Abstract
Linguistic descriptions of speech often exclude the concept of naturalness since they usually focus on ideal forms. As a consequence, they sometimes miss essential features of speech structure which not only contribute to its natural sound, but also support the coding of the meaning.
That is the case of intonation downtrends, which received the deserved level of attention only relatively recently. However, there are still uncertainties concerning the methodology of their quantification.
The present study evaluates four different approaches across two prosodic domains and two speech genres. We processed over 2400 prosodic phrases and over 850 declination units produced by 24 speakers.
The chief objective was to map behavior of downtrends in Czech spoken texts, and suggest a suitable method of their quantification facilitating future cross-linguistic comparisons. Moreover, reference data are provided on intonation downtrends for Czech, a West Slavic language of Central Europe.