Finally, I return to an issue held over from §§3.4, 4.4.1 and 5.6, namely, the specifics of the syntax of ‘what’ (and how it contrasts with that of ‘who’). There’s just one word for ‘who’: qui, which behaves like a regular wh phrase, as shown in column 1 in (255); it can undergo wh fronting or remain in situ; it can function as a direct object (255a–i), within an inherent-case-marked DP* (255j–r), and as a subject (255s–y). In
contrast, there are two words for ‘what’: quoi and que, as shown in columns 2 and 3 in (255):
(255)
(255) 1. QUI 2. QUOI 3. QUE/QU’
a. Jean voit qui? J. sees who
Jean voit quoi? J. sees what
*Jean voit que? J. sees what b. Qui il voit? who he sees *Quoi il voit? what he sees *Qu’il voit? what-he sees c. Qui voit-il? who sees-he *Quoi voit-il? what sees-he Que voit-il? what sees he d. Qui Jean voit-il?
who J. sees-he
*Quoi Jean voit-il? what J. sees-he
*Que Jean voit-il? what J. sees-he e. Qui voit Jean?
who sees J.
*Quoi voit Jean? what sees J.
Que voit Jean? what sees J. f. Je sais qui Jean voit.
I know who J. sees
*Je sais quoi Jean voit. I know what J. sees
*Je sais que Jean voit. I know what J. sees g. Je sais qui voit Jean.
I know who sees J.
*Je sais quoi voit Jean. I know what sees J.
*Je sais que voit Jean. I know what sees J. h. Voir qui? see who Voir quoi? see what *Voir que? see what i. Qui voir? who see Quoi voir? what see Que voir? what see j. Jean parle de qui?
J. speaks of who
Jean parle de quoi? J. speaks of what
*Jean parle de que? J. speaks of what k. De qui il parle?
of who he speaks
De quoi il parle? of what he speaks
*De qu’il parle? of what-he speaks l. De qui parle-t-il?
of who speaks-he
De quoi parle-t-il? of what speaks-he
*De que parle-t-il? of what speaks-he m. De qui Jean parle-t-il?
of who J. speaks-he
De quoi Jean parle-t-il? of what J. speaks-he
*De que Jean parle-t-il? of what J. speaks-he n. De qui parle Jean?
of who speaks J.
De quoi parle Jean? of what speaks J.
*De que parle Jean? of what speaks J. o. Je sais de qui Jean parle.
I know of who J. speaks
Je sais de quoi Jean parle. I know of what J. speaks
*Je sais de que Jean parle. I know of what J. speaks p. Je sais de qui parle Jean.
I know of who speaks J.
Je sais de quoi parle Jean. I know of what speaks J.
*Je sais de que parle Jean. I know of what speaks J. q. Parler de qui? speak of who Parler de quoi? speak of what *Parler de que? speak of what r. De qui parler? of who speak De quoi parler? of what speak
*De que parler? of what speak s. Qui pue? who stinks *Quoi pue? what stinks *Que pue? what stinks t. Je sais qui pue.
I know who stinks
*Je sais quoi pue. I know what stinks
*Je sais que pue. I know what stinks u. Jean dit que qui pue?
J. says that who stinks
*Jean dit que quoi pue? J. says that what stinks
*Jean dit que que pue? J. says that what stinks v. Qui il dit qui pue?
who he says that stinks
*Quoi il dit qui pue? what he says that stinks
*Qu’il dit qui pue? what-he says that stinks
In fact, there is one exception to this, namely (255i), where direct-object quoi precedes the infinitive. Given the 95
otherwise robust nature of the generalisation it’s tempting to conclude that the pre-infinitival position of quoi is due, not to full wh fronting, but to the fact that the verb, as an infinitive, doesn’t need to raise as high as a finite verb.
Note that que is ungrammatical in both subordinate contexts in (255f, g). Ungrammaticality is expected in (255f), since 96
que doesn’t immediately precede the verb, but not (255g), where it does, thanks to SI. Indeed, (255g) contrasts minimally with (255e). Of course, the interrogative is matrix in (255e) but subordinate in (255g). Recall from §5.6.1.1 that matrix and subordinate non-subject wh interrogatives differ with respect to the landing site of wh movement: SpecFocusP in matrix contexts, like (255e), SpecForceP in subordinate contexts, like (255g). It makes sense to explain the contrast between (255e, g) on this basis: although fronted que in (255e, g) is indeed immediately preverbal in terms of linear order, its position in SpecFocusP in (255e) means that it’s close enough to the verb, while its position in SpecForceP in (255g) means that it’s too far away from the finite verb. I return to this below.
In column 3 of (255d), the subject DP* Jean intervenes between fronted que and the verb, hence the ungrammaticality. Strangely, where the intervening subject DP* is ceci ‘this’ or cela ‘that’, the problem disappears (Plunkett 2000: 513), as in (i):
(i) Que cela/ceci veut-il dire? what that/this wants-it say ‘What does that/this mean?’
Note that, like pronominal clitics, que cannot be co-ordinated: *que ou qui ‘who or what’; cf. qui ou quoi (Plunkett 2000: 97
513).
The syntax of que diable ‘what the hell’ warrants special comment. It’s unlike regular que in at least two respects. First, 98
unlike simple que, que diable is compatible with CI: w. Qui dit-il qui pue?
who says-he that stinks
*Quoi dit-il qui pue? what says-he that stinks
Que dit-il qui pue? what says-he that stinks x. Qui Jean dit-il qui pue?
who J. says-he that stinks
*Quoi Jean dit-il qui pue? what J. says-he that stinks
*Que Jean dit-il qui pue? what J. says-he that stinks y. Je sais qui il dit qui pue.
I know who he says that stinks
*Je sais quoi il dit qui pue. I know what he says that stinks
*Je sais qu’il dit qui pue. I know what-he says that stinks
Typically, each who-context in column 1 containing qui has one grammatical what-counterpart in column 2 containing quoi or column 3 containing que (but see footnote 95). Realisation of direct-object ‘what’ is sensitive to two factors: wh fronting and position with respect to the verb. Direct-object ‘what’ is realised as quoi if it stays in situ (255a, h) (§5.6.2); bare quoi is incompatible with wh fronting (§5.6.1). Instead, fronted95
direct-object ‘what’ is realised as que, but only provided it immediately precedes the verb, as in (255c, e), where PI and SI have taken place, and in the infinitival in (255i). If fronted direct-object ‘what’ doesn’t immediately precede the verb, as in (255b, d), where CI has taken place, or there’s no inversion at all, ‘what’ isn’t realised as que. The realisation conditions of direct-object ‘what’ are such that some of the contexts in96
column 1 in (255) containing direct-object qui have no grammatical counterpart in columns 2/3 containing quoi/que. These are (255b, d, f, g), italicised in columns 2/3. Inherent-case-marked ‘what’ is realised as quoi throughout; que is systematically excluded, as in (255j–r), irrespective of whether wh fronting or inversion has taken place.
Before we turn to subject ‘what’, note that the above pattern can be understood by assuming that quoi/que share syntactic – as well as obvious morphological – properties with moi/me, toi/te, soi/se, which are pairs of corresponding non-clitic–clitic proforms (§§3.8, 4.4.1): moi/toi/soi are non-clitics; me/te/se are clitics. Non- clitics are compatible with independent stress and overt inherent-case marking; clitics are not. The distribution of quoi/que97 suggests that they are non-clitic and clitic, respectively: quoi appears within an inherent-case- marked DP* or as a non-fronted direct object (that is, in a position where it can bear stress); que is a fronted direct object, necessarily cliticising onto a verb (that is, in a position where it can’t bear stress).
Note further that neither moi nor me, etc., is compatible with the canonical subject position, as shown in (256b, c):
(256) a. Je suis fatigué. b. *Me suis fatigué. c. *Moi suis fatigué. I am tired me is tired me is tired ‘I’m tired.’
Instead, a member of a distinct set of subject proforms is used (§4.4.4), as in (256a). The relevance of the parallel between quoi/que and moi/me, etc., is that, if the former share relevant properties with the latter, then we expect them, too, to be incompatible with the canonical subject position (unlike qui). And this is indeed what we find, as shown in (255s–y). With 98 moi/me, etc., inability to occupy the canonical subject position isn’t
(i) a. *Que Pierre a-t-il fait? b. Que diable Pierre a-t-il fait? what P. has-he done what devil P. has-he done ‘What did P. do?’ ‘What the hell did P. do?’ Second, unlike simple que, que diable can appear in subject position: (ii) a. *Qu’a pu le motiver? b. Que diable a pu le motiver?
what-has been-able him motivate what devil has been-able him motivate ‘What could have motivated him?’ ‘What the hell could have motivated him?’
This begs the question, of course, of why no such proform exists. I leave this issue on the research agenda. 99
problematical because a set of subject proforms is available, as in (256a). In contrast, quoi/que have no corresponding subject form, and an alternative structure altogether is needed to ask, for example, ‘What99
stinks?’. Further, the alternative strategy used to ask such questions is exactly what we might expect. Here’s a thought experiment: Imagine that French subject proforms didn’t exist. How might a French speaker say ‘I’m tired’? Je suis fatigué, as in (256a), isn’t a possibility because the proform doesn’t exist. As we saw in (256b, c), neither are Moi suis fatigué and Me suis fatigué (the proforms are incompatible with SpecIP*). In the absence of a subject proform, the only way to say ‘I’m tired’ is to use a structure which allows me or moi to appear in a position other than the canonical subject position. Such a structure is a cleft (§5.5):
(257) C’est moi [qui – suis fatigué]. it-is me that is tired
‘I’m the one who’s tired.’/‘I’m tired.’
Here, the 1SG focus, underlined in (257), is the dependant of the copula être, and therefore realised as non- clitic moi. Exactly this clefting strategy provides a way of asking ‘What stinks?’. One grammatical alternative to the ungrammatical *Quoi/Que pue? is the cleft in (258a), while a second, in (258b), is derived from (258a) via wh fronting (§5.6.1) and PI (§5.7):
(258) a. C’est quoi [Op qui – pue]? b. Qu’est-ce – [Op qui – pue]? it-is what that stinks what-is-it that stinks ‘What stinks?’ = (258a)
The pattern illustrated in (258a, b) with subject ‘what’ has been generalised to subject ‘who’ and direct-obect ‘who’/‘what’, giving the full paradigm in (259a–d):
(259) a. C’est quoi qui pue? 6 Qu’est-ce qui pue? (subject ‘what’) it-is what that stinks what-is-it that stinks
b. C’est qui qui pue? 6 Qui est-ce qui pue? (subject ‘who’)
it-is who that stinks who is-it that stinks
c. C’est quoi qu’il voit? 6 Qu’est-ce qu’il voit? (direct-object ‘what’) it-is what that-he sees what-is-it that-he sees
d. C’est qui qu’il voit? 6 Qui est-ce qu’il voit? (direct-object ‘who’) it-is who that-he sees who is-it that-he sees
While (259a) gives the only ways of expressing a matrix subject ‘what’ interrogative (cf. columns 2 and 3 in (255s)), those clefts in (259b–d) are alternatives to non-cleft forms in column 1 in (255s), as well as (255c).
As for subordinate direct-object and subject ‘what’ interrogatives, we again find that neither quoi nor que can be used, as in (255f, g, t). With subject ‘what’ this is expected: quoi/que are non-subject forms. With object ‘what’ quoi is ruled out because quoi is incompatible with wh fronting, and que is ruled out since, although it can undergo wh fronting, the unavailability of PI or SI in subordinate interrogatives (see Table 5.2 on page 143) means that it doesn’t have a verb to cliticise onto. Instead, what is found in subordinate subject/direct- object ‘what’ interrogatives is a free relative (§5.6.1):
(260) a. Je sais [ce que Jean voit]. b. Je sais [ce qui pue]. I know that that J. sees I know that that stinks ‘I know what J. can see.’ ‘I know what stinks.’