Window model

Definition
The window model of co-articulation by Keating (1988) divides segments into those that are either positively specified for a certain feature (+), or negatively specified for a certain feature (-) or neutrally specified for that feature (0). The articulators move from one segment to the next, but their movements are constrained by time windows. These windows are narrow (along the time axis and/or in height) in case the segment is specified + or - for a particular feature, and broad in case the segment is specified neutrally. This means that + or - specified segments do not allow much variation in the movements of the particular articulator, whereas the movements are hardly constrained for neutral or underspecified segments.

This can be exemplified by the degree of nasalisation on French and English vowels preceding a nasal consonant. English does not have nasal vowels, and a vowel preceding a nasal consonant will therefore be neutrally specified for nasal. This translates into a broad window in which anticipatory nasalisation may occur early on in the vowel. French, on the other hand, does have nasal vowels, and to avoid confusion, nasalisation should not be perceptible too early on in the vowel. Thus, for the vowel in the French word 'haine', the window for co-articulation is much narrower than for the vowel in the English word 'pen'.

Links

 * Utrecht Lexicon of Linguistics