Connected Speech and Coarticulation (27-Nov-2003)

Speech is a stream of sounds rather than a series of discrete segments and in many cases the sounds merge or change, sometimes to anticipate the next sound and sometimes by being affected by a previous sound.

Assimilation

Assimilation refers to the effect where one sound segment is modified by its neighbour, and causes one sound becoming phonetically similar to the other. E.g. in English, in the sentence "I have to", the "v" of "have" is devoiced and is pronounced as an "f" since the first sound of the next segment is devoiced. When one sound changes to become more like the next, this is "regressive" assimilation; when one sound changes to become more like the preceding sound, this is "progressive" assimilation. Crystal also notes "coalescent", where two sounds mutually influence one another. Examples (from Crystal p166): Different languages exhibit different assimilatory rules. For example, in French, "avec vu" is is articulated "aveg vu", with the "c" beooming voiced. In Engish, anticipatory voicing does not happen so frequently, which is one of the reasons why a French speaker of English may have a noticeable "accent" - if he applies this kind of assimilation it will sound unusual to a native speaker.

Assimilation can be of different kinds:

Note that the assimilation process appears to be resulting in different phonemes being used, and so this ought to change the meaning of the words - why it doesn't is something that will be covered another day. However these assimilation rules are part of the phonology of the language, and so to a native speaker they are just a natural part of speech.

Assimilation might be seen as "lazy" articulation, at any rate it seems to be a result of trying to minimize the effort required.

Coarticulation

Coarticulation is related to assimilation; Roach says it's not easy to define. It is to do with the fact that when speaking, many different articulatory processes and components are involved, and these have different characteristics (e.g. tongue can move faster than vocal folds), and do not move discretely from one position to another, which means that certain aspects of articulation can be blurred. Because it's more a physical process, it's not something that can so easily be controlled, especially when speaking quickly. And because it's a physical process, it's not a characteristic of a language's phonology.

Anticipatory coarticulation occurs when one sound is affected by the following sound; perseverative coarticulation is when one sound is affected by the previous sound. In the word "screws", the /s/ and /z/ are articulated with rounded lips, but in "sees", they're not.

Elision

Elision is the phenomenon where sound segments are omitted completely, e.g. (from Crystal p166), "mashed potatoes" loses the "-ed" of "mashed". This may be a kind of coarticulation: electropalatography can show that in saying these words a speaker may, for example, move the tongue almost to the point where a sound would be articulated, without quite managing it: in other words, the sound is not deliberately omitted, it's just not given time to be made.

Useful book readings for this lecture:


Sounds, Grammar and Meaning page