Phonetics & Phonology

Foundations: phones, phonemes and evidence

Phonetics studies speech production, the transmitted signal and perception: articulatory, acoustic and auditory phonetics respectively. Phonology studies how a language organises sounds into contrasts and patterns. A phone is a speech sound; a phoneme is a contrastive unit in a particular language. A minimal pair differs in one phoneme in the same position and has different meanings: pin /pɪn/ and bin /bɪn/ establish the /p–b/ contrast. Positional variants of one phoneme are allophones (§2).

Keep physical measurements separate from their perceptual correlates:

These are relationships, not interchangeable labels: pitch is what a listener perceives, whereas F0 is a measurable property of a periodic signal. An accent concerns pronunciation; a dialect also includes grammar and vocabulary.

1.1 What separates a vowel from a consonant

• Phonetic definition: a vowel is produced with an open, unobstructed vocal tract (no audible friction or closure); a consonant involves a stricture (closure or narrowing) somewhere in the tract.
• Phonological role: vowels normally form the syllable nucleus; consonants normally occupy the onset/coda. Some consonants can themselves be syllabic, as [n̩] in button. The approximants /j w r/ are phonetically vowel-like but function as consonants; semivowel is used for /j w/.

1.2 Vowels — the classificatory parameters

Vowels are classified by:

1. Height / closeness of the tongue: close (high) – close-mid – open-mid – open (low).
2. Backness: front – central – back (which part of the tongue is highest).
3. Lip rounding: rounded vs unrounded (spread/neutral).
4. Length / quantity: long (ː) vs short. In this GB inventory, long and short vowel sets also differ in quality; actual duration depends on context (see pre-fortis clipping, §2.3).
5. Monophthong vs diphthong (steady vs gliding quality).

1.3 The 12 GB monophthongs

7 short vowels (schwa only in unstressed syllables):

5 long vowels:

Quality ≠ length alone. /iː/ vs /ɪ/ and /uː/ vs /ʊ/ differ in both quality (tongue position) and quantity. Beat /biːt/ is not just a longer bit /bɪt/ — the vowels sit in different places. Never describe the contrast as "just length."

1.4 The 8 GB diphthongs (classified by direction of glide)

A diphthong is a single vowel phoneme with a gliding quality: the tongue moves from a first (more prominent) element towards a second.

5 closing diphthongs (glide towards a close vowel):

3 centring diphthongs (glide towards /ə/):

Inventory convention: this guide uses the eight diphthongs listed above. Actual realisations and phoneme inventories vary by accent and speaker; use these symbols consistently when giving a GB transcription.

Triphthongs (5, supporting material): /eɪə/ player, /aɪə/ fire, /ɔɪə/ loyal, /əʊə/ lower, /aʊə/ power. In GB they are subject to smoothing (fire → [faə] ~ [faː]). Not part of the core 44.

1.5 Consonants — the classificatory parameters

Every consonant is specified by three things (the "VPM label"): Voicing – Place – Manner, e.g. /t/ = voiceless alveolar plosive, /z/ = voiced alveolar fricative.

Place of articulation (front → back):

Manner of articulation:

• Plosive (stop): complete closure + release — /p b t d k ɡ/.
• Affricate: closure released slowly into homorganic friction — /tʃ dʒ/.
• Fricative: narrowing producing turbulent friction — /f v θ ð s z ʃ ʒ h/.
• Nasal: oral closure, air escapes through the nose (velum lowered) — /m n ŋ/.
• Approximant: narrowing too wide for friction — /r j w/ (central) and /l/ (lateral approximant).

Voicing / fortis–lenis:

• Voiced means that the vocal folds vibrate; voiceless means that they do not. The labels fortis/lenis concern articulatory strength and the broader contrast, so partial devoicing need not turn a lenis phoneme into its fortis partner.
• Obstruents are plosives, affricates and fricatives. The paired fortis set is /p t k tʃ f θ s ʃ/ and the paired lenis set is /b d ɡ dʒ v ð z ʒ/. The voiceless fricative /h/ has no voiced partner in this inventory.
• Sonorant consonants are nasals and approximants /m n ŋ l r j w/. They are normally voiced, with contextual devoicing possible, and have no corresponding fortis series. Aspiration is a separate contextual rule for /p t k/ (§2.1).

1.6 The GB consonant chart (place × manner × voicing)

For paired obstruents: left = fortis/voiceless, right = lenis/voiced. Nasals and approximants are normally voiced; /h/ is unpaired and voiceless.

Notes: /h/ is voiceless glottal fricative (a "voiceless vowel" phonetically). /w/ is a labial-velar approximant (double place — shown in both columns). /r/ in GB is a post-alveolar approximant [ɹ], not a trill or tap. /ŋ/ is a single phoneme: sing /sɪŋ/ ends in /ŋ/, not a sequence /n/ + /ɡ/.

Example word for every consonant:

1.7 Common traps

• Calling /r/ a "trill" — the GB reference realisation is an approximant.
• Forgetting that /w/ has two places (labial-velar).
• Treating the /iː–ɪ/ and /uː–ʊ/ contrasts as purely quantitative.
• Mislabelling /θ ð/ as alveolar (they are dental).
• Counting /tʃ dʒ/ as two segments each — they are single affricate phonemes.

1.8 Exam-style questions

An allophone is a context-conditioned realisation of a phoneme. Contextual allophones are often described through complementary distribution (different environments); their difference does not itself establish a phonemic contrast in the language. The official topic list focuses on aspiration, pre-fortis clipping and devoicing; the rest is supporting.

2.1 Aspiration of /p t k/ — the headline allophonic rule

The fortis plosives /p t k/ are aspirated [pʰ tʰ kʰ] — released with a puff of air / a period of voicelessness before the following vowel's voicing begins (a long Voice Onset Time).

Where aspiration occurs (strong): at the beginning of a stressed syllable, especially before a vowel.

• pin [pʰɪn], tin [tʰɪn], kin [kʰɪn]; appear [əˈpʰɪə], attend [əˈtʰend], account [əˈkʰaʊnt].

Where aspiration is weak or absent:

1. After /s/ in a syllable-initial cluster — no aspiration: spin [spɪn], stay [steɪ], skin [skɪn]. The absence of aspiration does not change /p t k/ into the phonemes /b d ɡ/.
2. In an unstressed syllable — aspiration is weak: paper [ˈpʰeɪpə] (weak second /p/).
3. In syllable-final position — plosives are often unaspirated and may be unreleased [p̚ t̚ k̚] or glottally reinforced (§2.6).

Aspiration realised as devoicing of a following approximant: when /p t k/ are followed by /l r w j/, the "puff" devoices that approximant:

• play [pl̥eɪ], cry [kr̥aɪ], twin [tw̥ɪn], pure [pʰj̊ʊə], clean [kl̥iːn], quick [kw̥ɪk].
• But after /s/ the plosive is unaspirated, so the approximant stays voiced: splay [spleɪ], screw [skruː] — the /l r/ are not devoiced. (This neatly links aspiration and approximant-devoicing.)

Function of aspiration (frequent exam sub-question): aspiration is the principal perceptual cue that keeps the fortis–lenis contrast audible word-initially, because English lenis /b d ɡ/ are only weakly voiced there. Listeners hear pin vs bin mainly as aspirated vs unaspirated, not voiceless vs voiced.

2.2 Diagnostic: aspiration vs no aspiration

2.3 Pre-fortis clipping (pre-fortis shortening)

A vowel (or a sonorant) is shorter before a fortis consonant than before a lenis consonant in the same syllable. This is pre-fortis clipping.

• Minimal pairs (same vowel phoneme, different duration): seat /siːt/ (short [iˑ]) vs seed /siːd/ (full [iː]); right /raɪt/ vs ride /raɪd/; back /bæk/ vs bag /bæɡ/; cart /kɑːt/ vs card /kɑːd/; coat /kəʊt/ vs code /kəʊd/.
• Sonorants clip too: the /n/ in bent is shorter than in bend; /l/ in belt shorter than in bell.

Why it matters: final lenis obstruents may be substantially devoiced (§2.4), so the preceding vowel's relative duration is an important contrast cue. Both seat and seed contain the phoneme /iː/, but its realisation is shorter before fortis /t/. Clipping does not replace /iː/ with /ɪ/.

2.4 Devoicing of lenis obstruents

Lenis /b d ɡ dʒ v ð z ʒ/ are most fully voiced between voiced sounds. At utterance boundaries, especially finally, they may be partially or fully devoiced, marked with the voiceless diacritic ring:

• Final: bad [bæd̥], rose [rəʊz̥], give [ɡɪv̥], bridge [brɪdʒ̊].
• Initial: dog [d̥ɒɡ̊] — little voicing during closure.

Crucially, devoicing does not neutralise the contrast, because two other cues survive: (a) no aspiration on lenis (vs aspirated fortis) initially; (b) pre-fortis clipping of the preceding vowel finally. So bad stays distinct from bat via the longer vowel + weaker articulation, even though [d̥] is nearly voiceless.

Devoicing of approximants after fortis plosives (already met in §2.1): /l r w j/ → [l̥ r̥ w̥ j̊] after aspirated /p t k/. This is a progressive voicing effect.

2.5 Clear vs dark /l/

/l/ has two complementary allophones:

• Clear [l] before a vowel or /j/: leaf [liːf], light, value [ˈvæljuː].
• Dark / velarised [ɫ] before a consonant or a pause, and when syllabic: feel [fiːɫ], milk [mɪɫk], table [ˈteɪbɫ̩].

This is a textbook allophonic rule (complementary distribution). In Estuary/London speech dark /l/ is often vocalised to [o]/[ʊ] (milk → [mɪok]) — an accent feature, mention only as a variant.

2.6 Glottalisation and release types (supporting)

• Glottal reinforcement (pre-glottalisation): fortis /p t k tʃ/ are reinforced by a glottal stop [ʔ] when syllable-final before a consonant or pause: football [ˈfʊʔtbɔːl], not now, watch [wɒʔtʃ].
• Glottal replacement of /t/ (RP-adjacent/Estuary): bottle [ˈbɒʔl̩], button [ˈbʌʔn̩], Gatwick [ˈɡæʔwɪk].
• Nasal release: /t d/ + homorganic /n/ → release through the nose: button [ˈbʌtn̩], sudden [ˈsʌdn̩].
• Lateral release: /t d/ + /l/ → release over the sides of the tongue: little [ˈlɪtl̩], middle [ˈmɪdl̩].
• Nasalisation of vowels adjacent to nasals: man [mæ̃n].

2.7 Common traps

• Claiming /p t k/ are "always strongly aspirated" — aspiration is absent after /s/ and weak or absent in other contexts such as unstressed or final positions.
• Confusing devoicing with deletion: a devoiced [d̥] is still there; it has not vanished.
• Replacing a devoiced final lenis phoneme with a fortis one: [d̥] can still realise /d/. Preceding-vowel duration helps preserve the contrast.
• Forgetting that aspiration and approximant-devoicing are the same phenomenon (the plosive's VOT falling on the sonorant).

2.8 Exam-style questions

3.1 The nature of stress and the four factors of prominence

Stress is the relative prominence of a syllable. A stressed syllable is produced with more articulatory effort and stands out perceptually. Four factors (correlates) of prominence — the official sample asks you to name four:

1. Pitch (related to fundamental frequency) — especially a pitch movement/change on the syllable. Usually the strongest cue.
2. Loudness (intensity) — louder syllables are heard as more prominent.
3. Length (duration/quantity) — longer syllables sound more prominent.
4. Quality (vowel quality) — a full vowel can help a syllable stand out against surrounding reduced vowels; vowel reduction and stress are related but not identical.

The four prominence factors are pitch, loudness, length, quality. Use the perceptual terms when naming prominence factors and the acoustic terms when discussing measurements.

3.2 Levels of stress and their marks

• Primary stress ˈ (high vertical stroke before the syllable): ˈopen.
• Secondary stress ˌ (low stroke): ˌphotoˈgraphic, ˌacaˈdemic.
• Unstressed (no mark) — may contain a reduced vowel, a syllabic consonant, or an unreduced full vowel.

3.3 Vowel reduction and weak syllables

A syllable has an optional onset, a peak/nucleus, and an optional coda; rhyme/rime = peak + coda. Count sounds rather than letters: stone /stəʊn/ is CCVC, with /st/ as onset, /əʊ/ as one vowel peak and /n/ as coda. It is a closed syllable (coda present); go /ɡəʊ/ is open (no coda).

Phonotactics describes permitted sound distributions and combinations. English permits /str-/ initially (street) but not /ŋ-/; /ŋ/ occurs in codas (sing). Thus knowing the phoneme inventory is not enough to know which syllables are possible.

Distinguish strong syllables with full vowels from weak syllables, whose peaks include /ə/, weak /i u/, or syllabic consonants. [n̩ l̩] are common syllabic consonants in GB (button, little), with [m̩] also possible. Weak syllables are unstressed, but an unstressed syllable need not be weak or reduced: ˈphotograph retains a full /ɑː/ in its final syllable in this GB transcription. Weak /i u/ are useful symbols without a length mark, as in happy /ˈhæpi/ and weak to /tu/ before a vowel.

Compare photograph /ˈfəʊtəɡrɑːf/ → photographer /fəˈtɒɡrəfə/ → photographic /ˌfəʊtəˈɡræfɪk/. The family illustrates related changes in stress and vowel quality, but not a rule that every unstressed vowel becomes schwa. Rhythmic stress shift (§3.6) also does not mechanically replace the word's vowels.

3.4 Stress in complex words — three suffix classes

The behaviour of a suffix with respect to stress is a core exam point. Three classes:

(a) Stress-neutral (stress-preserving) suffixes — do not move the stress; the base keeps its stress. Mostly Germanic: -ness, -less, -ful, -ly, -er, -or, -ing, -ish, -y, -dom, -hood, -ship, -ment, -able (usually), -age.

• ˈhappy → ˈhappiness; ˈwonder → ˈwonderful; ˈgovern → ˈgovernment; ˈcoin → ˈcoinage.

(b) Stress-fixing / auto-stressed (stress-carrying) suffixes — the suffix itself takes the primary stress. Mostly of French/Romance origin: -ee, -eer, -ese, -ette, -esque, -aire, -oon, -ade.

• employˈee /ˌemplɔɪˈiː/, refuˈgee /ˌrefjuˈdʒiː/, engiˈneer, Japaˈnese /ˌdʒæpəˈniːz/, cigaˈrette /ˌsɪɡəˈret/, pictuˈresque, balˈloon.

(c) Stress-shifting / pre-accenting (stress-fixing on the stem) — the stress moves to the syllable immediately before the suffix: -ic, -ical, -ity, -ion/-tion/-sion, -ian, -ial, -ious/-eous, -graphy, -logy, -metry.

• eˈconomy → ecoˈnomic; ˈphotograph → ˌphotoˈgraphic; proˈductive → ˌproducˈtivity; ˈpolitics → ˌpoliˈtician. These are useful patterns with lexical exceptions; identify the actual suffix and check the word. The -al in ˈnation → ˈnational preserves the stress.

Worked example (the sample-test words):

So the sample-test answer is: stress-shifting suffixes = -ic, -ity, -ion; neutral suffixes = -age, -ly.

3.5 Compound stress: initial-stress compounds vs phrases

• Many compound nouns have primary stress on the first element (fore-stress): N+N ˈteacup, ˈsheepdog; Adj+N ˈblackbird, ˈgreenhouse, ˈhotdog. The word classes of the parts and the stress pattern are separate facts.
• Adjective + noun phrases normally have final prominence in a neutral reading: a black ˈbird, a green ˈhouse, a hot ˈdog. Contrastive focus can change that prominence.

Minimal pairs turning on stress placement:

Later stress is also an established pattern, especially in compound adjectives: ˌwell-ˈknown, ˌold-ˈfashioned, ˌgood-ˈlooking, ˌself-ˈconfident. Compare compound adverbs ˌdownˈstairs and ˌhead-ˈfirst. Learn these alongside early-stressed compound nouns.

Open combinations such as apple ˈpie and place names such as Park ˈLane also have later stress; ˈOxford Street illustrates the usual Street pattern. Stress is evidence for an analysis, not an exceptionless test of compoundhood. In a neutral reading, the contrast ˈblackbird versus black ˈbird is useful; in discourse, context and focus also matter.

Word class can change lexical stress. Compare noun ˈrecord with verb reˈcord, and noun ˈpermit with verb perˈmit. This is a lexical distinction between related words, not the contextual rhythmic stress shift below. It is a useful pattern for some two-syllable noun–verb pairs, not a rule for every English noun or verb.

3.6 Stress shift in connected speech (rhythmic reversal)

Some words with secondary + primary stress can redistribute prominence when followed by another prominent word. Earlier prominence is retained and the later word takes the main phrase stress. For example:

• ˌJapaˈnese → ˌJapanese ˈpeople
• ˌbad-ˈtempered → ˌbad-tempered ˈteacher
• ˌimpreˈcise → ˌimprecise ˈanswer

This is a contextual rhythmic pattern, not a compulsory change in every word with two stresses. The earlier mark in these phrase examples is secondary; the primary mark belongs to the following word. The citation form keeps its later primary stress, and rhythmic redistribution need not change vowel quality.

3.7 Common traps

• Writing ˈproductivity — no; -ity shifts stress to -tiv-: /ˌprɒdʌkˈtɪvɪti/.
• Assuming -ee behaves like other suffixes — it is auto-stressed (employˈee).
• Marking a compound like a phrase (or vice versa) — the stress placement is the answer.
• Missing the earlier prominence within the first word when the following word takes the main phrase stress.
• Naming only three prominence factors — the question wants four (pitch, loudness, length, quality).

3.8 Exam-style questions

4.1 Stress-timing

English is traditionally described as stress-timed: stressed syllables recur at roughly equal intervals (a tendency towards isochrony), regardless of how many unstressed syllables fall between them; the intervening weak syllables are compressed to keep the beat. The rhythmic unit is the foot — a stressed syllable plus all following unstressed syllables up to the next stress.

Contrast syllable-timed languages (French, Spanish, Czech is somewhere between), where syllables take roughly equal time.

Honest caveat (good to state): acoustic measurement does not confirm strict isochrony; stress-timing is best treated as a perceptual/pedagogical tendency, not a metronomic fact. But the compression of weak syllables — via weak forms and vowel reduction — is real and is what the exam tests.

4.2 Content words vs function words

Sentence stress falls on content words; function words are the natural sites of weak forms.

4.3 Strong and weak forms of grammatical words

Many function words have strong and weak pronunciations, sometimes with several variants: a strong form (stressed, full vowel) and a weak form (unstressed, reduced — usually to /ə/, or with elided consonants). The weak form is the default in connected speech.

Weak/strong form table (GB conventions). C = consonant sound; V = vowel sound. Forms ending in /ə ɜː ɔː/ may have linking /r/ before a vowel where the spelling has r (§6.4).

Articles & determiners

Conjunctions

Prepositions

Pronouns & possessives

Auxiliary & modal verbs, forms of BE

Existential/adverbial

Note: some prepositions (e.g. in, on, up) have no distinct weak form in this GB table; they simply stay /ɪn ɒn ʌp/.

4.4 When strong forms are used

The choice depends on grammatical role, focus and context, not only position:

1. Stranded prepositions and elliptical auxiliaries normally use a strong form: "What are you looking at?" /æt/; "I know you are." /ɑː/; "Where do you come from?" /frɒm/.
2. Contrast or emphasis: "The letter's from him, not to him." /frɒm/ … /tuː/.
3. Citation: "Don't begin a sentence with 'and'." /ænd/.
4. Contrasting coordinated terms: "for and against", "to and fro".

Final position does not make every grammatical word strong. Final weak pronouns are possible: "Don't look at me" /mi/; "I like them" /ðəm/. Likewise him can be weak /ɪm/ in "I can see him" if it is unstressed, or strong /hɪm/ when contrasted with someone else.

Role matters: demonstrative that uses /ðæt/; conjunction/relative that may use /ðət/. Relative who may be /hu/, while interrogative who uses /huː/. This guide distinguishes weak must for obligation (You must go: /məs/ before /ɡ/) from strong /mʌst/ for deduction (He must be ill).

4.5 Common traps

• Giving the strong form by default — the weak form is normal.
• Extending the strong-form rule for stranded prepositions/elliptical auxiliaries to all final pronouns: final me, him, them can remain weak.
• Reducing demonstrative that — conjunction/relative that can weaken.
• Reducing some when it means "certain/approximately" — that is the strong /sʌm/.
• Treating h-dropping as compulsory: weak pronouns/auxiliaries can lose /h/ medially; utterance-initial /h/ is normally retained in this GB model.

4.6 Exam-style questions

Intonation is the linguistically significant use of pitch patterns. Its domain is the tone unit, which can be as short as one word or extend over a clause or more. Unlike a lexical tone contrast, the choice here concerns how an utterance is organised and interpreted.

5.1 Functions of intonation

Intonation has four main functions:

1. Attitudinal — expressing emotions and attitudes: finality vs openness, politeness, surprise, boredom, sarcasm, reservation. (E.g. a fall sounds definite; a fall-rise sounds reserved — "yes, but…".)
2. Grammatical — signalling grammatical structure and sentence type: statement vs question ("You're leaving↘" vs "You're leaving↗"), wh- vs yes/no questions, marking clause boundaries, distinguishing defining from non-defining relative clauses (the latter set off in their own tone unit).
3. Accentual / focusing (highlighting) — the placement of the nucleus marks the most important word and the focus (new vs given information): "I bought a RED car" vs "I bought a red car."
4. Discourse / cohesive (interactional) — organising conversation: turn-taking, signalling given/new, topic structure, presenting information as new or shared, and helping signal completion or continuation in context.

An intonation analysis involves three connected decisions:

Analyse these together: a change of boundary can change the available nucleus and the interpretation of the tone.

5.2 The tone unit (intonation phrase) and its structure

A tone unit carries one intonation pattern. Its maximal structure (Roach):

(Pre-head) (Head) — Nucleus — (Tail)

• Nucleus (tonic syllable): the most prominent syllable, bearing the main pitch movement. Obligatory — every tone unit has exactly one.
• Head: from the first stressed syllable (the onset) up to, but not including, the nucleus.
• Pre-head: any unstressed syllables before the head.
• Tail: any syllables after the nucleus (they continue the nuclear tone).

Boundaries are marked | (minor) and ‖ (major). Example: | the ˈonly ˈthing I ˈwant is ˈREST ‖ — pre-head the, head ˈonly ˈthing I ˈwant is, nucleus REST, no tail.

5.3 The five nuclear tones and their typical meanings

Illustrations:

• "It's ˈtrue↘" — plain assertion (fall).
• "Are you ˈready↗" — yes/no question (rise).
• "I ˈlike it↘↗ (…but)" — reservation (fall-rise).
• "You ˈwon↗↘!" — impressed/surprised (rise-fall).
• "Well↔…" — hesitation (level).

Default tone by clause type (useful heuristics): statements, wh-questions, commands → fall; yes/no questions → rise; continuation → often rise; reservation/implication → often fall-rise. These are contextual tendencies, not fixed tone-to-meaning equations. A wh-question can rise to encourage a response, and a yes/no question can fall when insistent.

Question tags: "You know her, don't you↗?" genuinely checks; "You know her, don't you↘?" invites confirmation of an assumption. State both the tone and the speaker's stance.

5.4 Nucleus placement = focus (the accentual function in practice)

The nucleus normally falls on the last lexical (content) word of the tone unit (neutral/"broad" focus). Moving it earlier creates contrastive/narrow focus:

• "I said a GREEN car" (not a red one) — nucleus on green.
• "I said a green car" (not you) — nucleus on I.

Given information can be de-accented; contrast can put prominence on a word even when it has already been mentioned. Identify the discourse context before assigning focus.

Discourse function: a proclaiming fall typically presents information as new; referring rise/fall-rise presents it as shared or recoverable. A parenthetical insertion may be intonationally subordinate: lower pitch, narrower pitch range and faster delivery. In a context such as "What I need is a holiday", the given what I need can form a referring unit and new a holiday a proclaiming unit. These choices depend on what speaker and listener already share.

5.5 Common traps

• Saying a tone unit can have several nuclei — it has exactly one.
• Treating the usual rising yes/no pattern as compulsory: a fall can convey insistence or a different stance.
• Treating fall-rise as a plain rise, or assuming it always means the same thing: reservation and reference to shared information are possible interpretations.
• Forgetting that nucleus placement (not just tone choice) carries meaning (focus).

5.6 Exam-style questions

In connected speech, sounds can influence adjacent sounds within and across words. Four processes: assimilation, coalescence/assibilation, elision, linking.

6.1 Assimilation

Assimilation = a sound becomes more like an adjacent sound. Classify along two axes.

By direction:

• Regressive (anticipatory) — the following sound changes the preceding one (change points backwards: C ← C). The commonest type in English.
• Progressive (perseverative) — the preceding sound changes the following one (C → C). Rarer.
• Coalescent (reciprocal) — the two sounds fuse into a third (see §6.2).

By feature affected:

• Assimilation of place — the dominant English type. Word-final alveolars /t d n/ (and /s z/) can adjust towards the next consonant:

• Regressive assimilation of voice: devoicing in short grammatical words occurs, e.g. have to /hæv tə/ → /hæf tə/ and of course /əv ˈkɔːs/ → /əf ˈkɔːs/. The following /t/ or /k/ affects the preceding consonant. These are regressive, not progressive.
• Progressive assimilation of voice: in the analysis used here for regular -(e)s and -ed, the preceding stem-final sound conditions the ending: cats /kæts/ versus dogs /dɒɡz/; worked /wɜːkt/ versus played /pleɪd/ (§7.2–3).
• Assimilation of manner: regressive good night /ɡʊd naɪt/ → /ɡʊn naɪt/; progressive in the /ɪn ðə/ → /ɪn nə/ or all their /ɔːl ðeə/ → /ɔːl leə/ in fluent informal speech.

Boundary limit: do not generalise the grammatical-word rule to every lexical word, or import Czech-style voicing changes. Contrast have to with leave paper (keep /v/) and back door (keep /k/, not /ɡ/). This does not exclude partial phonetic devoicing [v̥]; distinguish that from replacing /v/ with fortis /f/ (§2.4).

Canonical exam case: /n/ → [m] before a bilabial — ten pounds → [tem ˈpaʊndz], in Berlin, green paint. Regressive, of place.

6.2 Coalescent assimilation / assibilation (yod coalescence)

When /t d s z/ meet a following /j/ (typically the /j/ of you/your, or word-internal /juː/), they coalesce into a palato-alveolar — a coalescent (reciprocal) assimilation, traditionally called assibilation or yod coalescence:

6.3 Elision

Elision is the loss of a sound present in a fuller pronunciation. Rate, style and consonant context matter; the examples below are possible realisations.

• /t d/ in consonant clusters: last night → /lɑːs ˈnaɪt/, next day → /neks ˈdeɪ/, hand made → /hæn ˈmeɪd/, friendship → /ˈfrenʃɪp/. The alveolar plosive is between consonants, but not every C–/t d/–C sequence requires deletion. The GB pattern used here retains it before /h/ (left hand, wild horse); before /r w l/ it allows retention or deletion (last week). Before a vowel, keep the consonant and link (last apple).
• Weak /ə/ elision: police /pˈliːs/; compression in family /ˈfæmli/ and camera /ˈkæmrə/. These compressed forms do not automatically contain syllabic consonants. A genuine syllabic-consonant example is button [ˈbʌtn̩], where [n̩] forms a syllable peak.
• /h/ elision in weak pronouns/auxiliaries medially: tell him /ˈtel ɪm/, give her /ˈɡɪv ə/, does he /ˈdʌz i/.
• /v/ elision in weak of before a consonant: lots of them /ˈlɒts ə ðəm/, waste of time /ˈweɪst ə ˈtaɪm/.

Processes can combine: handbag /ˈhændbæɡ/ → /ˈhænbæɡ/ (elision) → /ˈhæmbæɡ/ (regressive place assimilation). Explain each step rather than calling every change assimilation.

6.4 Linking

Linking connects neighbouring words smoothly. It can join an existing final consonant to a following vowel, or involve /r/ and transitional glides between vowels.

• Consonant-to-vowel linking (liaison): connect the final consonant smoothly to the following vowel: an apple /ən‿ˈæpl̩/, get up /ɡet‿ˈʌp/, run off /rʌn‿ˈɒf/. This is pseudo-resyllabification: the connected articulation does not erase the lexical word boundary.
• Linking /r/: in non-rhotic GB, a word-final orthographic ⟨r⟩ is silent before a pause/consonant but pronounced before a vowel: far away /fɑːr‿əˈweɪ/, four eggs /fɔːr‿ˈeɡz/, mother and father /ˈmʌðər‿ən ˈfɑːðə/, here and there /hɪər‿ən ˈðeə/.
• Intrusive /r/: an /r/ inserted where there is no ⟨r⟩ in spelling, after a word ending in /ə ɑː ɔː/ before a vowel (by analogy with linking /r/): law(r) and order /ˈlɔːr‿ən ˈɔːdə/, idea(r) of /aɪˈdɪər‿əv/, media(r) attention, vanilla(r) ice cream, draw(r)ing. It is a pronunciation variant; identify it when it occurs rather than inserting it into every possible context.
• Linking glides /j/ and /w/: a smooth transition may bridge two vowels; it need not be a separate phoneme in a broad transcription:
  • after /iː i/ or a diphthong ending towards /ɪ/ — /eɪ aɪ ɔɪ/ → linking /j/: I agree /aɪ‿j‿əˈɡriː/, the end /ði‿j‿ˈend/, say it /seɪ‿j‿ɪt/.
  • after /uː u/ or a diphthong ending towards /ʊ/ — /əʊ aʊ/ → linking /w/: go out /ɡəʊ‿w‿ˈaʊt/, you are /ju‿w‿ɑː/, two apples /tuː‿w‿ˈæpl̩z/, how old /haʊ‿w‿ˈəʊld/.

6.5 Common traps

• Confusing the three: assimilation = a sound changes; elision = a sound is lost; linking/liaison = a sound is kept/added to bridge words.
• Calling English assimilation "progressive" by default — place assimilation is overwhelmingly regressive.
• Confusing direction and feature: have to is regressive devoicing, while the suffix in cats illustrates progressive voice conditioning. Do not voice /k/ to /ɡ/ in back door.
• Inserting every possible linking glide or intrusive /r/ mechanically: transcribe the pronunciation being represented.
• Eliding /t d/ before a vowel — there they are retained and link.

6.6 Exam-style questions

These tasks occur in the published sample. They are useful practice, not a guarantee of the contents of every sitting.

7.1 The definite article: /ðə/ vs /ði/ (vs strong /ðiː/)

The form of the depends on the following sound, not the spelling:

Trap — sound, not letter: the university takes /ðə/ (the word begins with the consonant /j/), while the honour takes /ði/ (the ⟨h⟩ is silent, so it begins with a vowel). This is exactly the sample-test point.

Sample-test answer (the other hand vs the lexicon): the other → /ði/ because other begins with the vowel /ʌ/; the lexicon → /ðə/ because lexicon begins with the consonant /l/.

7.2 Regular -s allomorphs (plural / 3rd-person sg / genitive): /s/, /z/, /ɪz/

Apply in order:

1. /ɪz/ after a sibilant /s z ʃ ʒ tʃ dʒ/ (a vowel separates the stem-final sibilant from the suffix; this is an ending-specific pattern, not a ban on all adjacent sibilants): buses /ˈbʌsɪz/, roses, wishes, garages, churches, judges, processes /ˈprəʊsesɪz/.
2. /s/ after any other fortis (voiceless) consonant /p t k f θ/: cats /kæts/, books /bʊks/, states /steɪts/, months, cliffs.
3. /z/ elsewhere — after any voiced sound (vowels, non-sibilant lenis obstruents and sonorant consonants): dogs /dɒɡz/, forms /fɔːmz/, ways /weɪz/, factors /ˈfæktəz/, frogs /frɒɡz/, boys /bɔɪz/.

This is progressive voicing assimilation (the suffix agrees in voicing with the preceding sound), plus epenthesis in the sibilant case.

Regular plural pronunciation rule, one-sentence version (for the exam): "The regular plural is /ɪz/ after sibilants, /s/ after other voiceless consonants, and /z/ after all other (voiced) sounds."

Sample-test set (forms, ways, processes, factors, states, frogs):

7.3 Regular past-tense / -ed allomorphs: /t/, /d/, /ɪd/

Parallel logic:

1. /ɪd/ after alveolar plosives /t d/ (epenthesis): wanted /ˈwɒntɪd/, needed /ˈniːdɪd/, decided.
2. /t/ after any other fortis consonant /p k f θ s ʃ tʃ/: stopped /stɒpt/, worked /wɜːkt/, laughed /lɑːft/, washed /wɒʃt/, watched /wɒtʃt/.
3. /d/ elsewhere — after any voiced sound: played /pleɪd/, called /kɔːld/, loved /lʌvd/, opened /ˈəʊpənd/, robbed /rɒbd/.

7.4 Transcribing a sentence into tone units (worked model)

The official sample asks for an integrated transcription. Method:

1. Divide into tone units: choose meaningful groups, mark | / ‖, and allow for speech rate and focus.
2. Choose weak forms for unstressed function words, accounting for their role and the following sound.
3. Mark stressed syllables with ˈ and identify the nucleus of each unit.
4. Show linking with ‿ where words are joined. Add allophonic detail in [ ] when requested.

Worked example — the sample-test bold sentence (lines 6–8): "On the other hand, the lexicon of English contains many words whose precise form or meaning could not be predicted on the basis of the current state of the language."

One possible two-unit reading follows. The second unit is long; line breaks inside it only wrap the transcription, while | marks an actual boundary. More units are possible with a slower or differently focused reading.

| ɒn ði ˈʌðə ˈhænd↘↗ |
| ðə ˈleksɪkən‿əv‿ˈɪŋɡlɪʃ kənˈteɪnz ˈmeni ˈwɜːdz
  huːz prɪˈsaɪs ˈfɔːm‿ə ˈmiːnɪŋ kəd ˈnɒt bi prɪˈdɪktɪd‿ɒn
  ðə ˈbeɪsɪs‿əv ðə ˈkʌrənt ˈsteɪt‿əv ðə ˈlæŋɡwɪdʒ↘ ‖

Explain the choices:

• Weak forms: the = /ði/ before the vowel of other but /ðə/ before consonants; of = /əv/ (three times); could = /kəd/; be = /bi/; or = /ə/.
• Linking: lexicon‿of, of‿English, form‿or, predicted‿on, basis‿of and state‿of illustrate consonant-to-vowel linking. A transitional [j] is also possible in the other [ði j ˈʌðə].
• Nuclei and tones: hand bears a fall-rise connecting the contrasting preface to what follows. The final unit has its nucleus on lan- in language; the fall signals completion. The arrow after the word represents the contour beginning at that nuclear syllable and continuing through the tail.
• Allophony, if requested separately: strong aspiration on /t/ in contains [kənˈtʰeɪnz] and /k/ in current [ˈkʰʌrənt]; pre-fortis clipping in state and precise. Weak could /kəd/ is not a suitable example of strong stressed-onset aspiration in this reading.

This is a model application, not an official answer key. Tone-unit boundaries and focus are choices to justify; the printed sentence alone does not specify a unique spoken performance.

7.5 Exam-style questions

Foundations and GB inventory (44 phonemes)

• Phonetics = production/signal/perception; phonology = language-specific contrasts and patterns. Minimal pairs establish phonemic contrasts.
• F0/intensity/duration/formants are acoustic measures; pitch/loudness/length/quality are perceptual correlates.
• 12 monophthongs: short /ɪ e æ ʌ ɒ ʊ ə/ + long /iː ɑː ɔː uː ɜː/. Classify by height, backness, rounding, length.
• 8 diphthongs: closing /eɪ aɪ ɔɪ əʊ aʊ/ + centring /ɪə eə ʊə/. Classify by direction of glide.
• 24 consonants: VPM label each. Paired fortis obstruents /p t k tʃ f θ s ʃ/ vs lenis /b d ɡ dʒ v ð z ʒ/; unpaired /h/; sonorants /m n ŋ l r j w/.

Allophony

• /p t k/ aspirated at stressed-syllable onset; unaspirated after /s/, weak when unstressed/final. Aspiration → devoiced /l r w j/ (play [pl̥eɪ]).
• Pre-fortis clipping: vowel shorter before fortis (seat < seed); main length cue finally.
• Devoicing of final lenis (bad [bæd̥]); contrast kept by clipping + lack of aspiration.
• Clear [l] (pre-vocalic) vs dark [ɫ] (pre-consonant/pause).

Word stress

• Four prominence factors: pitch, loudness, length, quality.
• Suffix classes: neutral (-ness, -ful, -ly, -age…) / auto-stressed (-ee, -eer, -ese, -ette, -esque) / stress-shifting/pre-accenting (-ic, -ity, -ion, -ian, -ial…).
• Typical noun compound = fore-stress (ˈblackbird); phrase = end-stress (black ˈbird), with lexical and contextual exceptions.
• Stress shift redistributes prominence in context: ˌJapaˈnese → ˌJapanese ˈpeople (prominence notation).
• Syllable = onset + peak + coda; rhyme = peak + coda. Stone is CCVC. Unstressed does not automatically mean schwa.

Rhythm & weak forms

• English is stress-timed (tendency); feet run stress-to-stress.
• Content words stressed; function words → weak forms (default). Learn the table.
• Strong forms: stranded prepositions/elliptical auxiliaries, contrast, citation. Final pronouns can remain weak. Check role: demonstrative that is strong; use strong must for deduction.

Intonation

• Functions: attitudinal, grammatical, accentual/focusing, discourse.
• Tonality = grouping; tonicity = nucleus placement; tone = pitch pattern.
• Tone unit = (pre-head)(head) nucleus (tail); one nucleus per unit.
• Typical uses: fall completion/new information; rise checking/continuation; fall-rise reservation/shared information; rise-fall strong feeling; level routine. Interpret in context. Rising tag checks; falling tag assumes agreement.

Connected speech

• Assimilation (change): mostly regressive, of place — /n/→[m] before bilabials, /t/→[k] before velars, /s/→[ʃ] before /ʃ,j/.
• Assibilation / yod coalescence (coalescent): /t d s z/ + /j/ → /tʃ dʒ ʃ ʒ/ (did you /ˈdɪdʒu/).
• Voice assimilation: regressive grammatical-word devoicing (have to) vs progressive ending choice (cats/dogs). Keep /k/ in back door.
• Elision (loss): /t d/ can be lost in clusters; retain before /h/ in the pattern used here. Distinguish schwa deletion/compression from a syllabic consonant.
• Linking (add/keep): C→V liaison; linking /r/ (spelt) vs intrusive /r/ (unspelt); linking /j/, /w/ glides.

Cross-cutting rules

• the = /ði/ before vowel sound, /ðə/ before consonant sound (sound not spelling); strong /ðiː/ for emphasis.
• -s allomorphs: /ɪz/ after sibilants, /s/ after voiceless, /z/ after voiced.
• -ed allomorphs: /ɪd/ after /t d/, /t/ after voiceless, /d/ after voiced.
• Tone-unit transcription: divide |, choose weak forms, mark stress + nucleus, mark linking ‿.