assimilation (n.)

A general term in PHONETICS which refers to the influence exercised by one sound segment upon the ARTICULATION of another, so that the sounds become more alike, or identical. The study of assimilation (and its opposite, DISSIMILATION) has been an important part of HISTORICAL LINGUISTIC study, but it has been a much neglected aspect of SYNCHRONIC speech analysis, owing to the traditional manner of viewing speech as a sequence of DISCRETE WORDS. If one imagines speech to be spoken ‘a word at a time’, with PAUSES corresponding to the spaces of the written language, there is little chance that the assimilations (or assimilatory processes) and other features of CONNECTED SPEECH will be noticed. When passages of natural conversation came to be analyzed, however, assimilation emerged as being one of the main means whereby fluency and RHYTHM are maintained.

Several types of assimilation can be recognized. It may be partial or total. In the phrase ten bikes, for example, the normal form in colloquial speech would be /tem baIks/, not /ten ba}ks/, which would sound somewhat ‘careful’. In this case, the assimilation has been partial: the /n/ has fallen under the influence of the following /b/, and has adopted its BILABIALITY, becoming /m/. It has not, however, adopted its PLOSIVENESS. The phrase /teb baIks/ would be likely only if one had a severe cold! The assimilation is total in ten mice /tem maIs/, where the /n/ is now identical with the /m/ which influenced it.

Another classification is in terms of whether the change of sound involved is the result of the influence of an adjacent sound or of one further away. The common type is the former, as illustrated above: this is known as contiguous or contact assimilation. An example of the opposite, non-contiguous or distance assimilation, occurs in turn up trumps, where the /-n/ of turn may be articulated as /-m/ under the influence of later sounds. It also occurs in languages displaying VOWEL harmony, where a vowel in one part of a WORD may influence other vowels to be articulated similarly, even though there may be other sounds between them.

A further classification is in terms of the direction in which the assimilation works. There are three possibilities: (a) regressive (or anticipatory) assimilation: the sound changes because of the influence of the following sound, e.g. ten bikes above: this is particularly common in English in ALVEOLAR consonants in word-FINAL position; (b) progressive assimilation: the sound changes because of the influence of the preceding sound, e.g. lunch score articulated with the s- becoming , under the influence of the preceding -ch; but these assimilations are less common; (c) coalescent (or reciprocal) assimilation: there is mutual influence, or FUSION, of the sounds upon each other, as when don’t you is pronounced as  – the t and the y have fused to produce an AFFRICATE. In standard GENERATIVE PHONOLOGY, assimilation is characterized through the notion of FEATURE COPYING: SEGMENTS copy feature specifications from neighboring segments. In NON-LINEAR models, a FEATURE or NODE belonging to one segment (the trigger) is viewed as SPREADING to a neighboring segment (the target). The assimilation is UNMARKED when a rule spreads only features not already specified in the target (a ‘feature-filling’ mode); if the rule applies to segments already specified for the spreading features (thereby replacing their original values), it is said to apply in a ‘feature-changing’ mode. Further types of assimilation can be recognized within this approach, based on the identity of the spreading node: if a ROOT node spreads, the target segment acquires all the features of the trigger (total or complete assimilation); if a lower-level class node spreads, the target acquires only some of the features of the trigger (partial or incomplete assimilation); and if only a TERMINAL feature spreads, just one feature is involved (single-feature assimilation).