Abstract
The present study addresses the unresolved issue of speech melody in Czech question-word questions (i.e., specific, information or content questions). Some renowned authors claim that these questions use the same melodic courses as statements, whereas others suggest salient modifications.
We contribute to the debate with the results of an experiment, in which 24 speakers were recorded producing context-embedded utterances that formed latent question-statement pairs. The pairs under scrutiny were not obvious to the speakers but they structurally differed in just one or two phonemes and preserved identical stress-group and syllabic patterns.
The analysis of the melodic contours shows that although both statements and interrogatives are generally (even if not always) produced with a falling contour, the placement of the melodic peak and its magnitude differ significantly. While statements are typically furnished with a late intonational nucleus, question-word questions more commonly display a higher early peak.
Additionally, various pragmatic components of the messages are suggested that correlate with the individual tune shapes.