Reviewed by Jason D. Haugen, Oberlin College

In this monograph, Roland Pfau aims to test the psychological plausibility of distributed morphology (DM) by analyzing its theoretical architecture from a mentalistic perspective. P’s essential question is whether DM is amenable to integration with psycholinguistic models of language production, and he approaches the issue from a novel source of linguistic data: spontaneous speech errors in language use (primarily from German and English). P ultimately concludes that the DM model not only accounts for the speech error evidence but also makes correct predictions about possible and impossible speech errors—for example, the processing of morphosyntactic features (e.g. gender, number) playing a role in language production but categorial information (e.g. noun, verb, adjective) not.

A crucial element of DM that makes it compatible with multilevel processing models is the separation of morphosyntactic processes from morphophonological ones. The book begins with an introduction, a review of mentalistic approaches to grammar, and an extensive theoretical background discussion on psycholinguistic models and DM. P proceeds by addressing different speech error types along the time-course assumed in the models: ‘…from semantic planning and selection of items…(Chapter 4) via grammatical encoding and manipulation of morphosyntactic features (Chapter 5) towards morphological processes, Vocabulary insertion, and phonological readjustment (Chapter 6)’ (81).

With respect to semantic planning and selection (Ch. 4), P considers semantic substitutions, where a semantic competitor is inserted into a syntactic slot intended for a different root (e.g. Radiergummi ‘eraser’ ← Spitzer ‘pencil sharpener’), as well as semantic anticipations and perseverations, where a meaning-related concept is activated by a target and is inserted into another slot in a sentence (e.g. They even fly on the wing ← sleep on the wing). P concludes that such evidence requires a pre-syntactic conceptual level, where roots must be attached to some kind of conceptual content (in parallel with the lemma level in psycholinguistic production models), which contrasts with most DM research, which has maintained that root content is irrelevant to the syntax.

In Ch. 5, P convincingly makes the case that particular morphosyntactic features in language production are manipulable (as per DM), and as such they can be specifically implicated in speech errors. Spontaneous speech errors related to various types of mismatch in feature copying are examined (e.g. subject-verb agreement, pronoun agreement). In general, errant triggers are usually more proximate (at deep structure or surface structure) than the actual targets, although long-distance (across clause boundaries) agreement errors also occur. Other types of errors relating to morphosyntactic features include slips (stranding features in their base position) and shifts (exchanges, perseverations, or anticipations of features).

Although he introduces and illustrates a descriptive four-way typology for accommodation types (phonological, morphophonological, morphological, and morphosyntactic), in Ch. 6 P contends that the notion of accommodation as a psycholinguistic process involving repair operations in speech production is theoretically superfluous. He proposes instead to account for morphology-related speech errors by way of an appeal to DM-based morphological operations such as feature copy, licensing, morpheme insertion, and phonological readjustment.

In sum, this book provides much food for thought worthy of careful digestion by linguists who desire psycholinguistic compatibility for their grammatical theories. It would be very interesting to see how competing morphological theories would account for P’s speech error data utilizing alternative theoretical machinery. I have little doubt that P’s thought-provoking work will be instigating very lively debates in the field.