Finding Closure and Looking Outward
I have argued in favor of an analysis of German monoclausal multiple wh-questions that involves topic movement of the two wh-elements into the left periphery of the clause in the overt syntax. The starting point was the so-called intervention effect (Beck 1996a), which I argued against on theoretical and empirical grounds. Especially the latter demonstrated that any attempt to capture the ‘‘Beck e¤ect’’ using some notion of intervention under generates, hence fails. The alternative developed here capitalizes on a property other than quantification that illegitimate in-betweeners share: they are all nontopicalizable. Fleshing out this notion led to further empirical justification of the validity not only of this generalization, but also of the theoretical proposal whereby both wh-phrases (as well as a legitimate in-betweener, for example) move to a left-peripheral topic position. One possible conclusion to draw from the analysis is to deny that multiple wh-questions exist in German in the first place, by analogy with Italian, on the basis of finer discourse properties distinguishing German from English, for example (Pesetsky 1987). But this is not the only necessary conclusion, and it need not be the final word either. Finally, I connected the structures under discussion to clitic-doubling wh-phrases in (colloquial) Bulgarian. It has been argued that here too topicality in a wider sense plays a role (Jaeger 2004), something Cedric Boeckx and I have independently explored on the basis of languages such as Romanian (Boeckx and Grohmann 2004). How deep this wh/clitic connection goes remains to be seen. Let me finally comment on some possibilities.
As noted above, the phenomenon of doubling wh-phrases with a clitic is not unique to Bulgarian, and Jaeger’s study does not really break new ground on an empirical level. Nor does the attempt to tie in doubled wh phrases with some notion of topicality. What is new and highly intriguing, however, is the double explicitness of Jaeger’s study.
First, in that it explicitly spells out such intuitions rather than mentioning them as a side remark. Bear in mind that the language under con sideration is Bulgarian; while there has been a tremendous amount of research on clitic strategies in Romance languages, including the distinction between clitic doubling and clitic left-dislocation, the body of research on related phenomena in Slavic is considerably smaller. The consensus seems to be that Slavic has one strategy only, which is commonly referred to as ‘‘clitic doubling.’’ The relation between these clitic strategies and information structure is as obvious as it is contested. While the two obviously go hand in hand, it is not necessarily clear which (type of) strategy goes with which type of discourse function. In Romance, for example, there is a long-standing tradition of associating clitic left dislocation with topicality (in the present sense), but not necessarily and/ or exclusively clitic doubling (see Villalba 2000 for a summary). Whichever way the dice fall, the phenomenon at hand is clearly one that shows topicality effects, if not even restrictions akin to D-linking or DRQ. For what it is worth (see also, e.g., Leafgren 1997), I assume that Bulgarian clitic doubling is of the ‘‘right’’ variety to correlate with this kind of function. Future research will surely shed more light on the issue.
It is in this sense that Jaeger’s results are doubly interesting. To the extent that the tentative yet explicit connection to my main proposal holds up to scrutiny, there would be independent support for some of the wilder stabs at German left-peripheral syntax taken above. This is the topic of ongoing research—and if nothing else, this hypothesis and some correlating predictions are empirically testable.
Note, incidentally, that the parallelism between clitic doubling and wh phrases, the type of wh/clitic connection I am interested in (for further interpretations, see Boeckx and Stjepanovic´ 2005; Poletto and Pollock 2002), is not restricted to the Balkan Sprachbund, and neither is the notion of a wh-topic, even in the explicitness presented above (for both Bulgarian and German).
Another language that has been argued to contain wh-topics is (Modern) Greek. On the basis of a wide range of properties, Androulakis (1998) suggests that some wh-phrases are in fact wh-topics—namely, (at least) those that are doubled by a clitic. Her analysis accounts for the ambiguity she observes between a fronted wh-phrase and a lower quantifier in regular questions, which disappears when the wh-phrase is doubled, but it also extends to data pertaining to parasitic gaps, weak crossover effects, and others. Importantly, she also explores a notion of topicality, presuppositionality, and discourse properties of such wh-topics basically driven by the need to say something in addition to more or less well-established concepts like D-linking (see my remarks above).
And just to stress that the connection between wh-phrases and topic discourse properties is not the figment of limited imagination, let me mention Reglero’s (2003) and Lambova’s (2003) work on multiple wh questions in Basque and Bulgarian, respectively, which bears heavily on these issues, as well as Poletto and Pollock’s (2002) explorations in Ro mance (Northern Italian dialects, predominantly). The latter work in particular picks up on the wh/clitic connection made explicit in Boeckx and Stjepanovic´ 2005, though it also relates to present concerns. In the larger picture, then, we might add ‘‘topicality’’ to the connection and explore even more intricate patterns.
While I do not wish to appear to take too functional a perspective on grammar, one wonders how narrow syntax is and how narrow syntax should be—or, more rhetorically, following up on the discussion in Chomsky 2000 and much subsequent work: what exactly is narrow syntax?