The question of whether intonation contours directly signal meaning is an old one. We revisit this question using vocatives in Colombian Spanish (Bogotá). We recorded speakers' productions in three pragmatic conditions – greeting, confirmation-seeking, and reprimand – and compared proper names (vocatives) to situation-specific one-word utterances, such as ¡Hola! ‘Hello’ (non-vocatives). Intonational analyses showed no direct one-to-one correspondence between the pragmatic conditions and intonation contours: (a) for vocatives, e.g. a rising–falling contour occurred in both greetings and reprimands; and (b) for non-vocatives, e.g. a step-down contour (a.k.a. calling contour) occurred in both greeting and confirmation-seeking conditions. Looking beyond intonation to consider other phonetic variables – spectral tilt, duration, alignment of tonal targets, f0-range, f0-slope – the results showed that the intonation contours that occurred in more than one pragmatic condition differed in phonetic realisation, e.g. rising–falling vocatives showed differences in f0-range of the rise and spectral tilt. However, the corresponding non-vocatives did not show the same differences. Furthermore, vocatives in greeting contexts were realised differently from non-vocatives in greeting contexts. In sum, the pragmatic condition affects the prosodic realisation of (non-)vocatives, but the relationship is complex. The results are discussed in the light of prosodic constructions, leading to the conclusion that the prosodic realisation of vocatives and non-vocatives in Bogotá Colombian Spanish cannot be easily modelled by prosodic constructions.

Strings of words can correspond to more than one interpretation or underlying structure, which makes them ambiguous. Prosody can be used to resolve this structural ambiguity. This dissertation investigates the use of prosodic cues in the domains of fundamental frequency (f0) and duration to disambiguate between two interpretations of ambiguous structures when speakers addressed different interlocutors. The dissertation comprises of three production studies and one comprehension study. Prosodic disambiguation was studied with a focus on German name sequences of three names (coordinates) in two conditions: without (Name1 and Name2 and Name3) and with internal grouping of the first two names ([Name1 and Name2] and Name3). The study of coordinates was complemented with production data of locally ambiguous sentences with a case-ambiguous first noun phrase. Variability was studied in a controlled setting: Productions were elicited with a within-subject manipulation of context in a referential communication task in order to evoke prosodic adaptations to different conversational contexts. Context had five levels and involved interlocutors in three age groups (child, young adult, elderly adult) with German as L1 in the absence of background white noise, the young adult with background white noise, and a young adult without German as L1. Variability was explored at different levels: within a group of young individuals (intra-group level), within and between young individuals (intra-individual level and inter-individual level, respectively), and comparing between the group of young and a group of older speakers (inter-group level). Our data replicate the use of the three prosodic cues (f0-movement, final lengthening, and pause) in productions of young adult speakers and extend their use to productions of older adult speakers. Both age groups distinguished consistently between the two coordinate conditions. Prosodic grouping in production was evident not only on the group-final Name2 but also at earlier stages in the utterance, on the group-internal Name1 (early cues). For some speakers, some listeners were able to decode these early cues effectively as they were able to reliably predict the upcoming structure after listening to Name1 only. Thus, prosodic grouping appears as a globally marked phenomenon building up along the utterance. The internal structure of coordinates was disambiguated irrespective of the conversational context. In our data, speakers only slightly modified the prosodic cues marking the disambiguation in the different contexts. Listeners were unable to identify to which interlocutor the sequence had been produced. We interpret this intra-individual consistency in the production of disambiguating prosodic cues as support for a strong link between prosody and syntax. The findings support models in favour of situational independence of disambiguating prosody. All speakers reliably marked the distinction between the grouping conditions with at least one of the three prosodic cues investigated and most of the speakers used at least two of these cues. Further, individual differences in prosodic grouping did not lead to difficulties in recovering the grouping in comprehension. Taken together, these findings support the existence of a phonological category of prosodic grouping.