Laryngeal Cooccurrence Restrictions (review)

Margaret MacEachern’s book is concerned with the cooccurrence restrictions of consonants with laryngeal specifications (voiceless aspirates, voiced aspirates, ejectives, implosives, h, and ʔ) within a certain domain, often the root or the word. The main goal, as described by M, is to show that the restrictions reflect prohibitions on the cooccurrence of similar segments, and that the crosslinguistic variation results from the different thresholds of acceptable similarity set by different languages.

After outlining the typological patterns and presenting the gist of the analysis (Chs. 1 and 2), M discusses in detail the laryngeal cooccurrence restrictions in eleven languages (Ch. 3). There are two crucial observations: (1) Laryngeal cooccurrence restrictions seem to be subject to an implicational hierarchy based on similarity, for example [t^h] is more similar to [t’] than to [h]; and for all languages that ban the cooccurrence of [t^h] and [h], the cooccurrence of [t^h] and [t’] is also banned; (2) Identical segments behave differently from merely similar ones: They are sometimes allowed in languages that ban similar segments.

M argues against earlier feature-geometric analyses of cooccurrence restrictions in Ch. 4 and provides her optimality-theoretic analysis in Chs. 5 and 6. The restriction patterns derive from the interplay between featural faithfulness constraints (Max-IO [feature], Dep-IO [feature]) and a set of OCP constraints and their conjunctions. The legality of identical segments is accounted for by a constraint requiring consonants within a domain to be identical (BeIdentical). The reader is likely to be alarmed at this point as, apparently, nothing in the proposed system ensures that the coveted implicational hierarchy will emerge.

M takes up this point in Ch. 7 in a surprising way. Upon observing that the factorial typology of the proposed constraints predicts unattested patterns, M concludes that the implicational hierarchy does not arise from the synchronic grammar but from acoustic salience and the nature of language acquisition. M then discusses the acoustic similarity among consonants with laryngeal specifications and suggests, with reservation, that cooccurrence restrictions of similar segments might have originated from dissimilatory sound changes.

As one of the first works to identify implicational hierarchies in cooccurrence restrictions and to consider their functional basis, this is a very valuable book. It is also one of the first to explicitly analyze the segmental identity effect, and it has spawned some recent works on similar issues (e.g. Kie Zuraw, ‘Aggressive reduplication’, Paper presented at the annual meeting of the LSA, Chicago, 2000; Sharon Rose and Rachel Walker, ‘Consonant agreement at a distance’, Paper presented at the North East Linguistic Society, Washington, DC, 2000.) But I find M’s stance on the irrelevance of similarity to synchronic grammar unfortunate. Given this stance, we now need an explicit model for either the diachronic change or the learning mechanism that will give us the crosslinguistic regularity in laryngeal cooccurrence restriction patterns, and M does not deliver this. I also doubt that the origin of cooccurrence restrictions is dissimilation. It seems more likely that BeIdentical is what drives the system to avoid similar but nonidentical segments, as envisioned by Zuraw, Rose, and Walker. But this alternative requires similarity to play a role in the synchronic system and is thus unavailable to M.