L'inversion dans la subordonnée en français contemporain By Kerstin Wall (review)

BOOK NOTICES 237 Recherches sur les constructions imbriqu ées relatives et interrogatives en français. By Juhani Härmä. (Annales Academiae Scientiarum Fennicae , Dissertationes humanarum litterarum, 20.) Helsinki: Suomalainen Tiedeakatemia, 1979. Pp. 303. FM 60.00. This book deals, first synchronically, then diachronically, with cases of double embeddings in French, where the first level is a relative clause or a question, and the second level an object clause. Examples are la maison que je crois que Marie habite 'the house I think Mary lives in' ; Quepenses-tu que nous devrionsfaire? 'What do you think we should do?'; Dis-moi ce que tu veux que nous fassions 'Tell me what you want us to do.' These constructions appear only with verbs of 'propositional attitude' (croire, penser), declaration (dire), perception (voir, sentir), wish (vouloir), or appreciation (regretter , être content). In these constructions, the first connector's function is not in the clause it directly introduces, but in the second embedded clause; the second connector is merely a 'void' complementizer. H first reviews the position of traditional grammarians on this question, then accounts for the phenomenon in terms of WH-movement. Then he introduces a set of constraints on that transformation (he examines works by Ross, Chomsky, Cattell, and Horn), more or less covering most of the unacceptable sentences. But H's main concern is the related 'construction subjective', apparently composed of a relative clause embedded in another relative clause: L'homme que je vois qui vient est mon voisin 'The man I see coming is my neighbor.' However , by analogy to the preceding structures, one would expect "L'homme qui je vois que vient, if it is true that the first connector bears the grammatical function and the second connector is only a link. This relative clause is in fact an object clause. The surface form qui may be explained by a generalization of a rather adhoc transformation originally proposed by Gross and Kayne for attributive relative clauses; qui appears instead of que by a rule which seems specific to French, stating that a finite clause must have an expressed subject. In fact, the transformation has a diachronic background. Using an enormous corpus of Old and Middle French, H shows that, from the 12th century on, wavering occurred among three constructions, que ... que, que ... qu'il, and que . . . qui. The first disappeared in the 16th century, because of the need for a clause to have a subject . The que . .. qui construction has also tended to disappear or become marginal since the 17th century—probably because of the strange plurality of the functions ofqui, concurrently complementizer and surface subject. This syntactic monster has now been replaced by looser constructions : the infinitive with verbs of perception (l'homme queje vois venir), the parenthetical construction (l'homme qui, je crois, vient), or the dont ... que construction for the other verbs (un homme dontje sais qu'il est capable de tout). H's book is remarkable because of his comprehensive knowledge of synchronic and diachronic syntax. He is completely at ease among the dustiest German dissertations from the end of the 19th century, as well as the most recent and technical works. Although at first sight one may suspect a discrepancy here between syntax and philology, H's book is one of the most successful endeavors to bridge the gap. I can think ofno better example ofa work treating a specific syntactic problem in depth. [Jacques M. Julien , University of Texas, Austin.] L'inversion dans la subordonnée en français contemporain. By Kerstin Wall. (Acta Universitatis Upsaliensis , Studia Románica Upsaliensia , 30.) Stockholm: Almqvist & Wiksell, 1980. Pp. 179. Kr. 66.50. Dealing with verb-subject inversion in subordinate clauses, this book is essentially a statistical confirmation of Le Bidois's L'inversion du sujet dans laprose contemporaine (1952) and Blinkenberg's L'ordre des mots en français moderne (1969). Starting from a corpus based on 30 novels and 30 newspapers from the 60's and 70's, Wall draws the following conclusions: (1)Inversion is favored when the subject is longer than the verb, and when the verb itself is semantically 'weak'. Blinkenberg is criticized for neglecting the former parameter, Le Bidois for neglecting...