The Atoms of Phonological Representations

Phonology describes spoken language as a combinatorial system typically comprising 30–40 basic units. The smallest known inventories make use of about 10 contrastive units, the largest no more than 150 (Maddieson 1984). Phonological theories capture, on the one hand, the nature of these basic units themselves (the atoms of speech production and perception) and, on the other hand, the language-specific patterns that the units show when they combine into larger structures. One of the key problems in relating this combinatorial system to the phonetic, physical side of speaking lies in the spatio-temporal structure of speech, which does not seem to mirror the units of linguistic analysis. When we inspect instrumental records of speech, the discrete combinatorial system that underlies these records is by no means obvious. We see continuously changing, highly context-dependent articulator motion and spectral energy patterns, with no clear boundaries between the individual sounds. The discrepancy between the symbolic units used to capture grammatical regularities and the dynamic complexities of speech as a motor act is so fundamental that some researchers have adopted the view that phonology and phonetics best be studied as separate sciences, since it seems difficult to see how the insights of each discipline could be relevant for the other. Current phonological and phonetic theories are, however, informed by the insight that phonetics and phonology are intertwined to such an extent that one cannot be understood without the other. How the relationship between phonetics and phonology can be negotiated has therefore been one of the key research topics in both disciplines (Pierrehumbert 1990; Boersma 1998; Flemming 2001; Hume and Johnson 2001; Keating et al. 2003; Hayes et al. 2004; Prince and Smolensky 2004; to name but a few); see also Kingston and Beckman (1990) and subsequent Laboratory Phonology volumes. Keywords: phonology; formal grammars