This book is a sequel to Feigenbaum & Kurzon 2002, which had a similar but somewhat more parochial title. The terminological change from 'prepositions' to 'adpositions' signals an ambition to 'broaden the perspectives' to include also postpositions and the less common circumpositions. Although the editors do not define the book as a contribution to language typology, it appears in a series of typological studies, and therefore it may not be out of place to say that the perspective is still a bit narrower than could be expected from a book in that series. Most of the chapters focus on one single Eurasian language. In only one of the chapters is there an attempt at a larger sample of languages, and even in that case, the languages are overwhelmingly from Eurasia. In addition, the topics addressed in many of the chapters are quite narrow. In some, adpositions even play a rather indirect role. I am afraid that a reader who expects to get a general view of adpositions as a word class and their place in the grammars of human languages will be somewhat disappointed (cf. a very similar remark on Feigenbaum & Kurzon 2002 in Lander 2004).Several of the chapters, however, do discuss issues of considerable general interest.
Noting the difficulty of assigning the English word ago to any traditional part of speech, DENNIS KURZON (209–27) looks at a sample of twenty-six languages from different language families. He finds that there are clear examples of both adpositions and adverbs in the function of ago and concludes that the postpositional analysis of English ago is acceptable, although an analysis as a 'less prototypical adverb' is also possible. Kurzon starts the paper by saying that 'it may be assumed' that all languages have something corresponding to English ago. I would submit that this should be demonstrated rather than assumed, given that ago constructions are typically used with numerical expressions such as two years and may be of little use in languages without numerals or with very restricted numeral systems. On the whole, it may be observed that words such as ago tend not to belong to any natural classes of expressions. It is not obvious that it makes much sense to try to make them fit into traditional word classes.
Christopher Wilhelm (289–300) discusses the grammaticalization of the Classical Armenian preposition z- 'concerning, around' into a prefix with characteristics of a case-marking morpheme. Although most of the Indo-European case system was still preserved in Classical Arme nian, the distinction between nominative and accusative was neutralized in nouns. The preposition z-, which was always proclitic to the following word, was instead used as a marker of definite direct objects. What is of particular interest is that z- sometimes appears on both modifiers and head nouns in noun [End Page 448] phrases, which shows how NP-internal agreement may arise. Mentioning Spanish, Romanian, and some Australian languages, Wilhelm notes that while postpositions frequently grammaticalize into case markers, this is 'far less frequently' found with prepositions. It is somewhat strange that Wilhelm does not relate object marking in Armenian to the general phenomenon of differential object marking, since it has been discussed extensively in typological work over the last decades and is found in geographically close languages such as Turkish and Persian. An atypical trait of Arme nian object marking is the source meaning of the grammaticalizing adposition, 'concerning, about'. Regrettably, Wilhelm does not give any details about the early historical development of z-.
Languages with complex case systems sometimes exhibit intricate patterns of case marking of noun phrases governed by adpositions. ALAN REED LIBERT (229–55) describes such systems in Turkic languages. The topic has been somewhat neglected in typology. The only general work I have found is an earlier paper by Libert, but curiously it is not referred to in his chapter. Essentially, Libert tries to survey all types of case systems that occur in Turkic languages, but the reader does not...