Phonetic variations driven by prosodic structure and their function in speech production

Abstract

The present study investigates how prosodic strengthening driven by prosodic boundaries is manifested in relation to its linguistic function in the production of different speech styles, focusing on English plosives /p, t, k, b, d, g/ and nasals /m, n/ in IP-initial position compared to IP-medial position in interactive and read speech. Boundary-induced prosodic strengthening has been observed in domain-initial positions compared to domain-medial positions in articulatory and acoustic dimensions. This phenomenon has been accounted for by the syntagmatic contrast enhancement account, which stipulates that the contrast between neighboring consonants and vowels is enhanced, and by the paradigmatic contrast enhancement account, which posits that phonological contrasts such as voicing and manner of articulation contrasts are enhanced. By testing these two accounts in read and interactive speech, the present study aimed to provide a more comprehensive understanding of the mechanisms behind boundary-induced prosodic strengthening and investigate potential factors that modulate boundary-induced prosodic strengthening.In an interactive speech task, disyllabic English words that contained word-initial plosives and nasals were elicited in IP-initial and IP-medial positions in a task where a pair of participants interacted to find out what was going on in Animal Village. In each trial, one participant was given pictorial information, and the information needed to be verbally delivered to the other participant whose task was to ask a question and choose the scene among the given pictorial options based on the verbal description. In a read speech task that was administered with a few days later, the same target words were elicited in IP-initial and IP-medial positions in a task where participants were asked to read written sentences. The present study analyzed the data of 18 participants who participated in both the interactive and read speech tasks. Prosodic strengthening was evaluated by comparing durational, amplitudinal, and spectral measurements of the initial consonant in IP-initial and IP-medial positions. The analysis revealed that the patterns of prosodic strengthening were contingent on speech sounds and its acoustic correlates. For plosives, the VOT results showed evidence of syntagmatic contrast enhancement in read speech whereas they showed evidence of paradigmatic contrast enhancement in interactive speech. Spectral acoustic correlates showed paradigmatic contrast enhancement in both interactive and read speech when there was prosodic strengthening. Based on the fact that syntagmatic contrast enhancement was only observed in read speech, speech style influences prosodic strengthening and its linguistic function such that speakers may simply mark a prosodic juncture without the need for directly delivering messages to listeners in read speech, rather than having to help listeners identify the speech sound better by making it phonologically more distinct from other speech sounds as in interactive speech. Unlike plosives, except for durational acoustic correlates, prosodic strengthening on nasals showed the pattern of syntagmatic contrast enhancement in both read and interactive speech possibly due to fact that English nasals are phonetically and phonologically distinct enough from non-nasal consonants so speakers may not need to enhance the distinction between them when delivering messages to listeners in interactive speech. Overall, the present study suggests that prosodic strengthening driven by prosodic boundary and its linguistic function can be modulated by whether or not speech style is interactive (i.e., interactive vs. read speech) and by whether speech sounds are phonetically and phonologically distinct enough when compared to other speech sounds. Prosodic strengthening driven by prosodic boundary and its linguistic function should be understood in the interaction with different factors that modulate its patterns.