Turkish as a (non)- SOT Language
Luages differ as to how embedded tense gets interpreted. English, for example, is classified as a Sequence-of-Tense (SOT) language, and this classification is based on the observation that past-under-past in English is interpreted vacuously (see Sharvit 2013). Tense embedding thus yields two distinct temporal interpretations in an English sentence like ‘John said that Mary was sick’ where one interpretation is such that Mary is sick at the moment o f speech (i.e. the simultanous reading) or that she -was sick before the utterance time (i.e., the backshifted reading). Translation of this SOT sentence from English to Turkish produces a single interpretation, where only the backshifted interpretation is available, particularly when the embedded clause is of the finite complement clause type that is morphologically marked with the past tense+direct evidential marker [-DI], This evidence alone would suggest that Turkish should classified as a non-SOT language. We argue in the present paper that classifying Turkish as a non-SOT language on the basis o f this simple observation yields incorrect results, and taking into consideration another type of complementation in Turkish compels a reevaluation o f the initial observation that Turkish is a non-SOT-Ianguage. Specifically, we show that evidence based on nominal complement clauses in Turkish allows an alignment of Turkish with SOT languages. The implication of this is that whether a language is an SOT-language or not is contingent at least upon the type of complementation. We further argue that another factor also contributes to the interpretation of embedded tense in Turkish, therefore to its classification as a (non-)SOT language, namely the fact that the tense embedded under Past is morphologically either [past+direct evidential] or [past+indirect evidential].
The following license files are associated with this item: