The beauty in a beast: Minimising the effects of diverse recording quality on vowel formant measurements in sociophonetic real-time studies

Abstract

Sociophonetic real-time studies of vowel variation and change rely on acoustic analyses of sound recordings made at different times, often using different equipment and data collection procedures. The circumstances of a recording are known to affect formant tracking and may therefore compromise the validity of conclusions about sound changes made on the basis of real-time data. In this paper, a traditional F1/F2-analysis using linear predictive coding (LPC) was applied to the vowels /i u a/ extracted from spontaneous speech corpora of Glaswegian vernacular, that were recorded in the 1970s and 2000s. We assessed the technical quality of each recording, concentrating on the average levels of noise and the properties of spectral balance, and showed that the corpus comprised of mixed quality data. A series of acoustic vowel analyses subsequently unveiled that formant measurements using LPC were sensitive to the technical specification of a recording, with variable magnitudes of the effects for vowels of different qualities. We evaluated the performance of three commonly used formant normalisation procedures (Lobanov, Nearey and Watt-Fabricius) as well as normalisations by a distance ratio metric and statistical estimation, and compared these results to raw Bark-scaled formant data, showing that some of the approaches could ameliorate the impact of technical issues better than the others. We discuss the implications of these results for sociophonetic research that aims to minimise extraneous influences on recorded speech data while unveiling gradual, potentially small-scale sound changes across decades.

Remote sociophonetic data collection: Vowels and nasalization from self‐recordings on personal devices

“…If the measure of interest is systematically affected by a certain type of device, adjustments could be calculated on participant data to improve comparability across speakers. If speakers make multiple recordings in different sessions, ask them to use the same device each time, and instruct them on device placement and minimizing background noise to increase both audio quality and comparability across sessions. Environmental noise was not a factor in the current study, but it varies considerably in natural settings and could be problematic for many types of acoustic measurements (e.g., De Decker, 2016; Rathke et al., 2017). Consider other creative solutions like mailing participants equipment, instructing them on how to disable speech‐processing apps on their device before recording, asking them to record on two devices simultaneously, or sending them a small device like an inexpensive keychain loaded with a short audio clip containing multiple variations of the measurements of interest and asking them to play it aloud before recording.…”

Linguistic fieldwork in a pandemic: Supervised data collection combining smartphone recordings and videoconferencing

“…Measurements of mean F0 and SCPP (smoothed cepstral peak prominence) proved relatively robust across devices, but jitter and shimmer did not. The findings of Jannetts et al (2019) and Rathcke et al (2017) indicate that ideally, device specifications should be the same for all participants. This poses a problem for projects such as SDATS, as there are currently more than 20 major vendors within the smartphone industry (Metodiev 2019).…”

“…They found that recordings on a first-generation iPhone were suitable for formant analysis, as the overall shape of the vowel space does not seem to be affected substantially. More recently, Rathcke et al (2017) studied the effects of different equipment on F1 and F2 measurements in real-time sociolinguistic analyses, reporting that formant measurements using LPC tracing are sensitive to the technical specifications of the microphones used at different points in time. Their key recommendation is that linguists should keep the recording equipment controlled in order to maximize comparability between recording sessions.…”

Sharing innovative methods, data and knowledge across sociophonetics and forensic speech science

“…Some sociophonetic studies have considered the effects of poor quality recordings on acoustic analysis methods (e.g. Rathcke et al 2016), and this research is incredibly valuable to practitioners who need to assess what methods are suitable given the recording characteristics. Sociophonetic data collection is, of course, conducted according to the requirements of the study.…”