engleski

Lingual coarticulation in voiced and voiceless postalveolar fricatives in cochlear implant users: EPG evidence from Croatian

There are several reasons for investigating coarticulation in atypical speech - it can reveal new insights into typical sensorimotor behaviour, enhance existing coarticulatory models and theories and finally, improve the diagnosis and the treatment of atypical speech (Hardcastle & Tjaden, 2008). Researchers have so far focused on several groups of patients and hearing-impaired persons are one of the most frequently targeted groups in this respect [1, 2]. The common finding is that prelingually deaf persons with cochlear implants (CI) might have different gestural organization from persons with typical hearing (TH). The nature and details of that difference is still largely unknown and further research is needed. Lingual aspects of gestural organization are the most important aspects of speech production and electropalatography (EPG) is currently the only available technique specifically designed to investigate linguopalatal contact patterns [3, 4]. Therefore, in this investigation we use EPG to investigate lingual coarticulation in voiced and voiceless postalveolar fricatives /ʃ/ and /ʒ/ in prelingually deaf CI users. We shall compare the results with previously published findings on TH persons [5, 6]. It is our hypothesis that analysing coarticulation in speech sounds which require complex gestural organization and auditory control will reveal some differences in the nature of gestural organization between CI and TH persons. Research into speech of CI users is challenging in several important aspects – small and heterogeneous groups, which are hard to identify and recruit. This investigation is no different in this respect, so we treat each speaker as a separate experiment. Three prelingually deaf CI users participated in this investigation. All three participants were female and they were native speakers of Croatian, aged 19 (I1), 23 (I2) and 27 (I3). They differed with respect to age at implantation (I1=3 ; I2=5, I3=10) and age at rehabilitation (I1=2 ; I2=5.7 ; i3=3.6) and speech audiometry results (I1=90% at 45 dB, i2=70% at 65Db, I3=60% at 50dB). Target sounds were transcribed by broad phonetic transcription by the two authors independently. The agreement in judgements was well above 90% and all cases in which the judgements differed were discussed and re-transcribed. Speech material was collected by means of a modified map-task experiment whereby quasi-spontaneous speech was elicited. In such a task speakers repeated each fricative (/ʃ/ and /ʒ/) four times in three symmetrical vowel contexts (/ii/, /aa/, /uu/). Recording, annotation and data analysis were performed via Articulate Assistant software ([7] Articulate Instruments Ltd., 2008), while statistical analysis and data visualization were done via MS Excel. Four EPG parameters were measured (CoG, amount of contact, central groove, fricative duration) and they were analysed at temporal midpoint of the fricative (in order to analyse overall articulatory configuration) and at 5 equally spaced sample points throughout the fricative (in order to analyse articulatory dynamics in normalized duration). The results showed that there are observable inter-speaker differences in the way they articulate and coarticulate voiced as opposed voiceless postalveolar fricatives. These differences and their relation to participants’ hearing status will be discussed Lingual correlates of voicing difference are similar to those reported for TH persons when measured at fricative mid-point, but observable difference between CI and TH persons can be seen when articulatory dynamics is taken into account.

cochlear implant (CI), coarticulation, fricatives, electropalatography (EPG)

nije evidentirano