This study investigates the phonetics and phonology of phrasal domains in Turkish prosody and discusses its implications for the syntax-prosody mapping. The findings and proposals are based on a corpus of 1152 sentences extracted from a total number of 1144 spoken dialogues. The dialogues were acted out by nine native speakers who speak the standard variety of modern Turkish.
In the light of the data, an inventory of pitch accents and edge tones is proposed for Turkish. The findings indicate that Turkish prosody governs a separate and single level of phrasing above the Phonological Phrase, namely the Intonational Phrase (IP). The evidence is from boundary tone placement, linguistic pause distribution, the position of head-prominence, and phrase final lengthening of vowels at IP-final positions.
Based on the structures where IP-formation is and is not induced, a new theory of intonational phrasing is proposed. It is shown that every structure with illocutionary force yields intonational phrasing in phonology. Considering that the same structures are also the loci of the so-called "clausal tunes", we further question the clause-typer status of intonation, as pursued in some studies in the literature. With evidence from a variety of grammatical processes, specifically complementation, it is shown that intonation cannot be envisaged as a clause-typer. Rather than being sentential force indicators, such tunes are argued to be the reflexes of illocutionary force. In relation to these proposals, the conception of a single ForceP layer (Rizzi 1997) is rejected, and a two-way partitioned representation of ForceP is proposed: an outer ForceIllocutionaryP layer, which dominates an inner ForceSententialP layer. In this mechanism, clause-typing operates at ForceSentential0, whereas ForceIllocutionary0 specifies speaker intentional meaning (Grice 1957, Searle 1965, Chierchia and McConnell-Ginet 1990). This proposal captures not only why both intonational phrasing and the "clausal tunes" are restricted to the same structures, i.e. those with the ForceIllocutionaryP layer, but also why the structures lacking both phenomena are devoid of illocutionary force. The latter are argued to be truncated from ForceSententialP.
Intonational phrasing facts are also the source of another proposal regarding relativization. Based on the phonological, semantic and pragmatic disparities between prerelatives and the ki-headed postrelatives, a new classification for relativization is proposed for Turkish. Prerelatives are integrated relatives and ki-relatives are supplementary relatives in the sense of Potts (2003, 2005). ki-relatives and the ki-headed clauses that only act as parentheticals are subsumed under the supplemental ki-clause category, and they are contrasted with the ki-headed finite complement clauses (ki-FCC). It is shown that supplemental ki-clauses initiate IP-formation, whereas ki-FCCs are prosodically integrated structures. This disparity is attributed to the nature of the ForceP layers in their syntax.
Finally, the prosody of arguments is discussed. Contra impressionistic approaches to prosody, it is shown that the phrasing behaviors of arguments are not as rigid as they are envisioned to be. It is argued that some of these patterns yield distinct classes of meanings, while some of them are semantically vacuous structures.