Hitherto, we have looked at sounds individually. But language in everyday use is not conducted in terms of isolated, separate units; it is performed in connected sequencesof larger units, in words, phrases and longer utterances. Now we shall be concerned with what happens to sounds not only within words, but also when the words are connected into larger units. There are actually some remarkable differences between the pronunciation of a word in isolation and of the same word in a block of connected speech. These changes are mostly quite regular and predictable. Still the problem of defining the phonemic status of sounds in connected speech is by far too complicated because of the numerous modifications of sounds in speech. These modifications are observed both within words and at word boundaries. As you may know from the practical course of phonetics, speech sounds influence each other in the flow of speech. As a result of the intercourse between consonants and vowels and within each class there appear such processes of connected speech as assimilation, accommodation, vowel reduction and elision which is sometimes termed deletion.
The adaptive modification of a consonant by a neighbouring consonant in the speech chain is known as assimilation,e.g the alveolar [t] followed by the interdental  becomes dental: eighth, at three.
The term accommodationis often used by linguists to denote the interchanges of "vowel + consonant type" or "consonant + vowel type", for instance, some slight degree of nasalization of vowels preceded or followed by nasal sonorants: never, men; or labialization of consonants preceding the vowels [o] and [y] in Russian: больно, конь, думать, лучше.
One of the wide-spread sound changes is certainly vowel reduction.Reduction is actually qualitative or quantitative weakening of vowels in unstressed positions, e.g. board — blackboard, man —postman.
Elision or complete loss of sounds, both vowels and consonants, is often observed in English. Elision is likely to be minimal in slow careful speech and maximal in rapid relaxed colloquial forms of speech.
The processes involved cannot be neglected in defining the phonemic status of speech sounds. These phenomena manifest the economy of pronouncing efforts on the part of the speaker. The speaker and the listener are two participants of communication. The addresser's aim is to inform the addressee of something. The letter's wish is to comprehend the idea. The simplifications themselves go quite unnoticed by the listener, as they do not affect the meaning. The listener is mainly interested in the meaning the speaker aims to convey and not in the precise phonetic organization of a block of connected speech. So long as the meaning is recognizable, the listener is satisfied. We have to regard the omissions and reductions then as a kind of economy on the part of the speaker, who aims not to give more information than necessary. The speaker assumes usually correctly that the listener will not notice the omissions. A question then arises whether such kind of simplification leads to excessive ambiguity. On the face of it, one would expect a great-deal of ambiguity to arise, particularly in cases where a whole word is reduced (weakened) to a single sound, for example, [z] for the whole word has. The load carried by a single sound can become enormous, [z], for example, can represent the reduced forms of has, is and even does, the plural and possessive for nouns, the third person singular for verbs. The sound [э] can represent the reduced forms of are, or, her and sometimes of (as in six o'clock), as well as the indefinite article a, the comparative degree of adjectives (shorter), the suffix of a noun (teach-er), etc. But in spite of the meaning load carried by the same sounds, ambiguity rarely arises because the syntactic functions are quite different and the context makes the intention clear. On hearing a sequence like [z'nik Чслгшп] the listener unmistakably reconstructs: Is Nick coming?; in [5э 'boiz 'skeit] the sound [z] must be the plural form, while in [дэ 'boiz '(Un it] it must be derived from has.
Now let us see which qualitative features of consonant sounds may be changed in the process of their interrelation in a speech chain.
Consonants are modified according to the place of articulation. Assimilation takes place when a sound changes its charac ter in order to become more like a neighbouring sound. The characteristic which can vary in this way is nearly always the place of articulation, and the sounds concerned are commonly those which involve a complete closure at some point in the mouth that is plosives and nasals which may be illustrated as follows:
a) The dental [t], [d], followed by the interdental ,  sounds (partial regressive assimilation when the influence goes backwards from a "later" sound to an "earlier" one), e.g. "eighth",
"at the", "breadth", "said that".
b) The post-alveolar [t], [d] under the influence of the post- alveolar [r] (partial regressive assimilation), e.g. "free", "true""that right word", "dry", "dream", "the third room".
c) The post-alveolar [s], [z] before [f] (complete regressive as similation), e.g. horse-shoe ['ho:JJu:], this shop [Qi/'/op], does she
4. The affricative [t + j], [d + j] combinations (incomplete regressive assimilation), e.g. graduate ['graedjueit], congratulate' [kan'graetfuleit], did you [didju:], could you ['kudju:], what do you say ['wot3u:'sei].
It is easy to see from the examples above that the sounds commonly changing their place of articulation are alveolar stops. Nasal consonants are not less susceptible to assimilation. The place of articulation of nasals also varies according to the consonant that follows, e.g.
In camp [m] remains bilabial before another bilabial as well as in man before'a vowel.
Similarly in cent [n] is alveolar before another alveolar as well as in net.
But in "symphony" [m] is actually labio-dental followed by the labio-dental [f].
In "seventh" [n] becomes dental, before the interdental [в].
In "pinch" [n] is palato-alveolar corresponding to the following affricate [tf].
In "thani" [n] assimilates to the velar consonant becoming velar [ŋ].
We should like to note here that by analogy with alveolar consonants nasal assimilation operates not only within the morpheme as in "thank but also across syllable boundaries as in "symphony", across morpheme boundaries, for example, in prefixes in-, ил- as in "incomplete", "ungrateful"; "impractical", where [n] assimilates to [p] and becomes bilabial [m], in the stressed prefix con- as in "conquer". Assimilation of nasals seems to be also optional across word boundaries, e.g. m case, in fact.
The manner of articulation is also changed as a result of assimilation, which may be illustrated as follows:
a) Loss of plosion. In the sequence of two plosive consonants the former loses its plosion: glad to see you, great trouble, and old clock (partial regressive assimilations).
b) Nasal plosion. In the sequence of a plosive followed by a nasal sonorant the manner of articulation of the plosive sound and the work of the soft palate are involved, which results in the nasal character of plosion release: "sudden", "not now", "at night", "lerme see" (partial regressive assimilations).
c) Lateral plosion. In the sequence of a plosive followed by the lateral sonorant  the noise production of the plosive stop is changed into that of the lateral stop: "settle", "table", "at last"
(partial rogressive assimilations). It is obvious that in each of the occasions one characteristic feature of the phoneme is lost.
The voicing value of a consonant may also change through assimilation. This type of assimilation affects the work of the vocal cords and the force of articulation. In particular voiced lenis sounds become voiceless fortis when followed by another voiceless sound, e.g.:
1. Fortis voiceless/lenis voiced type of assimilation is best manifested by the regressive assimilation .in such words as newspaper (news [z] + paper); gooseberry (goose [s] + berry). In casual informal speech voicing assimilation is often met, e.g. have to do it ['hasf ta 'du:j, five past two ['faif past 'tit]. The sounds which as similate their voicing are usually, as the examples show, voiced lenis fricatives assimilated to the initial voiceless fortis consonant of the following word. Grammatical items, in particular, are most affected: [z] of has, is, does changes to [s], and [v] of of, have be comes [f], e.g.
She's five. Of course.
She has fine eyes. You've spoiled it.
Does Pete like it?
2.The weak forms of the verbs is and has are also assimilated to the final voiceless fortis consonants of the preceding word thus the assimilation is functioning in the progressive direction, e.g.
Your aunt's coming.
What's your name?
(partial progressive assimilation)
3. English sonorants [m, n, r, 1, j, w] preceded by the fortis voiceless consonants [p, t, k, s] are partially devoiced, e.g. "smart", "snake", "fray", "qruick", "fmns", "ptey", "pride" (partial progressive assimilation).
The voiced/voiceless type of assimilation is well developed in the Russian language, e.g. сдавать, сбросить, французский, абсолютный. The positional devoicing of final consonants is especially constant, e.g. клуб, снег, мороз.
It should be noted that the interference of the Russian voiced/voiceless regressive type of assimilation results in a typical mistake in English: "black dog", "this day", "gets dark", "much better", "let's go". In English assimilation usually results in changing voiced lenis consonants into voiceless fortis, e.g. of course [afkors]. The change of voiceless fortis consonants into voiced lenis as a result of assimilation is not typical. Thus teachers of English should be aware of it and be ready with special exercises to prevent the errors.
Lip positionmay be affected by the accommodation, the interchange of consonant + vowel type. Labialisation of consonants is traced under the influence of the neighbouring back vowels (accommodation), e.g. pool, moon, rude, soon, who, cool, etc. It is possible to speak about the spread lip position of consonants followed or preceded by front vowels [i:], [i], e.g. tea — beat; meet — team; feat — leaf, keep — leak; sit — mz'ss (accommodation).
The position of the soft palateis also involved in the accommodation. Slight nasalization as the result of prolonged lowering of the soft palate is sometimes traced in vowels under the influence of the neighbouring sonants [m] and [n], e.g. and, morning, men, come in (accommodation).
To summarize so far, assimilation affecting the place of articulation is considered to be most typical of the English sound system and assimilation affecting the work of the vocal cords (voiced/voiceless type) is most typical of the Russian speech.
It is to be noted that the described allophonic realizations of phonemes are marked in Received Pronunciation as obligatory and stable for all the members of the speech community in every phonetic style. It is perfectly natural that all sorts of sound adaptation are more frequent in informal colloquial flow of speech than in formal speech. This tendency is a matter of style, not correctness. In informal casual discourse assimilation involves the alveolar stops [t], [d] before another stop at border junctions, e.g.
that place [ð æ p 'pleis] Or: hard problem ['ha:d 'problem]
that book [' ð æ p'buk] hard blow ['ha:b Ъ1зu]
that kind [' ð æ k 'kaind] hard case ['ha:g 'keis]
that golfer [ð æ k 'golfa] hard ground ['ha:g 'graund]
In these examples [t] retains its voicelessness, [d] accordingly retains its voiced character, but both of them shift their articulation in symphony with the articulation of the following stop. It should be noted that the velar stops [k], [g] are not subjected to the assimilation of this kind.
The alveolars [s], [z] and [t], [d] assimilate in informal casual speech more often than in slow careful speech to palato-alveo-lars when followed by the palatal [j], e.g.
face your friend ['feijja Trend] as you like [33 ju 'laik] can't you do it ['ka:nt/9 'du: it] on duty [on 'cfeurti]
The examples above illustrate the changes affecting the place of articulation.
In informal casual speech complete type of assimilation is often observed, e.g.
ten minutes ['tern 'mmits] nice shoes ['naif 'Jkz] one more [Чулт 'тэ:] good-bye ['gub 'bai] let me ['lem mi-]
We would like to point out right here that elision or complete loss of sounds, both vowels and consonants, is observed in the structure of English words. It is typical of rapid colloquial speech and marks the following sounds:
1. Loss of [h] in personal and possessive pronouns he, his, her, him and the forms of the auxiliary verb have, has, had is wide spread, e.g. What has he done? ['wot эг i- xdAn].
2.  tends to be lost when preceded by fo:], e.g. always ['o:wiz], already [o:'redi], all right [d: 'rait].
3. Alveolar plosives are often elided in case the cluster is followed by another consonant, e.g. next day ['neks 'dei], just one ['45AS vah], mashed potatoes ['maef pa'teitsuz]. If a vowel follows, the consonant remains, e.g. first of all, passed in time. Whole syllables may be elided in rapid speech: library ['laibn], literary [Чип].
Examples of historical elision are also known. They are initial consonants in write, know, knight, the medial consonant [t] in fasten, listen, whistle, castle.
In sum, we may say that in the process of speech the degree of sound modifications may be different, varying from partial assimilation, when one sound feature is modified, like in "tenth" (alveolar [n] becomes dental) to actual loss of a sound. For example: listen ['lisn], next day ['neks 'dei], complete sound adaptation being the intermediate state: ten minutes ['tern 'mmits], nice shoe ['naif'Ju:].
Describing the interrelation of sounds in cannected speech we would like to mention one more remarkable phenomenon.
While the elision is a very common process in connected speech, we also occasionally find sounds being inserted. When a word which ends in a vowel is followed by another word beginning with a vowel, the so-called intrusive "r" is sometimes pronounced between the vowels, e.g.
Asia and Africa ['eijbr and 'aefrika] the idea of it [dtai'diar эу it] ma and pa ['ma:r and 'pa:]
The so-called linking "r," is a common example of insertion, e.g. clearer, a teacher of English.
Thus it is clear that the linking and intrusive [r] are both part of the same phonetic process of [r] insertion.
When the word-final vowel is a diphthong which glides to [i] such as [ai], [ei] the palatal sonorant [j] tends to be inserted, e.g saying ['seijirj]; trying ['trajin].
In case of the [u]-gliding diphthongs [зи], [аи] the bilabial sonorant [w] is sometimes inserted, e.g. going ['gauwin], allowing [s'lauwin].
The process of inserting the sonorants [r], [j] or [w] may seem to contradict the tendency towards the economy of articulatory efforts. The explanation for it lies in the fact that it is apparently easier from the articulatory point of view to insert those sounds than to leave them out.
The insertion of a consonant-like sound, namely a sonorant, interrupts the sequence of two vowels (VV) to make it a more optional syllable type: consonant + vowel (CV). Thus, insertion occurs in connected speech in order to facilitate the process of articulation for the speaker, and not as a way of providing extra information for the listener.
Now by way of conclusion we should like to say that we understand the sound quality as a set of characteristics which are in constant interrelation and compensation. In case one of the features of a phoneme is lost there remains a sufficient number of characteristics of a phoneme and its status and function are not lost. Thus modifications of sounds in a speech chain are of allo-phonic character, that is they are realizations of allophones of phonemes.