Reviewed by: Abstract Phonology in a Concrete Model: Cognitive Linguistics and the Morphology-Phonology Interface Frank Y. Gladney Tore Nesset. Abstract Phonology in a Concrete Model: Cognitive Linguistics and the Morphology-Phonology Interface. Berlin: Mouton de Gruyter, 2008. pp. x, 250. [Cognitive Linguistics Research, 40.] The book is narrowly focused on stem alternations in the Russian verb, as in pisa-l ‘wrote’ ~ piš-et ‘writes’. Nesset’s central claim is that /a/ ~ /Ø/ truncation and /s/ ~ /š/ softening constitute a conspiracy which has the meaning here of [PRESENT]. His broader purpose is to argue that Cognitive Grammar (CG) accounts for Russian verbal morphophonemics better than other frameworks. There are 12 chapters. Chapter 1 states the book’s main argument, that CG accommodates the “morphology-phonology interface” and that morphophonemic alternations are meaningful. Chapter 2 presents the CG linguist’s “toolbox” (represented mainly as boxes and arrows). Chapter 3 is about the CG approach to Russian phonemes and allophones. Chapter 4 presents verbal stem alternations. Chapter 5 claims that CG provides a synthesis of the phonologically based one-stem description and the semantically based two-stem description. Chapter 6 treats infinitives, the straightforward pisa-t’, the morphophonemically more complex gres-ti (~ greb-u), and the hard-to-segment peč’ (~ pek-u). Chapter 7 treats past-tense forms, the problem of soxla (~ soxnut’) losing -nu-, of vel (~ vedu) losing -d-, and of nes (~ nesla) losing -/. Chapter 8 is on the imperative and the problem of opacity, igrajte being opaque because the j doesn’t truncate before a consonant and bros’ because suffixal -i- does truncate even though the conditioning vowel ending does not appear. Chapter 9 is on palatalization and lenition. Kupit’ ~ kuplju is said to show lenition because /pl’/ ends in a continuant and prjatat’ ~ prjaču because [č’] “does not have a complete closure” (174). Chapter 10 is about opacity and non-modularity. Pišet is opaque because the suffix in pisa- which conditions the transitive softening of š is absent from this form. Non-modularity refers to the claim that CG handles forms like pišet effectively because it does not require separate morphology and phonology modules. Chapter 11 is titled “The [End Page 311] meaning of alternations: The truncation-softening conspiracy.” Chapter 12 is concluding remarks. The key principle of CG is the so-called Content Requirement, namely, that language consists solely in sound, meaning, and their symbolic association. Grammar is the network of schemas and their elaborations (or extensions). At the semantic pole, [DOG], for example, is schematic for its elaboration [POODLE] as well as for its metaphorical extension to [INFERIOR PRODUCT], [UNATTRACTIVE PERSON], etc., while itself being an elaboration of [ANIMAL], which in turn is an elaboration of [CREATURE], ultimately of [THING]. Proceeding downward into specificity, poodle can be more narrowly categorized as my poodle, my sister’s poodle, and the poodle that kept me awake last night by barking. In this way words categorize phrases, which categorize clauses, which categorize complex sentences and even discourses. At the phonetic pole, phonemes are schematic for allophones, while allophones are schematic for their various realizations. An important postulate of CG is that everything above the phoneme level is meaningful, including, in the above examples, ‘s, the, that, by, and -ing. This may be what underlies the author’s claim that morphophonemic alternations also have meaning. CG has been most fully worked out for expressions at the word level and higher. In the most recent exposition, Langacker 2008, sublexical matters are dealt with on only a few of its 540 pages. This puts Nesset at a slight disadvantage in applying CG to Russian verbal morphophonemics. An example of how CG handles verbal morphology is provided by the forms seeded and sat. Langacker (1987: 340) presents each as an elaboration of the schema [VERB-PAST]. Seed-ed elaborates it straightforwardly, sat less so. How to handle the [PAST] component of sat is a problem that has been around for a while. Chomsky and Halle (1968: 11) regard [PAST] as an abstract morpheme which is sometimes realized as -ed and sometimes deleted in favor of a readjustment rule that effects ablaut changes like /i/ /æ/. But abstract morphemes and syntactic features are ruled out in CG...