1. English-speakers, more than the speakers of most other languages, tend to give great weight to important syllables and much of the time to treat relatively neglectfully the less important syllables of words. This custom has very notable effects on the changes of form which words undergo in English. The kinds of process that bring about these frequent alterations to the shapes of words can be summarised under the headings of assimilation (including coalescence), elision, compression and liaison /li`eɪzn/.
2. Uttered entirely alone, words take their "citation" or "lexical" forms as shown in dictionaries, but the majority of the most commonly used English words frequently change in form (losing and/or changing phonemes) as they undergo reduced articulations. These reductions are mainly the results of pressures exerted by the process of rhythmically integrating groups of words which the speaker wishes to convey to be unified grammatical entities.
3. Fortunately, only a tiny fraction of these reduced forms of words need regularly be adopted by EFL users. Only forty-odd words are the exceptions which constitute this fraction. These are dealt with under "Weakform Words & Contractions" elsewhere on this Website. Let's look briefly at each of the categories we listed above .
4. Assimilation (ie alteration of an original sound by the influence of an adjacent one: the term was recorded as applied to consonants in OED from 1871) is a topic that the ordinary EFL user needn't much worry about because practically all the assimilations that arise in continuous speech are "optional". That is to say, among native speakers they are very frequent but sporadic rather than invariable. Perhaps the most regularly occurring types are from / s / or / z / to / ∫ / or / ʒ / as in apprenticeship /ə`prentɪʃʃɪp/, bus-shelter / `bʌʃ ʃeltə /, dress shop /`dreʃ ʃɒp/, ice-show /`aɪʃ ʃəʊ/, less sure /'leʃ `ʃɔː/, Miss Jones /mɪʃ `ʤəʊnz/, S-shaped /`eʃ ʃeɪpt/ and has she / `hæʒ ʃi /etc. The only important thing for the EFL user to remember is not to artificially slow down articulation and thus spoil the fluency of an utterance in order to produce for example an unchanged /z/ in is she, or /s/ in horse shoe / `hɔːʃ ʃuː/ or `tortoise-shell /`tɔːtəʃ ʃel /. With fluent rhythm the articulation can usually be left to take care of itself.
Certain assimilations involve coalescence, notably of /t/ and /d/ with following /j/ as in did you uttered as / `dɪʤu /. Sometimes Tuesday is represented (usually by sneering writers) as uttered by some speakers as Chewsday and dew as Jew, etc. Expression of disapproval of such usages is now very old-fashioned. EFL users should feel free to adopt these coalescent forms if they find them more comfortable to use.
5. One of the very small number of assimilations that it would sound abnormal not to make is the sharpening to / s / of the s of used to when it means accustomed to. The verb use has / z / for its s in all other circumstances. The same sort of thing often happens to supposed to but it won't sound strange if you don't sharpen the s there nor sharpen the traditional /v/ to / f / in have to or of course or the /z/ to /s/ in has to – though a great many speakers do make those assimilations very frequently.
6. You may be surprised to see that dictionaries show as the usual forms for gooseberry, raspberry, fivepence, newspaper / `gʊzbri, `rɑːzbri, `faɪfpəns, `njuːspeɪpə / but these pronunciations are not continuous-speech adaptations, just simply common invariable forms for most speakers. (In Northern England one finds in some cases differences from General British in such items.)
7. The outstanding thing that makes assimilation a notable EFL topic is the strong tendency for speakers of Dutch, French, Greek and various Slavonic and African languages to soften a sharp consonant at the end of a word followed by another beginning with a soft consonant. Such people tend to produce versions which suggest the spellings bag `door, `baizeball, `buzz route, `eyesberg, `Jews' bottle, `wodgeman, `rizzwatch, robe `ladder, rose `beef and on `whore's back for back `door, `baseball, `bus route, `iceberg, `juice bottle, `watchman, `wristwatch, rope `ladder, roast `beef and on `horseback.
8. The reverse of this is done by some eg Dutch speakers. They may seem to be saying dretful, fock patches or flackpoles when they are aiming to say the English words dreadful, fog patches or flagpoles. Yorkshire people are the only ones in the English-speaking world with this tendency: eg they (including in his day the well-known author J. B. Priestley) tend to pronounce Bradford as Bratford, actually usually [`bræʔfəd]. Many EFL users tend to produce yet other types of assimilations that sound very abnormal such as I'd love one with a /v/ instead of the normal / w / beginning the word /wʌn/.
9. Elision (the omission of a speech sound; linguistic use noted by OED from 1581) is likewise (weakform words aside) not an important topic for the EFL user — with some very minor exceptions. Expressions like `breaststroke, first `stop, masked `ball, next `time naturally tend to lose the final / t / of the first word. Any obvious slowing down of the natural appropriate rate of utterance in order to manage to produce such a / t / is best avoided. Even what might be written good `eel, take `air, pry `minister and extra `tension for good `deal, take `care, prime `minister and extra at`tention are commonplace native-speaker variants.
10. It is now, despite the impression given by many dictionaries, quite unusual to make four syllables of the many common ultimately Latin-derived adverbs like actually, generally, obviously, usually etc or even to make temporarily different from temporally, though in this word many British speakers now postpone the main stress saying /tempə`rerəli/ as Americans generally do.
11. English speakers generally do not elide the final plosive of a word which is closely followed by another word that begins with an obstruent (ie a plosive, affricate or fricative) or nasal consonant but they do not release it. If one does release such a plosive, the unfortunate result is strikingly unnatural in pronunciations like take care with the first /k/ released and postman, windmill or grandfather similarly produced. These sound too much like take a care, posterman, windowmill and grander father.
12 One needs to be constantly on guard against being misled by our very archaic spelling into restoring any of the very numerous historical elisions which are to be seen in words such as blackguard / `blægɑːd /, Christmas / `krɪsməs /, cupboard / `kʌbəd /, evening / `iːvnɪŋ /, several / `sevrl / soften / `sɒfn / etc (contrast often which has a very common spelling-influenced pronunciation with its former / t / restored ). One now even hears an / l / in calm from a very small minority of English people as one can from many Americans. EFL users should also be careful not to elide the second element of the English affricates / ʧ / and / ʤ / which would produce outlandish expressions sounding like what one might write as villid `church, 8 G `Wells and `what chain rather than village `church, H. G.`Wells and `watch chain.
13. It's necessary also to avoid simplifying awkward consonant sequences in ways that English speakers don't adopt. For example, for months / mʌnθ / is quite unacceptably abnormal whereas omitting the / θ / to give / mʌns / is perfectly alright. Fifths and twelfths tend to contain co-articulatory blendings of two or more consonants produced simultaneously by native English speakers. LPD includes the versions / fɪθs / and / twelθs / which it records (one may be confident from the refraining from adverse comment) as unremarkable and must be far safer for the EFL speaker to adopt than to attempt a triple or quadruple consonant ending. With the form / sɪks / for sixths LPD is less tolerant labelling it as "casual" possibly bowing to the many who would profess to be shocked if a teacher were found to be recommending a pronunciation which failed to distinguish sixths from six. Something which most GB speakers and many Americans had come to accept very widely by the end of the twentieth century was the reduction in any word of final / -sts / to simply /-st /. Such versions now constantly pass unnoticed even in isolated words, though the dictionaries and textbooks are being very slow getting round to acknowledging this so widespread failure to differentiate singular and plural in such words.
14. Compression (first proposed as a institutionalised linguistic term in Windsor Lewis 1969 A Guide to English Pronunciation) is the reduction of articulatory movement typically resulting in elimination of a syllable as when / `reɪdiəʊ / becomes /`reɪdjəʊ/ or / `fɒləʊɪŋ / becomes / `fɒlwɪŋ /. This can also largely be disregarded by the EFL user. However, it is desirable to have the normally reduced versions not making three clear syllables of closing-diphthong-plus-schwa when an unstressed syllable follows as in words like fireman and powerful or they may well sound over-carefully enunciated. Similarly words like empire and `rush hour won't sound natural if uttered as three syllables. Most word-internal syllabic consonants as in eg fattening or settler or civilisation are constantly compressed into unsyllabic ones. Note the compression loops used in the LPD (the Longman Pronunciation Dictionary) eg between the / l / and the final schwa at settler /set ̬lə/ to convey this alternation and the useful account in LPD of the phenomenon at its alphabetical entry "compression". Younger speakers are currently increasingly tending to avoid such "de-syllabications". In extremes not recommendable for imitation they may introduce schwas which can sound pedantic or fussy and where some have probably never before existed eg in accidentally, amply, assembly, trembling, fattening, maddening, handling, spindly, wobbling etc.
15. Liaison (in OED as an institutionalised linguistic term first recorded in 1884; "linking r" is not noted by the OED before 1950 but Daniel Jones introduced its use in the 1920s; it seems to have appeared first in print in Ida Ward's Phonetics of English in 1929). This phenomenon is the linking to a following word of a preceding one by employment of a final sound not present in the isolate form of the first word. It's found in English when a word ending in a non-close vocalic sound, / ɑː, ɔː, ɜː/ or a schwa [ə], is closely followed by a word beginning with any vocalic sound (ie vowel or diphthong). Most such words have a final -r (or -re) in their spelling; if not, the / r / used is often labelled 'intrusive' and is sometimes criticised by purists (ie people who set themselves up as arbiters of how other people should speak), even when its use is virtually universal as in the phrase the idea / r / of it.
16. Omission of linking / r / is in many cases quite "optional" but its absence from common expressions usually uttered with close rhythmic integration can sound quite strange. So EFL users are best advised to cultivate it in eg phrases such as our own, or else, better and better, later on, far off, other end, pair of etc. The purists like to complain that people say Laura Norder instead of law and order. Note also the jokey mock book-title "Dog's Delight by Nora Bone" (gnaw / r / a bone). But huge numbers of people of unimpeachable education say such things and even also draw/r/ing and I saw / r / it without noticing that they're doing so and without being noticed as doing so by the great majority of other speakers. These last two types need not be particularly cultivated by EFL users.