or even four consonants. However, the consonant phonemes that can function as the third or fourth consonant of such clusters are very limited. They are /t, d, s, z, θ/. The list is limited because very often these represent suffixes:

• Past tense verbs or participles, e.g., lapse > lapsed
• Plural nouns, e.g., lamp > lamps
• Third person singular present tense verbs, e.g., ask > asks
• Possessive nouns, e.g., student > student’s
• Contractions of is or has, e.g., The bank’s opening, The bank’s opened
• Ordinal numbers, e.g., twelve > twelfth
• Quality nouns, e.g., warm > warmth

In short, English is a language that makes extensive use of inflexions, derivations, and contractions, many of which contribute to the size of syllable-final clusters. Syllable structure is thus connected with grammar and morphology here.

It is worthwhile remembering that languages change over time and that the syllable structure rules in this paper are those of modern-day English speakers. For example, the spelling of gnat and knight contains vestigial g and k letters, because these words used to be pronounced /gn, kn/, as some people still do with gnu (and compare with German Knecht).

Number of syllables

Learners need to be able to say how many syllables multisyllable words contain. Several books (e.g., Gilbert 2001: 12–18) contain exercises in stating how many syllables words contain and in tapping out the syllables.

Separating the onset and rhyme

We have seen how the onset position functions somewhat independently of the rhyme (peak and coda positions). The phenomena of alliteration and rhyme relate to these two components respectively. Activities such as those in Vaughan-Rees (2010) can be used to reinforce these features. Questions such as the following can be asked: “Which word does not rhyme: spoon, book, tune?”, “Which word has a different first sound: chair, call, kick?”

In terms of literacy, the spelling of the onset is largely independent of the rhyme. Activities can be used that highlight this. For instance, learners can be asked to put the consonants /b, kl, d, g, h, dʒ, m, p, s, st, sl, θ/ before the rhyme /ʌmp/ or, in spelling terms, the letters b, cl, d, g, h, j, m, p, s, st, sl, th before ump.

Separating the peak and coda

Similar activities can be used for distinguishing the vowel in the peak from the consonant(s) in the coda. In short, the three positions and their constituent sounds can be worked on independently or in various combinations: “Here is a picture of a bell. Finish the word for me: /bɛ … /”, “Say spoon. Now say it again, but instead of /u:/ say /ɪ/”, “Say trip. Now say it without the /r/ sound”, “Say pink. Now say it again but do not say /p/.”

Dealing with clusters

Consonant clusters are a major problem, especially for learners from languages that do not permit clusters. Various (combinations of) strategies, similar to those described above for the integration of loanwords, may be resorted to by the learner. The following examples relate to the English word street /stri:t/.

• Final consonants may simply be omitted: /stri:/.
• Extra vowels may be added to final consonants: /stri:ti:/.
• Extra vowels may be inserted to break up initial clusters: /si:tri:t/.
• Extra vowels may be inserted before initial clusters: /i:stri:t/.

The fact should not be forgotten that words are not said in isolation, but in stretches that are linked together (see Chapter 9 on connected speech processes). Thus, if learners omit the final /k/ of link, they should be given practice in pronouncing the word with something following that begins with a vowel sound, e.g., link it, linking. The word division is largely irrelevant here; link it ends like (rhymes with) trinket.

Likewise, linked phrases can be used to combat the above learners’ strategies. In street address, the final /t/ is linked to the following vowel, avoiding deletion of the /t/ or insertion of a vowel. In this street, the fact that this ends in the same /s/ consonant as at the beginning of street allows the two /s/s to be joined, avoiding intrusive vowels.