The standard identification of the elementary particles of phonological analysis with distinctive features, inherited from the work of Trubetzkoy and Jakobson in the first half of the century, has recently been called into question. The alternative is the adoption as phonological primes of certain basic segments undivided, their combination giving the rise to more complex segments.
The starting point of all such holistic theories is the vowel space, which they divide by means of two axes, the vertical aperture axis, already referred to in connection with Lund Swedish vowel lowering, and the horizontal tonality parameter, which expresses front vs. back/round polarity. Holistic theories simply elevate the ends of these axes, namely I, U, and A (representing maximal palatality, labiality, and openness, respectively), to the rank of phonological primes. An analogy with colour will be apposite. In colour, red, blue and yellow (or perhaps green) are undecomposable, and thus the primary colours out of which all other colours are made up. Similarly with I, U, A for sound, according to holistic doctrine. Phonetically, these primes correspond to the three “equal” vowels of Stevens (1972, 1989). Stevens’s discovery was the correspondence of each of the three vowels in question with an area where both the acoustic effect of articulatory variation and the perceptual effect of acoustic changes are negligible. By contrast, the boundaries between the three areas are quite sharp, i.e. catastrophic in the technical sense. Such simple facts of physics would accordingly explain the cross-linguistic robustness of the canonical three-vowel system.
The differences between the various holistic approaches hinge on such aspects as the role of recursion and the calculus of element combination. Three perspectives on these matters stand out, which we shall refer to by the labels “Particle”, “Dependency” and “Government”, given to them by their respective practitioners. In traditional generative phonology, the primitive elements of segmental representation are the distinctive features—e.g. [high], [low], [back], [round]. This system of representation is atomistic, and (at least at the classificatory level) the elements are binary: segments are composed of unordered features, and differ from one other in the values (plus or minus) of those features. But in particle phonology, the primitive elements are called particles or components. The simplest of the three, larger vowel inventories are constructed by means of recursive association of the three basic elements (particles). For example, /i/ is made up of I, /e/ is made up of AI, /u/ is made up of IU and /o/ is made up of AIU.
Roca, Iggy. 1994. Generative Phonology. London: Routledge. Walsh, D.L.1997.Phonology Year Book 1:129-155.
1. Government Phonology