The H-cluster reductions are various consonant reductions that have occurred in the history of English, involving consonant clusters beginning with /h/ that have lost the /h/ (or become reduced to /h/) in some or all dialects.
The cluster /hw/ (spelled ?wh? since Middle English) has been subject to two kinds of reduction:
The Old English consonant clusters /hl/, /hr/ and /hn/ were reduced to /l/, /r/, and /n/ in Middle English. For example, Old English hl?f, hring and hnutu become loaf, ring and nut in Modern English.
In some dialects of English the cluster /hj/ is reduced to /j/, leading to pronunciations like /ju:d?/ for huge and /'ju:m?n/ for human, and making hew a homophone of ewe or you. This is sometimes considered a type of glide cluster reduction, but it is much less widespread than wh-reduction, and is generally stigmatized where it is found. Aside from accents with general H-dropping, in the United States this reduction is mostly found in accents of Philadelphia and New York City; it also occurs in Cork accents of Irish English. In other dialects of English hew and yew remain distinct; however, the cluster /hj/ of hew, human, etc. is often reduced from [çj] to just [ç] (a voiceless palatal fricative).
Y-cluster reductions are reductions of clusters ending with the palatal approximant /j/, which is the sound of ⟨y⟩ in yes, and is sometimes referred to as "yod", from the Hebrew letter yod(h), which has the sound [j]. Many such clusters arose in dialects in which the rising diphthong /?u/ (the product of the merger of several Middle English vowel sequences) became the falling diphthong /ju:/. (For more information, see Phonological history of English high back vowels.) They were thus often found before the vowel /u:/, as in cube /kju:b/ - which was in some cases modified to // or /?/ before (historical) /r/, as in cure, or weakened to /?/ or /?/ as in argument. They also occurred in words ending in -ion and -ious, such as nation and precious.
This change from /?u/ to /ju:/, which had occurred in London by the end of the 17th century, did not take place in all dialects. A few dialects, notably in Wales, as well as in some parts of northern England, New England and the American South, still retain a (falling) /?u/ diphthong where standard English has /ju:/ - these dialects therefore lack the clusters with /j/ and have not been subject to the reductions described here.
The diphthongs /ju:/ or // are most commonly indicated by the spellings eu, ew, uCV (where C is any consonant and V is any vowel), ue and ui, as in feud, few, mute, cue and suit, while the historical monophthong /u:/ is commonly indicated by the spellings oo and ou, as in moon and soup.
Yod-dropping is the elision of the /j/ from certain syllable-initial clusters of the type described above. Particular cases of yod-dropping may affect all or some of the dialects which have the relevant clusters.
The change of [?] to [j] in these positions (as described above) produced some clusters which would have been difficult or impossible to pronounce; this led to what John Wells calls Early Yod Dropping, in which the [j] was elided in the following environments:
Problems playing this file? See media help.
The previously mentioned accents that did not have the [?]->[j] change were not subject to this process. Thus, for example, in much Welsh English pairs like chews/choose, yew/you and threw/through remain distinct: the first member of each pair has the diphthong // while the second member has /u:/:
Conversely an initial /j/ does not appear in Welsh English before /i:/ in words such as yeast and yield.
Many varieties of English have extended yod-dropping to the following environments, on condition that the /j/ be in the same syllable as the preceding consonant:
Yod-dropping in the above environments was formerly considered nonstandard in England, but is today also heard among well-educated RP speakers. (The /j/ after /s/ is not normally dropped in RP in medial positions, however - compare pursuit /p?'sju:t/.) In General American yod-dropping is found not only in the above environments but also:
Glide retention in these contexts has occasionally been held to be a shibboleth distinguishing Canadians from Americans. However, in a survey conducted in the Golden Horseshoe area of Southern Ontario in 1994, over 80% of respondents under the age of 40 pronounced student and news without yod.
General American thus undergoes yod-dropping after all alveolar consonants. Some accents of American English, however, preserve the distinction in pairs like loot/lute and do/dew, because (as in the Welsh English dialects discussed above) they retain a diphthong // in words where RP has /ju:/, thus /lut~lt/, /du~d/, etc.
However, in words like annual, menu, volume, Matthew, continue, etc., where there is a syllable break before the /j/, there is no yod-dropping. The same applies accordingly to British and other accents - while the yod is often dropped after initial /l/, for example, it is not dropped in words like volume or value. (British speakers do omit the /j/ in figure, however, although many Americans retain it.)
Some East Anglian accents such as Norfolk dialect extend yod-dropping not only to the position after /t/, /d/ or /n/, but to the position after nonalveolar consonants as well, so that pairs like beauty/booty, mute/moot, cute/coot can be homophonous. A well-known series of UK television advertisements beginning in the 1980s featured Bernard Matthews describing his turkeys as "bootiful" (for beautiful). Such accents only pronounce a /j/ in words like "use", "unit" etc. where there is no consonant before the /j/.
|Luke||look||'lu:k||With foot-goose merger.|
|suit||soot||'su:t||With foot-goose merger.|
|fuel||fool||'fu:l||With vile-vial merger.|
Yod-coalescence is a process that palatalizes the clusters [dj], [tj], [sj] and [zj] into [d?], [t?], [?] and [?] respectively (for the meanings of the symbols, see English phonology). The first two of these are examples of affrication.
Unlike yod-dropping, yod-coalescence frequently occurs with clusters which would be considered to span a syllable boundary. It thus commonly occurs before unstressed syllables. For example, in educate, the /dj/ cluster would not usually be subject to yod-dropping (in General American, say), as the /d/ is assigned to the previous syllable, but it commonly coalesces to [d?]. Below are a few examples of universal yod-coalescence.
In some other words the coalesced pronunciation is common but not universal in the standard accents, for example:
Coalescence can even occur across word boundaries, as in the colloquial "gotcha" (for got you) and "whatcha" (for what you).
In some accents, yod-coalescence also occurs in stressed syllables, as in tune and dune. This happens in Australian, Cockney, Estuary English, Hiberno-English (some speakers), Newfoundland English, South African English, and to a certain extent in New Zealand English, Scottish English (many speakers), and even some varieties of English in Asia, like Philippine English (many speakers, due to influence by the phonology of their mother languages). It results in pronunciations such as these:
In certain varieties such as Australian and Ugandan English, stressed /sj, zj/ can also coalesce:
This can lead to additional homophony; for instance, dew and due come to be pronounced the same as Jew.
Yod-coalescence has traditionally been resisted in RP. While it has certainly become established in words of the first group listed above (nature, soldier, pressure, etc.), it is not yet universal in those of the second group (educate, etc.), and does not generally occur in those of the third group.
|duty||Judy||'d?u:?i||With intervocalic alveolar flapping.|
Old and Middle English had an initial /wr/ cluster, hence the spelling of words like write and wrong. This was reduced to just /r/, apparently during the 17th century. An intermediate stage may have been an [r] with lip rounding.
As a result of this reduction, pairs of words like rap and wrap, rite and write, etc. are homophones in practically all varieties of Modern English. They remain distinct in the Doric dialect of Scots, where the wr- cluster is pronounced /vr/.
Old English also had a cluster /wl/, which reduced to /l/ during Middle English. For example, the word lisp derives from Old English wlisp(ian).
Middle English initial /kn/ is reduced in modern English to /n/, making pairs like knot/not and knight/night homophones.
The /kn/ cluster was spelled cn- in Old English; this changed to kn- in Middle English, and this spelling survives in Modern English, despite the loss of the /k/ sound. Cognates in other Germanic languages usually still sound the initial /k/. For example, the Old English ancestor of knee was cn?o, pronounced /kne:o?/, and the cognate word in Modern German is Knie, pronounced /kni:/.
Most dialects of English reduced the initial cluster /kn/ to /n/ relatively recently; the change seems to have taken place in educated English during the seventeenth century. The cluster is preserved in some Scots dialects. 
The Middle English initial cluster /?n/ is reduced to /n/ in Modern English. Like the reduction of /kn/, this seems to have taken place during the seventeenth century. The change affected words like gnat, gnostic, gnome, etc., the spelling with gn- being retained despite the loss of the /?/ sound. The cluster is preserved in some Scots dialects. 
The song The Gnu jokes about this silent g and other silent letters in English. In fact the g in gnu may always have been silent in English, since this loanword did not enter the language until the late 18th century. The trumpeter Kenny Wheeler wrote a composition titled Gnu High, a pun on "new high".
In some types of Caribbean English, the initial clusters /sp/, /st/, and /sk/ are reduced by the loss of /s/. The following stop is then subject to regular aspiration (or devoicing of the following approximant) in its new word-initial environment. Some examples of such pronunciations are:
|spit||-> 'pit||['sp?t]||-> ['pt]|
|stomach||-> 'tomach||['st?m?k]||-> ['tm?k]|
|spend||-> 'pen||['sp?nd]||-> ['pn] (also affected by final cluster reduction)|
|squeeze||-> 'queeze||['skwi:z]||-> ['kw?i:z]|
NG-coalescence is a historical sound change by which the final cluster /n?/, pronounced  (the /n/ being realized as a velar nasal by assimilation with the velar /?/), came to be pronounced as just [?] - that is, the final [?] was dropped, but the velar quality of the nasal remained. The change took place in educated London speech around the end of the 16th century, and explains why there is no [?] sound at the end of words like fang, sing, wrong and tongue in the standard varieties of Modern English.
The change in fact applies not only at the end of a word, but generally at the end of a morpheme. If a word ending in -ng is followed by a suffix or is compounded with another word, the [?] pronunciation normally remains. For example, in the words fangs, sings, singing, singer, wronged, wrongly, hangman, there is no [?] sound. An exception is the comparative and superlative forms of adjectives: in the words longer/longest, stronger/strongest, younger/youngest, the [?] is pronounced in most accents.
In other cases (when it is not morpheme-final), word-internal -ng- does not show the effects of coalescence, and the pronunciation  is retained, as in finger and angle. This means that the words finger and singer do not rhyme in most modern varieties of English, although they did in Middle English. The process of NG-coalescence might therefore be referred to as the singer-finger split.
Some accents, however, do not show the full effects of NG-coalescence as described above, and in these accents sing may be found with , the suffix -ing may be pronounced  (though G-dropping is another possibility), and singer may rhyme with finger. This is particularly associated with English English accents in an area of northern England and the Midlands, including the cities of Birmingham (see Brummie), Manchester, Liverpool (see Scouse), Sheffield and Stoke-on-Trent. It is also associated with some American English accents in the New York City area.
On the other hand, in some accents of the west of Scotland and Ulster, NG-coalescence is extended to morpheme-internal position, so that finger is pronounced /'fr/, thus rhyming with singer (although the [?] is not dropped before a stressed syllable, as in engage).
It is because of NG-coalescence that /?/ is now normally regarded one of the phonemes of standard English. In Middle English, the [?] can be regarded as an allophone of /n/, occurring before velar consonants, but in Modern English, in view of minimal pairs such as pan-pang and sin-sing, that analysis no longer appears to hold. Nevertheless, some linguists (particularly generativists) do regard a word like sing as being underlyingly /s?n?/, positing a rule that deletes [?] after a nasal before a morpheme boundary, after the nasal has undergone assimilation. A problem with this view is that there are a few words in which [?] is followed neither by a velar nor a morpheme boundary (such as gingham, and dinghy and Singapore for those speakers who pronounce them without [?]), and some in which the [?] is not deleted before a morpheme boundary (longer etc., as noted above).
The above-mentioned accents which lack NG-coalescence may more easily be analyzed as lacking a phoneme /?/. The same may apply to those where NG-coalescence is extended to morpheme-internal position, since here a more consistent [?]-deletion rule can be formulated.
G-dropping is a popular name for the feature of speech whereby /n/ is used in place of the standard /?/ in weak syllables. This applies especially to the -ing ending of verbs, but also in other words such as morning, ceiling, Buckingham, etc. G-dropping speakers may pronounce this syllable as [?n] or [?n] (reducing to a syllabic [n] in some cases), while non-G-dropping speakers have //, or more rarely // or /i?/.
Relative to the great majority of modern dialects, which have NG-coalescence, G-dropping does not involve the dropping of any sound, simply the replacement of the velar nasal with the alveolar nasal. The name derives from the apparent orthographic consequence of replacing the sound written ⟨ng⟩ with that normally written ⟨n⟩. The spelling -in' is sometimes used to indicate that a speaker uses the G-dropping pronunciation, as in makin' for making.
The pronunciation with /n/ rather than /?/ is a long-established one. Old English verbs had a present participle in -ende and a verbal noun (gerund) form in -ing(e). These merged into a single form, written -ing, but not necessarily spoken as such - the /n/ pronunciation may be inherited from the former distinct present participle form. The /n/ variant appears to have been fashionable generally during the 18th century, with the alternative // being adopted in educated speech around the 1820s, possibly as a spelling pronunciation.
Today, G-dropping is a feature of colloquial and non-standard speech of all regions, including stereotypically of Cockney, Southern American English and African American Vernacular English. Its use is highly correlated with the socioeconomic class of the speaker, with speakers of lower classes using /n/ with greater frequency. It has also been found to be more common among men than women, and less common in more formal styles of speech.
The fact that the /n/ pronunciation was formerly associated with certain upper-class speech is reflected in the phrase huntin', shootin' and fishin' (used in referring to country gentry who frequently engaged in such field sports). Further evidence that this pronunciation was once standard comes from old rhymes, as in this couplet from John Gay's 1732 pastoral Acis and Galatea, set to music by Handel:
Which was presumably pronounced "shepherd, what art thou pursuin', heedless runnin' to thy ruin", although this would sound very odd in an opera today. Similarly, in the poetry of Jonathan Swift (1667-1745), -ing forms consistently rhyme with words in /?n/, as in this verse of A Ballad on the Game of Traffic, where "lining" rhymes with "fine in":
In later Middle English, the final cluster /mb/ was reduced to just /m/ (the plum-plumb merger). This affects words such as lamb and plumb, as well as derived forms with suffixes, such as lambs, lambing, plumbed, plumber.
By analogy with words like these, certain other words ending in /m/, which had no historical /b/ sound, had a silent letter ⟨b⟩ added to their spelling by way of hypercorrection. Such words include limb and crumb.
Where the final cluster /mn/ occurred, this was reduced to /m/ (the him-hymn merger), as in column, autumn, damn, solemn. (Compare French automne, where the cluster has been reduced to /n/.) Both sounds are nonetheless still pronounced before vowels in certain derivatives, such as autumnal and solemnity.
|test||-> tes'||[t?st]||-> [t?s(?)]|
|desk||-> des'||[d?sk]||-> [d?s(?)]|
|hand||-> han'||[hænd]||-> [hæn(?)]|
|send||-> sen'||[s?nd]||-> [s?n(?)]|
|left||-> lef'||[l?ft]||-> [l?f(?)]|
|wasp||-> was'||[w?sp]||-> [w?s(?)]|
When a consonant cluster ending in a stop is followed by another consonant or cluster in the next syllable, the final stop in the first syllable is often elided. This may happen within words or across word boundaries. Examples of stops that will often be elided in this way include the [t] in postman and the [d] in cold cuts or band saw.
Historically, similar reductions have taken place before syllabic consonants in certain words, leading to the silent ⟨t⟩ in words like castle and listen. This change took place around the 17th century. In the word often, the [t] sound later came to be re-inserted by some speakers as a spelling pronunciation.
For many speakers, an epenthetic [t] is inserted in the final cluster /ns/, making it identical or very similar to the cluster /nts/. For example, the words prince and prints have come to be homophones or nearly so.
The epenthesis is a natural consequence of the transition from the nasal [n] to the fricative [s]; if the raising of the soft palate (which converts a nasal to an oral sound) is completed before the release of the tongue tip (which enables a fricative sound), an intervening stop [t] naturally results. The merger of /ns/ and /nts/ is not necessarily complete, however; the duration of the epenthetic [t] in /ns/ has been found to be often shorter (and the [n] longer) than in the underlying cluster /nts/. Some speakers preserve a clearer distinction, with prince having [ns], and prints having [nts] or [n?s]. The epenthesis does not occur between syllables, in words like consider.
The merger of /nz/ and /ndz/ is also possible, making bans and pens sound like bands and pends. However, this is less common than the merger of /ns/ and /nts/ described above, and in rapid speech may involve the elision of the /d/ from /ndz/ rather than epenthesis in /nz/.
Epenthesis of a stop between a nasal and a fricative can also occur in other environments, for example:
An epenthetic [p] often intervenes in the cluster /mt/ in the word dreamt, making it rhyme with attempt.
Some originally epenthetic consonants have become part of the established pronunciation of words. This applies, for instance, to the /b/ in words like thimble, grumble and scramble.
For the insertion of glottal stops before certain consonants, see Glottalization below.
|Finns||fends||'f?n(d)z||With pen-pin merger.|
|fins||fends||'f?n(d)z||With pen-pin merger.|
|Hans||hands||'hæn(d)z||Hans may also be pronounced /'h?:nz/ or /'h?:ns/.|
|Heinz||hinds||'ha?n(d)z||Heinz may also be pronounced /'ha?nts/.|
|hence||hints||'h?n(t)s||With pen-pin merger.|
|inns||ends||'?n(d)z||With pen-pin merger.|
|pins||pends||'p?n(d)z||With pen-pin merger.|
|rinse||rents||'r?n(t)s||With pen-pin merger.|
|since||cents||'s?n(t)s||With pen-pin merger.|
|since||scents||'s?n(t)s||With pen-pin merger.|
|tense||tints||'t?n(t)s||With pen-pin merger.|
|tins||tends||'t?n(d)z||With pen-pin merger.|
|Vince||vents||'v?n(t)s||With pen-pin merger.|
|whines||winds||'wa?n(d)z||With wine-whine merger.|
In English as in other languages, assimilation of adjacent consonants is common, particularly of a nasal with a following consonant. This can occur within or between words. For example, the /n/ in encase is often pronounced [?] (becoming a velar nasal by way of assimilation with the following velar stop /k/), and the /n/ in ten men likely becomes [m], assimilating with the following bilabial nasal /m/. Other cases of assimilation also occur, such as pronunciation of the /d/ in bad boy as [b]. Voicing assimilation determines the sound of the endings -s (as in plurals, possessives and verb forms) and -ed (in verb forms): these are voiced ([z], [d]) following a voiced consonant (or vowel), but voiceless ([s], [t]) after a voiceless consonant, as in gets, knocked.
While there are many accents (such as Cockney) in which syllable-final /t/ is frequently glottalized (realized as a glottal stop, [?]) regardless of what follows it, the glottaling of /t/ in clusters is a feature even of standard accents, such as RP. There, [?] may be heard for /t/ in such words and phrases as quite good, quite nice, nights. More precisely, it occurs in RP when /t/ appears in the syllable coda, is preceded by a vowel, liquid or nasal, and it is followed by another consonant except (normally) a liquid or semivowel in the same word as in mattress.
Another possibility is pre-glottalization (or glottal reinforcement), where a glottal stop is inserted before a syllable-final stop, rather than replacing it. That can happen before /p/, /t/ and /k/ or also before the affricate /t?/. It can occur in RP in the same environments as those mentioned above, without the final restriction so a glottal stop may appear before the /t/, as in mattress. It can also occur before a pause as in quite! spoken alone but not in quite easy. In the case of /t?/, pre-glottalization is common even before a vowel, as in teacher.
According to Wells, this pre-glottalization originated in the 20th century (at least, it was not recorded until then). Glottalization of /t/ spread rapidly during the 20th century.
Final consonant clusters starting with /s/ sometimes undergo metathesis, meaning that the order of the consonants is switched. For example, the word ask may be pronounced like "ax", with the /k/ and the /s/ switched.
This example has a long history: the Old English verb áscian also appeared as acsian, and both forms continued into Middle English, the latter, metathesizing to "ask". The form axe appears in Chaucer: "I axe, why the fyfte man Was nought housband to the Samaritan?" (Wife of Bath's Prologue, 1386), and was considered acceptable in literary English until about 1600. It persists in some dialects: for example, in Ulster Scots as /'aks/, and in Jamaican English as /'a:ks/, from where it has entered London speech as /'?:ks/.
S-cluster metathesis is common in African American Vernacular English, although it is one of the most stigmatized features of AAVE, often commented on by teachers. Examples of possible AAVE pronunciations include:
The form has been found to occur in Gullah and in the speech of some young African Americans born in the Southern United States. It is reported to be a highly stigmatized feature, with children who use it often being referred to speech pathologists.