Reviewed by: Grammatical change and linguistic theory: The Rosendal papers Michele Loporcaro Grammatical change and linguistic theory: The Rosendal papers. Ed. by Thórhallur Eythórsson. (Linguistik aktuell/Linguistics today 113.) Amsterdam: John Benjamins, 2008. Pp. 441. ISBN 9789027233776. $180 (Hb). The volume under review comprises the proceedings of two symposia held at Rosendal and Lysebu, Norway, in 2005. The contributors are specialists in diachronic syntax and/or morphology of various theoretical persuasions working on different languages. The structure of the contributions is diverse too. HENNING ANDERSEN, in 'Grammaticalization in a speaker-oriented theory of change', summarizes his own theorizing on grammar change. He subdivides grammatical change into grammation, regrammation, and degrammation, distinguishing it from innovation, classified into neologism, extension, adoption, and reanalysis. Andersen objects to functionalist grammaticalization theory because it 'has been divorced from the reality of language transmission among speakers' (11). He also objects to generative approaches to grammaticalization, as exemplified by van Gelderen 2004 and Roberts & Roussou 2003: the former is criticized for endorsing the 'perennially attractive idea…that speakers prefer their language economical' (16), whereas the latter is because their theory 'ignores the observable fact that in every language the new coexists with the old in the form of synchronic variation' (16). John Ole Askedal's '"Degrammaticalization" versus typology: Reflections on a strained relationship' focuses on the unidirectionality principle, refuting the eight alleged examples of attested 'antigrammaticalization' pointed to by Haspelmath (2004:29). Some of them are better described as function splits (such as Greek ksana 'again', originally a bound prefix that gave rise to a homophonous adverb), while others (like Irish -m(u)id '1PL subject affix' > muid 'we', as in molaimid > molann muid 'we praise') are not changes per se, but rather the 'corollary of a typological change from synthetic to analytic structure of an individual construction or of the overall language system' (71). In 'Cascading parameter changes: Internally-driven change in Middle and Early Modern English', THERESA BIBERAUER and IAN ROBERTS discuss the much-investigated reorganization of the English verb placement and auxiliary system through a series of parametric changes that happened between the twelfth and the seventeenth centuries. These are claimed to stand in a cascading relation, with each change causing the next. The whole cascade, it is maintained, was initiated by extrasyntactic factors, since 'syntax, by itself, is diachronically completely inert' (Longobardi 2001:277–78). The first change in the series (loss of VP-to-SpecvP movement) might have been triggered by replacement of earlier verb-particle constructions through French borrowings (90, 106). No examples of this are provided, however, as the paper focuses more on structural representations than on data. [End Page 219] In 'The rise and development of analytic perfects in Italo-Romance', MICHELA CENNAMO locates the rise of Romance perfective auxiliaries in a transition stage in which not only the inflectional exponents of formerly distinct morphosyntactic categories (say, nominative vs. accusative) were blurred, but, more radically, 'a disruption of grammatical voice and the concomitant loss of a firm notion of grammatical relations' (121) took place. The claim is substantiated with examples like that in 1 (122, from Anthimos's De observatione ciborum, 511–533 AD). (1) ficum contundito usque dum minutum fiat (Anthim. 890) fig.ACC cut.2PL.IMP until small.ACC become.PRES.IND.3SG 'Cut the fig until it is reduced to small pieces.' Incorrect glossing in this example (the imperative is a future 2SG form; fiat is subjunctive) has an impact on interpretation: in fact, to demonstrate that an 'accusative in subject function' occurs in this passage, one would need to make sure that ficus 'fig' is here a 4th or 2nd declension masculine (a more seldom variant with respect to the feminine, current in the classic language), since only in that circumstance would accusative ficum contrast with nominative ficus. But the 'fig' word also inflects as a 2nd declension neuter (ficum, nominative = accusative) in Late Latin, attested especially in medical texts, like Caelius Aurelianus's, fifth century (cf. ThLL 6.651). Since this variable is not controlled, the example is inconclusive, as are several others, so the evidence is not as...