The status of syntax in speech: comments on the Language Into Act Theory proposal

Language into Act Theory (CRESTI, 2000) is a pragmatic corpus-driven theory of spontaneous speech. The utterance is individuated as the reference unit of speech. It is defined as the smallest linguistic unit with pragmatic autonomy and interpretability in isolation. This theory states that prosody delimits the utterance, illocutions (AUSTIN, 1962) and information structure, that develops from tonal units. Each tonal unit corresponds in principle to an information unit at the pragmatic level. According to Cresti (2014), the domain of syntax would be constrained to the information units, resulting in the establishment of what the author calls syntactic-semantic islands. In this paper, we present a discussion about this approach. Through exposition of empirical data from corpora, as well as from languages of typologically varied languages, we argue on the possibility that the syntactic relations between the elements distributed along two or more information units may exceed the boundaries of such units. Data on predication, subordination, coordination, negation, and case markers illustrates this assumption, insofar as these relationships are not disrupted by the presence of a prosodic break carrying information value. Therefore, two interface levels must be distinguished, but they are not isomorphic: the prosodic-pragmatic and the syntactic levels. We argue that spoken syntax would be better studied by means of probabilistic criteria.