ETD

• legacy data
• reuse
• sociophonetics

This doctoral thesis presents a case of historical data reuse of Angela Spinelli’s historical oral archive, collected in the early 1980s in the rural area of Prato, Tuscany, to conduct a sociophonetic investigation of the phenomenon of rhotic consonant degemination. The distinguishing element of this research is the analysis of data that can no longer be elicited today due to the disappearance of the original speakers and the profound social changes that have taken place in the area. The reuse of the archive collected by the historian Angela Spinelli is motivated by its richness in audio recordings and accompanying materials as well as the methodological framework adopted in her investigation. Furthermore, this research work places the reuse of research data as a core element within a Findable, Accessible, Interoperable and Reusable (FAIR) framework.
The analysis of archival data focuses on bisyllabic words containing /rr/ in intervocalic position, that strongly relate to the topics addressed in Spinelli’s research: the period of the “war” (guerra), work on the “land” (terra), factory work with “iron” (ferro), the use of the “cart” (carro), and so on. To assess the presence of the phenomenon of degemination in the data objectively, a perceptual test was conducted including phoneticians with familiarity with the phenomenon to discriminate cases of phonologically geminate (/rr/) words produced as phonetically degeminated (/r/) or geminated. The responses were later evaluated calculating the inter-annotator agreement among participants.
Based on the outcomes of the perceptual test, the analysis then points at the observation of the durational parameters of rhotic consonants to determine whether, in words perceived as “degeminated,” a reduction in rhotic duration is observed compared to those perceived as geminated in light of a set of variables: i) social (gender, age, occupation, education level, place of residence, and geographic mobility); ii) linguistic (word and rhotic type, phrase-final or non-final position); iii) psycholinguistic (presence of accommodation to the interviewer, Spinelli). The aim of this investigation is to determine whether and to what extent these variables play a role in rhotic consonant duration reduction. Likewise, the duration of the preceding vowel, considered in the literature as a secondary marker of degemination, was also examined. In this case the focus is on whether vowel lengthening occurs as a compensatory strategy in response to consonant shortening as to compensate the overall syllable duration.
Finally, statistical tests employing linear mixed models are conducted to confirm the relevance of certain social variables in predicting the occurrence of degemination. The analysis of duration values confirms the presence of degemination among the subjects who participated in Angela Spinelli’s interviews. In particular, this phenomenon is mainly conveyed by the shortening of the duration of the rhotic consonant rather than the lengthening of the vowel preceding the consonant. Regarding the type of rhotic, for short consonantal durations the most frequent realization is as taps. In contrast, those realized as trills tend to appear in contexts with longer consonantal durations. Concerning the influence of psycho-social variables, statistically significant results on degemination of /rr/ emerged for the sex and employment parameter, presenting the phenomenon as male speakers who have held agricultural-related jobs as opposed to female speakers employed in factory-related jobs. Mobility and level of schooling showed no significant differences between the groups. Concerning collection points, no trend emerged that differentiated significant variations in rhotic duration between locations further north or further south in the investigated area. Finally, for the other variables, no significant results emerged for accommodation. However, due to the limited and unbalanced nature of the sample, the “significant” results obtained should be interpreted as indicative of a trend rather than as definitive findings.
In conclusion, despite the inevitable issues related to the reuse of data collected in analog format and for other purposes (such as non-optimal audio quality and sometimes difficulties retrieving sociolinguistic information), this research project shows that the reuse of a historical oral archive to shed light on past phonetic phenomena is possible.