Back to listing

ANU-CoEDL ZOOM Seminar: Production and perception of prominence in Drehu – a (psycho)acoustic study, Catalina Torres, 5 June

Australian National University

Date: 26 May 2020

Seminar: Production and perception of prominence in Drehu – a (psycho)acoustic study

Speaker: Catalina Torres, University of Melbourne

When: 5 June 2020, 3pm

Where: via zoom  (please email for zoom link invitation)


This presentation discusses methodological challenges fieldworkers may encounter when investigating phonological stress in an under-described language. The example of the language Drehu (Austronesian) is taken and suggestions on how to overcome potential biases are presented. Incongruencies between previous impressionistic descriptions and current acoustic analyses of prominence marking are presented. Additionally, methodological biases that can influence fieldworkers are discussed. In this talk, the current work on Drehu, as well as other considerations reported in the study of word phonology are taken into account (Röttger & Gordon, 2017). I argue that experimental methods are required for the investigation of word prosody. Firstly, detailed instrumental studies are necessary to determine relevant acoustic cues. Secondly, a perceptual investigation is required to validate hypotheses based on phonetic observations. 

Drehu represents an interesting case in the study of word prosody since early records report an unusual typological pattern within Austronesian languages. The first impressionistic descriptions claim there is word-initial stress (Lenormand, 1954; Tryon, 1968). In a survey on word accentual patterns in the Austronesian language family, 117 languages were evaluated, and it was found that in 89% of the cases, the stress domain is located at the right edge of the word. Further, it was found that in 75% stress is restricted to the penultimate or ultimate syllable (van der Hulst, Goedemans, & van Zanten, 2010). 

Two production studies and one perception experiment were carried out during fieldwork in Lifou. The speech materials of these experiments were designed to allow for teasing apart word-level stress from phrase-level prominence and investigate acoustic correlates. Experiment I examined how nouns in informational focus were realised in three different positions in carrier phrases, using a word insertion task. Four adult female speakers were recorded for this purpose. Experiment II aimed to test whether nouns embedded in fully-fledged sentences, in non-focal condition, would either show evidence for word-initial or final prominence. The study used token words placed phrase medially in two positions, and which were preceded by at least one monosyllabic function word. Twelve adolescent speakers were recorded. Results of Experiment I show that non-utterance-final tokens generally start with a low tone, and that there is an F0 rising movement towards the right edge. The tonal patterns LH, LHL1H1, LL1H, and HLH1 were found in 98% of the data. Experiment II shows that despite the manipulation of position and focus the initial low and final high tone are predominantly used on nouns. In Experiment III twenty-two Drehu listeners were invited to a forced-choice lexical decision task. Speech sequences were acoustically manipulated to investigate preferred duration patterns in word recognition.  

Acoustic and statistical analyses indicate that the final syllable of the word is prominently marked through means of duration and the modulation of fundamental frequency (F0). Results of the lexical decision task validate the relevance of durational cues at the right edge. These findings are incongruent with previous impressionistic descriptions. The lack of acoustic evidence for word-initial stress and results of three experiments suggest there is edge-prominence which is marked on the right side. Data from an under-described language like Drehu contributes to the understanding of typological variation found in the intonational phonology of the languages of the world. This study also provides evidence for right-edge prominence marking in the language which hadn’t been reported before. 


Lenormand, M. H. (1954). La phonologie du mot en lifou (îles Loyalty). Journal de la Société des Océanistes, 10(10), 91-109.

Roettger, T., & Gordon, M. (2017). Methodological issues in the study of word stress correlates. Linguistics Vanguard3(1).

Tryon, D. T. (1968). Dehu grammar.

van der Hulst, H., Goedemans, R., & van Zanten, E. (Eds.). (2010). A survey of word accentual patterns in the languages of the world. Walter de Gruyter.

  • Australian Government
  • The University of Queensland
  • Australian National University
  • The University of Melbourne
  • Western Sydney University