Acoustic correlates of word stress in Turkish converbs
Keywords: converbs, word stress, acoustics, Turkish
This study acoustically investigates word stress in Turkish manner -(y)ARAK and temporal -(y)INCA converbs using production experiment data. Analyzing f0, duration, and intensity via Linear Mixed-Effect Models (LMM), the expected findings suggest that the two converb types systematically differ in their acoustic correlates and stress realization patterns.
In Turkish, converbs function as non-finite verbal adjuncts that encode distinct semantic relations, such as manner with -(y)ARAK and temporal sequence with -(y)INCA (Sezer, 1983; Göksel & Kerslake, 2005; Bağrıaçık, 2018). Turkish converbs provide a suitable testing ground for investigating how suffixation shapes word stress (see Kabak & Vogel, 2001; Inkelas & Orgun, 2003; Kamali & Samuels, 2008; Güneş, 2015; Erguvanlı Taylan, 2015; Özçelik, 2017). Converbial suffixes (e.g., -(y)ArAk and -(y)IncA), which deviate from the default word-final stress, constitute a rich domain for examining the acoustic realization of stress in Turkish. This study aims to examine these two converb types acoustically. It investigates whether they differ in f0, duration, and intensity, proposing that converb type systematically shifts the acoustic realization of stress. To this end, the present study addresses two main research questions (RQs): (a) What are the acoustic correlates of stress in Turkish manner converb (-(y)ArAK) and the temporal converb (-(y)IncA)? (b) To what extent do these correlates differentiate stress between the two converb types in Turkish? To answer the RQs, a production experiment is conducted to examine the acoustic correlates of the manner converb (-(y)ArAK) and the temporal converb (-(y)IncA) embedded in the carrier sentences (see (1)):
Target sentences for the converb types
Manner converb
-(y)ArAk
Geçen
last
hafta
week
garson
waiter
adam-ın
man-GEN
hesab-ı
bill-ACC
öde-yerek
pay-CVB.MAN
çorba-yı
soup-ACC
iç-tiğ-i-ni
drink-NMLZ-3SG.POSS-ACC
belirt-ti
state-PST
‘Last week the waiter stated that the man drank the soup by paying the bill.’
Temporal converb
-(y)IncA
Geçen
last
hafta
week
garson
waiter
adam-ın
man-GEN
hesab-ı
bill-ACC
öde-yince
pay-CVB.TEMP
çorba-yı
soup-ACC
iç-tiğ-i-ni
drink-NMLZ-3SG.POSS-ACC
belirt-ti
state-PST
‘Last week the waiter stated that the man drank the soup when he paid the bill.’
Thirty native speakers of Turkish (24 F, 16 M) produced target sentences such as those illustrated in (1). The experiment consisted of two converb types in Turkish: manner (-(y)ArAk) and temporal (-(y)IncA). There were 120 stimuli in total: 60 target sentences (30 items × 2 converb types) and 60 fillers containing the negation marker (-mA) in Turkish. The production data were acoustically analyzed using the ProsodyPro script (Xu, 2013) package in Praat (Boersma & Weenink, 2026) for f0 (Hz), mean intensity (dB), and duration (ms). Data were analyzed using a linear mixed-effects model (LMM) in R (R Core Team, 2026) to determine the word stress pattern of converb types in Turkish affirmative sentences. Our results from the LMM showed a significant effect for the temporal converb (-(y)IncA) compared to the manner converb (-(y)ArAK). Overall results suggest that the two converb types in Turkish differ in their f0, duration, and mean intensity. Accordingly, these values were higher on the suffix than on the root in the manner converb, whereas the temporal converb did not bear a higher f0 peak on the suffix.
References
Bağrıaçık, M. (2018). Conversion in Turkish: An overview. Word Structure 11(2), 148–174. https://doi.org/10.3366/word.2018.0122
Bates, D., Mächler, M., Bolker, B. & Walker, S. (2015). Fitting Linear Mixed-Effects Models Using lme4. Journal of Statistical Software, 67(1), 1-48. https://doi.org/10.18637/jss.v067.i01.
Boersma, P., & Weenink, D. (2026). Praat: Doing phonetics by computer (Version 6.4.44) [Computer software]. https://www.fon.hum.uva.nl/praat/
Erguvanlı Taylan, E. (2015). The Phonology and Morphology of Turkish. Boğaziçi Üniversitesi Yayınevi.
Göksel, A. & Kerslake, C. (2005). Turkish: A Comprehensive Grammar. Routledge.
Güneş, G. (2015). Deriving Prosodic Structures. LOT Publications
Inkelas, S. & Orgun, C. O. (2003). Turkish stress: A prosodic constraint-based analysis. Journal of Linguistics, 39(1), 139-161.
Kabak, B. & Vogel, I. (2001). The phonological word and stress assignment in Turkish. Phonology, 18(3), 315-360.
Kamali, B., & Samuels, B. (2008). The syntax of Turkish pre-stressing suffixes (Unpublished presentation). Harvard University.
Özçelik, Ö. (2017). Prosodic Systems of Azerbaijani and Turkish: A Comparative Analysis.
R Core Team. (2026). R: A language and environment for statistical computing. [Computer software]. https://www.r-project.org/
Sezer, E. (1983). The Stress Pattern of Turkish. Cornell University Working Papers in Linguistics.