Özet:
This study investigates the use Sentence Topics in Turkish naturally occurring conversations. In particular, it focuses on the correlation between the pragmatic functions, informative status and the prosodic properties of Sentence Topics in Turkish. The findings of the study are based on the analysis of naturally occurring conversations extracted from three different recordings that are recorded with seven participants and in different settings and time. In line with previous studies in the literature, Sentence Topics are characterized into two groups, discourse old information bearing Sentence Topics and pragmatic aboutness bearing Sentence Topics. It is shown that there is a dichotomy in the functioning of Sentence Topics that is sensitive to the informative status of Sentence Topics. In particular, it is claimed that Sentence Topics that bear discourse old information are usually used to mark topic maintenance and pragmatic aboutness bearing Sentence Topics function as topic shift markers. It is also shown that the dichotomy in the functions of Sentence Topic is not reflected in the prosodic properties of these items. However, we show that there is another dependency at work in the tonal marking of the Sentence Topics. Sentence Topics, no matter what their function is, receive a rising tone (L*H) when they occur preverbally, and a downstepped falling tone (!H*L) when they occur postverbally. Another point that we observed was related to the correlation between the different functions of Sentence Topics and ellipsis. In particular, we stated that only topic maintenance marking Sentence Topics can be omitted if the speaker assumes that they are recoverable from discourse (accessible in the discourse). On the other hand, regardless of their informative status, Sentence Topics that mark topic shift can never be elided. Finally, we agreed with the claims that the informative value of Sentence Topics are encoded through the speaker's assumptions about the hearer's familiarity to the information conveyed. We further argued that this structuring mechanism may also be constrained by some other factors such as extra-linguistic context or information processing of the participants.
