676LANGUAGE, VOLUME 62, NUMBER 3 (1986) Apprehension: Das sprachliche Erfassen von Gegenständen. Teil I: Bereich und Ordnung der Phänomene. Edited by Hansjakob Seiler and Christian Lehmann . (Language universals series, 1:1.) Tübingen: Narr, 1982. Pp. x, 311. DM 96.00. Apprehension: Das sprachliche Erfassen von Gegenständen. Teil II: Die Techniken und ihr Zusammenhang in Einzelsprachen. Edited by Hansjakob Seiler and Josef Stachowiak. (Language universals series, 1:2.) Tübingen: Narr, 1982. Pp. iv, 342. DM 96.00. Reviewed by T. L. Markey, University ofMichigan Every empirical science necessarily has a taxonomic stage. Just as necessarily , it must build on that phase when it embarks on theory construction. The classificatory taxonomy of Carl von Linné (1707-78) was an essential prerequisite for Darwin's Origin of species (1859). Darwin, with his espousal of gradualism as a model of evolution, offered a program for all that Linné had merely labeled. This program has, in turn, been challenged by the non-gradualistic punctuationalism of Stephen Jay Gould. Linné's taxonomy, having been demoted as theoretically vacuous, could succeed only as an art form, as in the Jardin des Plantes where the great Bernard de Jussieu received the Swedish scientist. (Recall how our structuralists once 'graded' taxonomic phonemicizations in terms of their relative 'elegance'.) By and large, Darwin did not question taxonomy—but, much more significantly, the processes that had allowed it in the first place. In linguistics (a social science that has learned much from biological metaphor), taxonomy and its attendant extrapolation of typedefining universals (typology) have had long and distinguished careers, especially in historical studies of language. After all, these were the very issues that Rasmus Rask confronted when he wrote (1811-14) his prize essay on Icelandic. Given this foregrounding, we can see that the bulk of research to date on typology and universals—particularly that conducted by or inspired by Greenberg (as opposed to Chomsky's 'substantive universals', e.g. the 'A-over-A' principle, and his non-empirical autonomous grammar with its government and binding)—has been essentially taxonomic in scope. In the final analysis, this line of inquiry has been chiefly concerned with categorial universals and their type-defining contingencies, e.g. SOV languages and their contingent adj + N ordering; i.e., it is concerned with a word-order taxonomy which is, at bottom , epistemologically uninteresting. Alternatively, this line of inquiry has viewed processes or constructions as themselves categorial (such as ergative vs. non-ergative formations), or has been thrilled to find languages that complete a 'predicted' picture of classificatory speciation—e.g. the fact that, as an OVS language, Hixkaryana has finally supplied one of the six possible combinations of S, V, and O.1 1 A panoramic history of universals research has yet to be written; this is a major deficit. Some contextualization is, however, in order here. Greenberg, building on precursors like Carl Meinhof, entered the scene as a taxonomist; he classified African languages (1955), and word order was a convenient macro-feature for indexing genetic and allogènetic groupings. Word order was the theme REVIEWS677 The volumes reviewed here present part of an on-going effort to come to grips with the sorts of processual questions that troubled Darwin. These are research reports from the Cologne Universals and Typology (UNITYP) project, directed by H. Seiler.2 The principal thrust of this project is to uncover certain processual universals in a highly articulate fashion; then to determine their relative instantiation (in related and unrelated languages) by ranging them as implicational hierarchies along continua; and finally to specify directions of evolutive influence. UNITYP seeks a functionalist explanation of language processing, and it has employed the modalities of a discursive functionalism. It necessarily concerns itself with scaling, with operational dimensions as continua , with implicational hierarchies and their ordering (à la Louis Guttmann), and with turning points along continua—in fact, with the whole machinery of a 'geometricization' of lingual cognition as part of the human communicative code.3 UNITYP's aim is substantially more globally conceived than was Greenberg 's rather atomistic and piecemeal survey of individual strategies (e.g. negation , definiteness, numeral classifiers); and has emphasized langage (as opof his Dobbs Ferry conference...