Where do phonological features come from? Cognitive, physical and developmental bases of distinctive speech categories ed. by G. Nick Clements and Rachid Ridouane (review)

Rachid Ridouane writes in a brief preface that this volume is dedicated to the memory of his coeditor, G. Nick Clements, who died just as it was nearing completion. It is based on papers presented at the conference Where do Features Come From?, held in Paris in 2007, and supplemented with some additional papers. The volume is intended to give a state of the art of current research on feature theory, and in this it largely succeeds. What it reveals is a field advancing in different directions and employing diverse methodologies, but whose basic questions remain very much unresolved.

In their overview, Rachid Ridouane and G. Nick Clements list a series of questions about distinctive features that the papers in the volume address: How do they originate? How are they cognitively organized? How do they pattern in phonological systems? How are they extracted from the signal? How are they enhanced? What role do they play in language acquisition? Following this introduction, the volume is organized into the following parts: general and cognitive issues; acoustic and articulatory bases of features; extracting features from the signal; and features in phonological development.

The paper by Abigail C. Cohn is an interesting review of the role of distinctive features in generative phonology, and touches on most of the major themes of the volume. One of the questions she takes up is whether segments or features are phonological primitives. She considers four possible answers to this question: the primitives are (i) segments, (ii) features, (iii) neither, or (iv) both. After discussing the pros and cons of each of these answers, she concludes, somewhat surprisingly, that the correct answer is none of the above. This suggests that something may be amiss in the formulation of the question.Aplace to look may be the term ‘primitive’, which may be understood in different ways. One sense of ‘primitive’ is of a unit that cannot be decomposed further; in this sense, segments could not be primitives if they are analyzed into features. There is another notion in play, however: the question of whether the phonology can refer to segments as wholes, in which case segments could be primitives even if they can be decomposed into features. A third sense of ‘primitive’ refers to epistemological priority: do learners have features from the beginning (in which case they would be primitive in this third sense), or are they acquired from ‘finer-grained phonetic primes’ (36) (in which case features are not primitive in sense three, though they may come to be primitives of the acquired phonology in sense one)? Tied up with these issues are the notions of innateness and universality. Cohn’s conclusion that ‘the characterization of segments in generative phonology as combinations of universally-defined distinctive features is approximately, but not literally, correct’ (16) is thus a bit enigmatic, but perhaps accurately reflects current thinking.

Clements has argued that sound systems do not simply maximize the differences between segments, but are characterized by an economical use of features. Scott Mackie and Jeff Mielke present the results of the first large-scale study of feature economy, applying several different ways of measuring it in natural and synthetic inventories. They find that natural inventories are more economical than would be expected by chance; however, they also find that artificial vowel systems created without features are at least as economical as real ones, indicating that economy does not necessarily depend on features.

Björn Lindblom, Randy Diehl, Sang-Hoon Park, and Giampiero Salvi ask where feature economy, what they call ‘the re-use principle’, comes from. In a wide-ranging and inventive paper, they develop measures of...