The sigmatic aorist in Indo-European: Evidence for the space-time hypothesis By Bridget Drtnka (review)

232 LANGUAGE, VOLUME 74, NUMBER 1 (1998) linguistics' (339-67) by Kaplan addresses errors researchers make in relying on intuitions from descriptive theories, computational work, and psycholinguistics , namely procedural, substance, and interaction seductions. [Kim Honeyford, University of Waterloo.] The sigmatic aorist in Indo-European: Evidence for the space-time hypothesis. By Bridget Drtnka. (Journal of IndoEuropean Studies monographs, 13). Washington: Institute for the Study of Man, 1995. Pp. 154. The space-time hypothesis (i.e. the possibility of making an in-depth description of Indo-European with chronological and dialectal differences) has gained wide acceptance among Indo-Europeanists in the last years, and with her book Bridget Drinka adds valuable evidence for testing its validity. As she explains in the introductory chapter (1-7), handbooks and grammars usually reconstruct an s-aorist for the proto-language on the basis of Greek and Indo-Iranian ¿-aorists, Latin ¿-perfects, and some ¿-preterites of other Indo-European languages. D sets out to prove that the ¿-aorist is an innovation rather than an archaism, and I believe that she achieves her goal. In the introduction she also provides a general sketch of the space-time hypothesis. Just for the sake of accuracy, I would like to point out that although it is customary now to attribute the paternity of that theory to Wolfgang Meid, that is not the way the story really goes (see Francisco R. Adrados, "The new image of Indo-European', Indogermanische Forschungen 97.1-28, 1992). The core ofthe book consists ofa series ofchapters devoted to the analysis ofthe ¿-preterites ofparticular Indo-European languages: Indo-Iranian (8-33); Old Church Slavonic (34-48); Latin (49-92); Greek (93-1 19); and Hittite, Tocharian, Armenian, and Old Irish (120-34). She demonstrates that the lengthened grade of the root, usually posited as one of the characteristics of those ¿-constructions, is not an inherited feature from the protolanguage but, rather, an independent innovation of Old Indian, Old Church Slavonic, and Latin as a result of phenomena of compensatory lengthening in conditioning environments which are different from one language to the other. The chapter on Latin in especially interesting: starting from a reanalysis of ¿-forms, D sketches a global history of the Latin perfect which no doubt deserves attention. She also reformulates anew the conditions in which the so-called Lachmann' s law operates: The lengthening shown by some Latin perfects would be due to the phonetic loss of -«- and -d- markers, analogically extended from the present to the perfect. As for Celtic, D concentrates exclusively on Old Irish, but a short discussion of Continental Celtic ¿formations such as Gaulish legasit, readdas, or prinas (see Pierre-Yves Lambert, La langue gauloise, Paris: Errance, 1994) and Celtiberian ampitiseti would not be out of place. To those one could add now Celtiberian ausanto, attested in the third inscription from Botorrita, recently edited by Francisco BeItran , Javier de Hoz, and Jürgen Untermann (El tercer bronce de Botorrita (Contrebia Belaisca), Zaragoza: Diputación General de Aragón, 1996). Anyway, she argues convincingly that Irish f-preterites cannot continue the Indo-European ¿-aorist. The final chapter (135-44) presents the general conclusions ofD's study concerning the nonantiquity of the ¿-aorist, its vocalism, and its limited areal diffusion—just Greek and Indo-Iranian—with aspectual value. [Eugenio R. Lujan, Universidad Complutense de Madrid.] The possibility oflanguage: A discussion of the nature of language, with implications for human and machine translation . By Alan K. Melby with C. Terry Warner. (Benjamins translation library , 14). Amsterdam: John Benjamins , 1995. Pp. xxvi, 274. In this book, Melby argues in favor of three hypotheses : • Current techniques in machine translation cannot be extended so as to achieve fully automatic highquality translation of unrestricted text. • Domain-specific language differs in essential aspects from general language. • Generative grammar does not and cannot cover general language. The first claim is introduced by a very eclectic, very briefoverview ofthe history ofmachine translation . It is observed that either working systems are based on controlled language or they require postediting . The justification for the claim is closely linked to the second hypothesis. M argues that there are two quite different entities, lexicographical units and terminological units. Terminological units have a...