1 Especially Aspects of the theory of syntax (Cambridge, Mass., 1965). Some of the theoretical notions incorporated in this paper were first suggested to me by Paul M. Postal in a course given at the 1964 Linguistic Institute. A shorter version of the paper was read at the Summer Meeting of the Linguistic Society in Ann Arbor, 31 July 1965.

2 Some of my rules may have to be restricted further on the basis of tense.

3 Acting like the Q posited for English by Jerrold J. Katz and Paul M. Postal in An integrated theory of linguistic descriptions (Cambridge, Mass., 1964). It may be advantageous to set up both a question marker Q and wh.

4 An example sentence marked with an asterisk is ungrammatical.

5 This three-step analysis was first suggested by R. B. Lees, to whose instruction this paper owes a great deal. Lees, F. M. Jenkins, and Dan. A. Wilson were kind enough to read the paper in manuscript.

6 This can be strongly supported in a detailed examination of French determiners, which is beyond the scope of this paper. The actual underlying nominal would be something like wh+quelque+un+de+le+N. The deletion of de+le is quite general in determiners, as is the deletion of un. When both have been deleted, wh+quelque+N is rewritten as quel+N. When neither has, an obligatory rule transforms wh+quelque+un to le+wh+quelque, this becoming lequel by independently required morphophonemic rules. This accounts for the lequel interrogatives, e.g. Lequel des tableaux préférez-vous? ‘Which of the pictures do you prefer?‘, otherwise not discussed here.

There is also evidence from selectional restrictions. For example, quel is not used to question pronouns or the demonstrative cela, but quelque does not precede them either, so that no additional statements are needed to prevent *quel cela? and the like from being generated.

7 This formulation of the problem is oversimplified; but the restrictions on possible sources of quel+N in complex sentences have not been investigated and need not concern us here.

8 Katz and Postal (1964) use the same device in the corresponding rule for English.

9 Some questions syntactically similar to 25 are avoided because of ambiguity. Thus Quelle femme aime Henri? may mean either 'What woman loves Henry?' or 'What woman does Henry love?' It is assumed that this is a stylistic but not a grammatical problem. In any case, such sentences are certainly not deviant in the clearly syntactic sense that some qui-que questions are, for example *Qu'est tombé? 'What fell?'.

10 However, it is possible that proper nouns, and for that matter pronouns, are in fact preceded on an abstract level by a determiner which is later in most cases obligatorily deleted.

11 In addition it may be possible that T5 can be independently motivated for the derivation of reflexive pronouns.

12 See Katz and Postal (1964) for a discussion of the use of pro elements in a grammar.

13 With minor restrictions, like the absence of a sentence adverb.

14 Aspects of the theory of syntax. This convention is probably not satisfactory, but it serves my present purposes.

15 Comment on parle en français 131 (Paris, 1927).

16 See Albert Dauzat, Grammaire raisonnée de la langue française ² 286, 431 (Lyon, 1947); W. H. Fraser and J. Squair, A French grammar 32 (Boston, 1906); Félix Gaiffe et al., Grammaire Larousse du xx^e siècle 76, 187 (Paris, 1936); Walter von Wartburg and Paul Zumthor, Précis de syntaxe du français contemporain ² 37 (Bern, 1958); Georges and Robert Le Bidois, Syntaxe du français moderne 1.363 (Paris, 1935); Maurice Grevisse, Le bon usage ⁶ 406, 702 (Paris, 1955); Albert Valdman, French 19 (Boston, 1961); and Lucien Tesnière, Eléments de syntaxe structurale 198 (Paris, 1959).

17 Cf. Robert B. Lees, ‘Analysis of the “cleft sentence” in English’, Zeitschrift für Phonetik, Sprachwissenschaft und Kommunikationsforschung 16.371-88 (1963). Lees gives a separate embedding rule for this construction, a possible alternative to my proposed source.

18 ‘Le système des pronoms qui, que, quoi en français’, Le français moderne 29.241 (1961).

19 T2 may have to be further restricted so that the emphasized NP is quelque +[+PRO] only when the matrix string is headed by wh, i.e. when a wh question can be formed on it. This depends on whether a sentence like C'est quelque chose que j'ai vu ‘It's something that I saw’ is considered syntactically deviant.

20 Some added restrictions on T2 are required.