We study two question particles (donc and ça) in a large French corpus. We show that they are specialized for direct questions, and that wh-ça is possible both in situ and in fronted position, contrary to previous literature (Cheng and Rooryck 2001). We also show that ellipsis is the most common use of wh-ça, with sluicing (Ross 1969) and a new construction that we call wh-stripping, while fronting is the most common use of wh-donc. In many examples, a non elliptical question would be impossible or quite different, which seem to cast doubt on a deletion under identity approach of sluices (Ross 1969) and to favor a Direct Interpretation approach (Ginzburg and Sag 2000). Looking at the preceding context, in about half of wh-ça and wh-donc sluices, the wh-word has a correlate, and in the majority of cases, the correlate is definite. We suggest that the particle ça favors elliptical reprise questions, compared to bare wh words, while the particle donc favors biased questions.
1. Introduction
2. Previous corpus studies
3. Our corpus of contemporary French
4. A new corpus study of wh-ça
5. A corpus study of wh-donc
6. General discussion
7. Conclusion
References