Acoustic Correlates of Parental Role and Gender Identity in the Speech of Expecting Parents

Melanie Weirich, Adrian P. SimpsonInterspeech 2017

TL;DR

This study explores how expecting parents' speech varies not just by gender, but also by their self-perceived femininity and their planned parental leave duration. The findings suggest that these social factors, beyond biological sex, influence acoustic speech characteristics like vowel space size and sibilant contrast, particularly in male speakers.

What Problem Does This Paper Address?

While much research has focused on acoustic differences between male and female speech, this paper delves into the subtler variations within genders. Specifically, it investigates how social factors, such as gender identity (how feminine individuals perceive themselves) and parental role (the planned duration of parental leave), might contribute to these intra-gender acoustic differences. The authors aim to disentangle biologically driven sex differences from socially influenced gender-specific variations in speech.

Methods

The researchers analyzed the speech of 29 expecting parents (14 females, 15 males) from Jena, Germany. They used a picture description task to elicit specific target words containing vowels and sibilants. Acoustic analyses focused on two key areas:

  • Vowel Space Size: Measured by calculating the area of the polygon formed by the mean formant frequencies (F1 and F2) of the vowels /i:, ɛ, a, u:/.
  • Sibilant Contrast: Quantified using Discrete Cosine Transformation (DCT) coefficients to analyze the spectral shape of the sibilants /s/ and /ʃ/. Euclidean Distances in the DCT1xDCT2xDCT3 space were calculated for paired /s/-/ʃ/ productions.

Gender identity was assessed using the German version of the Extended Personal Attributes Questionnaire (GEPAQ-F+), which measures self-ascribed positive feminine attributes. Parental role was defined by the planned duration of parental leave.

Datasets

The study utilized speech data collected as part of the ongoing longitudinal AVARE project. The dataset consists of adult-directed speech from 29 German speakers (expectant mothers and fathers) recorded before the birth of their child. No new benchmark datasets were introduced.

Key Insights and Ideas

  • Beyond Biological Sex: The study highlights that social factors like gender identity and parental role can influence speech acoustics, contributing to variations beyond those explained by biological sex differences.
  • Fathers' Speech as a Bridge: Fathers planning to take extended parental leave showed acoustic characteristics (vowel space size and sibilant contrast) that fell between those of mothers and fathers who planned to return to work sooner. This suggests a potential influence of caregiving roles on speech.
  • Femininity and Male Speech: A particularly interesting finding is the positive correlation between self-ascribed femininity and larger vowel space sizes in male speakers. This implies that men who identify more strongly with feminine attributes may exhibit acoustic features traditionally associated with female speech.
  • Sibilant Contrast Nuances: While mothers generally showed the largest sibilant contrast, and control fathers the smallest, the fathers planning parental leave fell in between. A tendency for a positive correlation between femininity and sibilant contrast was observed in males, though it did not reach statistical significance.

Key Results

  • Vowel Space Size: Expectant mothers had significantly larger vowel space sizes than control fathers. Fathers planning parental leave had vowel space sizes intermediate to mothers and control fathers.
  • Gender Identity and Male Vowel Space: For male speakers, a significant positive correlation was found between self-ascribed femininity scores and vowel space size (r = 0.54, p < .05).
  • Sibilant Contrast: Mothers exhibited a significantly larger acoustic sibilant contrast than control fathers. No significant differences were found between the father groups or between fathers and mothers.
  • Gender Identity and Male Sibilant Contrast: A non-significant tendency for a positive correlation was observed between self-ascribed femininity and sibilant contrast in male speakers (r = .38, p = .07).

Limitations and Future Work

The authors acknowledge that the GEPAQ F+ score is a limited measure of gender identity and that planned parental leave doesn't necessarily reflect actual involvement in childcare. Future work within the larger AVARE project will involve gathering more detailed information on childcare involvement and conducting listening experiments to assess how listeners perceive masculinity/femininity based on these acoustic variations.

Summary updated

BibTeX
@inproceedings{weirich2017acoustic,
  author = {Melanie Weirich and Adrian P. Simpson},
  title = {{Acoustic Correlates of Parental Role and Gender Identity in the Speech of Expecting Parents}},
  booktitle = {Interspeech},
  year = {2017},
}
Figure