CONNECTED SPEECH
The various features of articulation interact, overlap and influence the
pronunciation of words. These so-called coarticulations are especially present in
connected speech. The pronunciation of connected words is particularly prone to
alteration across word boundaries, i.e. where one word meets another
immediately following word. Consequently, the sounds that are most affected are
the sounds at the ends of words and the sounds at the beginning of words. For
example, the phrase that pen /ðæt pɛn/ sounds more like [ðæp pʰɛn] in connected
speech. Here the word-final /t/ of the word that assimilates to the place of
articulation of the word-final /p/ of the immediately following word pen. We will
examine three processes that operate at word boundaries:
         • ASSIMILATION
         • LIAISON
         • ELISION
           Following this, we will conclude by discussing a fourth process that
           similarly arises owing to the effects of rapidly articulated speech
           but which does not operate at word boundaries:
         • REDUCTION
Before proceeding it is worth making a general point. Sounds across word
boundaries may alter but they do not necessarily have to. The speech of a person
who is particularly careful over their enunciation may not be appreciably affected
by processes such as assimilation, liaison, elision and reduction. However, it
would be unusual if none of a person’s connected speech were influenced. The
degree to which such processes impact on connected speech will, therefore, vary
from person to person.
Connected speech processes
 Processes affecting pronunciation in rapidly articulated connected speech are
discussed in this section..
                                        .
                                  A. ASSIMILATION
At the word level, assimilation takes place when one speech segment is transposed into another
owing to the influence of a neighboring segment within the same word. An example of this is
the dentalization of alveolars when they occur in close proximity to dental fricatives, e.g. width
/wɪdθ/ → [wɪd̪θ]. Here, the alveolar /d/ assimilates the place of articulation of the immediately
following dental fricative /θ/. It is also possible that a segment assimilates the voicing of a
neighboring segment. We have come across this already, when we considered de-voicing of
voiced consonants that appear in word-final position. An example is the word kid /kɪd/ being
realized as [kʰɪd̥]. As the word-final /d/ is adjacent to a neighboring silence (i.e. the end of the
word) it assimilates the voicelessness of this silence, thereby de-voicing the phoneme to create
the allophone [d̥]. De-voicing of word-final consonants is, therefore, another example of
an assimilatory process. There are also, as we shall see later, assimilations of the manner of
articulation of neighboring segments.
In connected speech, assimilation may take place at word boundaries. There are broadly two
types of assimilation:
       I.      Allophonic assimilation
       II.     Phonemic assimilation
I. Allophonic assimilation overview
Consider the following phrase.
bad thought                  /bæd θɔt/
There is a word boundary between the word bad and the word thought. The sound at the end of
the word bad is the alveolar /d/ and the sound at the beginning of the word thought is the dental
fricative /θ/. Now, just as alveolars may become dentalized within a word, the same process of
assimilation may occur across a word boundary. Consequently, the phrase may be realized as
follows.
bad thought                  /bæd θɔt/                    →          [bæd̪ θɔt]
In this instance, the final /d/ of bad is dentalized in anticipation of the upcoming dental /θ/ that
initiates the immediately following word thought. The alveolar /d/ is said to assimilate the place
of articulation of the neighboring dental fricative. This is an example of allophonic
assimilation because the new, transformed sound is an allophone of the original phoneme, i.e.
[d̪] is an allophone of /d/.
We will consider four types of allophonic assimilation in connected speech:
   1. dentalization
   2. labialization
   3. de-vocing of liquids
   4. disapplication
       1.Allophonic assimilation – dentalization To reiterate, assimilation occurs
       when one sound is altered owing to the influence of a neighboring segment. In
       the previous example we saw how alveolars may be dentalized across a word boundary
       when followed immediately by a dental sound. A further example should serve to
       consolidate this concept:
hot thing                  /hɒt θɪŋ/
       In the phrase hot thing the word boundary occurs between the first word hot and the
       immediately following word thing. The first word hot ends with the alveolar sound /t/.
       This sound, therefore, precedes the word boundary. The first sound following the word
       boundary is the dental consonant /θ/ in the word thing. The alveolar sound before the
       word boundary is, consequently, dentalized in anticipation of the upcoming dental
       sound that occurs immediately after the word boundary, i.e.
hot thing                  /hɒt θɪŋ/                   →          [hɒt̪ θɪŋ̃ ]
       Note that in this example the dentalized [t̪ ] is produced without aspiration as it is
       unreleased. That is to say, the oral cavity is occluded to create the stop consonant but
       there is no audible release in the plosive phase of its production, i.e.
hot thing                  /hɒt θɪŋ/                   →          [hɒt̪̪̚ θɪŋ̃ ]
       Further examples of dentalization across word boundaries include the following.
hit this                   /hɪt ðɪs/                   →         [hɪt̪̪̚ ðɪs]
ten things                 /tɛn θɪŋz/                  →         [tɛñ ̪ θɪŋ̃ z̥]
tell that                  /tɛl ðæt/                   →         [tɛl̪ ðæʔ]
       In summary, we see that as well as dentalization occurring within words, this process
       can also operate in connected speech across word boundaries. The above examples
       demonstrate how alveolars before a word boundary are dentalized if the initial sound of
       the next word is a dental. Under these circumstances, the alveolar assimilates the place
       of articulation of the dental. We can summarize this into a rule about dentalization
       across word boundaries:
       Rule ‘If the sound in word-final position of the word preceding the word boundary is
       an alveolar /t, d, n, s, z, l/ and if the sound in word-initial position in the word
       immediately following the word boundary is a dental /θ, ð/ then the alveolar is
       dentalized.’
       . This process of allophonic assimilation can be summarized as
       Word-final alveolars are dentalized before dental fricatives.
       Some further examples should make this clear:
can they?                  /kæn ðeɪ/                   →         [k̟ʰæ̃n̪ ðeɪː]
this thing                 /ðɪs θɪŋ/                   →         [ðɪs̪ θɪŋ̃ ]
those things                /ðəʊz θɪŋz/                  →          [ðəʊːz̪̪̊ θɪŋ̃ z̥]
all those                   /ɔɫ ðəʊz/                    →          [ɔːɫ̪ ðəʊːz̥]
       As well as dentalization, the assimilatory process of labialization can also occur across
       word boundaries.
       2. Allophonic assimilation – labialization
       Within words, the bilabial nasal /m/ or the alveolar nasal /n/ appears before a
       labiodental fricative /f, v/ they are likely to assimilate the labiodental place of
       articulation, e.g.
Comfort                     /kʌmfət/                     →          [k̠ʰʌ̃ɱfətʰ]
       A similar process of allophonic assimilation can also across a word boundary. This
       pattern is summarized in the following .
       Rule ‘If the sound in word-final position of the word preceding the word boundary is a
       nasal /m, n, ŋ/ and if the sound in word-initial position in the word immediately
       following the word boundary is a labiodental fricative /f, v/ then the nasal is labialized.’
       We can, therefore, summarize this rule with the following statement.
       Word-final nasals are labialized before labiodental fricatives.
       Examples of this rule in operation include the following.
home visit                  /həʊm vɪzɪt/                 →          [həʊ̃ːɱvɪzɪtʰ]
born free                   /bɔn fri/                    →          [bɔ̃ːɱfɹ̥ ɪ]
       The assimilation of a bilabial or alveolar nasal before a labiodental fricative is
       reasonably common in connected speech. The assimilation of the velar nasal to a
       labiodental nasal in this context is, perhaps, less common but it is still possible to find
       examples in rapidly articulated speech, e.g.
ring fence                  /rɪŋ fɛns/                   →          [ɹɪɱ
                                                                       ̃ fɛñ s]
       Another type of labialization occurs when labiodental fricatives occur in word-final
       position before a bilabial plosive, i.e.
       The IPA (the International Phonetic Alphabet) symbol [ɸ] is used to represent the
       voiceless bilabial fricative, and [β] to represent the voiced bilabial fricative. These two
       phones have no contrastive function in English and they are, therefore, allophones of
       the corresponding phonemes /f/ and /v/. Examples of this process in action include the
       following.
life peer                   /laɪf pɪə/                   →          [laɪɸ pʰɪəː]
live band                   /laɪv bænd/                  →          [laɪːβ bæ̃nd̥]
       Interestingly, this process does not occur within words in English but only across word
       boundaries in rapidly articulated connected speech.
       We can state this rule as follows.
       Word-final labiodental fricatives are labialized before bilabial plosives.
          3. Allophonic assimilation – de-voicing of liquids
          We know that the liquids /r, l/ are prone to being de-voiced in single words when they
          occur after voiceless consonants. This process can also function across a word boundary
          in connected speech. So, if a word-final voiceless consonant precedes a word-initial
          liquid across a word boundary, the liquid will be de-voiced, i.e..
          Consider the following.
Miss Wright                    /mɪs raɪt/
          The sound immediately preceding the word boundary is the word-final voiceless
          consonant /s/ of the first word Miss. This is followed immediately after the word
          boundary by the word-initial liquid consonant /r/ of the second word Wright. In this
          instance, it is the sound before the word boundary that affects the sound following the
          word boundary. Consequently, it is the liquid /r/ that assimilates the voicing of the
          preceding voiceless consonant, becoming de-voiced, i.e.
Miss Wright                    /mɪs raɪt/                   →          [mɪs̃ ɹ̥ aɪːʔ]
          We see, therefore, that this is an example of so-called left-to-right assimilation, as the
          sound to the left of the word boundary affects the sound to the right of the word
          boundary. Further examples of left-to-right de-voicing in which a voiceless consonant
          precedes a liquid include the following.
hot love                       /hɒt lʌv/                    →          [hɒt̪̚ l̥ ʌv̥]
hip length                     /hɪp lɛŋθ/                   →          [hɪpʰ l̥ ɛ̃ŋθ]
bike race                      /baɪk reɪs/                  →          [baɪkʰ ɹ̥ eɪs]
          In each of these examples it is the liquid sound, either /l/ or /r/, in word-initial position
          immediately following the word boundary, that is assimilated. In each case the liquid
          assumes the voicing of the preceding voiceless consonant and is de-voiced. We can state
          this rule as:
          Rule Word-initial /l, r/ are de-voiced after voiceless consonants.
          4. Allophonic assimilation – disapplication
          Velarization
          We have seen that /l/ is velarized (dark l) if it occurs in word-final position, e.g.
Kill                           /kɪl/                        →          [k̟ʰɪɫ]
          However, in connected speech, a word-final /l/ will not be velarized if it is followed
          across a word boundary by a word that has a vowel in word-initial position, e.g.
kill it                        /kɪl ɪt/                     →          [k̟ʰɪl ɪtʰ]
          In this environment, the rule that a word-final /l/ is velarized is disapplied
          (disapplication) , i.e. the /l/ is not velarized.
          .
          Further examples include:
fall over                   /fɔl əʊvə/                   →         [fɔːl əʊːvə↓]
pull on                     /pʊl ɒn/                     →         [pʰʊl ɒ̃n]
call anyone                 /kɔl ɛnɪwʌn/                 →         [k̠ʰɔːl ɛñ ɪw
                                                                                ̃ ʌ̃n]
        The rule governing this pattern can be stated as:
        Word-final /l/ is not velarized before a word-initial vowel.
        Word-final de-voicing.We know that word-final voiced plosives, fricatives and
        affricates are prone to de-voicing when they appear in words spoken in isolation, e.g.
Lab                         /læb/                        →         [læb̥]
Cave                        /keɪv/                       →         [kʰeɪːv̥]
Badge                       /bæʤ/                        →         [bæʤ̪̊]
        However, in connected speech, when they appear word-finally before a word-initial
        vowel or voiced consonant across a word boundary they are not voiced, i.e.
        The rule, is as follows.
        Word-final voiced plosives, fricatives and affricates are not de-voiced before a
        word-initial vowel or voiced consonant.
        A few examples should make this clear:
lab window                  /læb wɪndəʊ/                 →         [læb̪̚ wɪñ dəʊː]
cave drawing                /keɪv drɔɪŋ/                 →         [kʰeɪːv d̠ɹɔːɪŋ̃ ]
badge of office             /bæʤ əv ɒfɪs/                →         [bæʤ əv ɒfɪs]
        In the last example above – badge of office – there are two examples of a voiced fricative
        in word-final position preceding a word-initial vowel. The first occurrence is at the word
        boundary between the words badge and of. The word badge has a word-final voiced /ʤ/
        that precedes the word-initial vowel of the immediately following word of. Hence, the
        word-final voiced /ʤ/ is not de-voiced. The second occurrence is at the word boundary
        between the words of and office. The word of has a voiced /v/ in word-final position.
        This precedes the word-initial vowel /ɒ/ of the word office. In this environment,
        therefore, the /v/ is not de-voiced.
        Of course, if word-final voiced plosives, fricatives and affricates occur before a
        voiceless consonant in word-initial position they will be de-voiced, e.g.
lab coat                    /læb kəʊt/                   →         [læb̥ kʰəʊtʰ]
cave painting               /keɪv peɪntɪŋ/               →         [kʰeɪːv̥ peɪː̃ ntʰɪŋ̃ ]
badge pin                   /bæʤ pɪn/                    →         [bæʤ̪̊ pɪñ ]
        There are several other types of allophonic assimilation that affect pronunciation across
        word boundaries. However, the foregone discussion should be sufficient to give you a
        feel for the operation of this process. We will now turn our attention to the second type
        of assimilation, phonemic assimilation.