Development of an Afrikaans sentence perception test based on the CUNY topic-related sentences – phase 1 : sentence perception in noise

Scourfield, Jolanda (Stellenbosch : University of Stellenbosch, 2011-03)

Thesis (MAud)--University of Stellenbosch, 2011.

Thesis

ENGLISH ABSTRACT: Speech audiometry for diagnostic purposes is widely used by audiologists around the world, but its application is starting to shift more towards rehabilitative purposes for people with hearing impairment. This has created the need for the development of appropriate test materials, including speech-in-noise tests, in the first language of the person undergoing rehabilitation. This document describes a study entailing the first phase in the development of an Afrikaans sentence perception test in noise based on the City University of New York (CUNY) topic-related sentences (Boothroyd, Hanin & Hnath, 1985). The test is called Sinslyste in Afrikaans vir Volwassenes in Lawaai [Sentence lists in Afrikaans for Adults in Noise] (SAV-L). Twenty-seven sentence lists containing 12 sentences each were compiled and evaluated by eight speakers of Afrikaans for their naturalness before they were recorded together with three-talker babble as masking noise. The recorded material was then presented to six groups of ten participants each (with the exception of nine participants in one of the rounds) and adjusted in intensity to improve inter-list reliability. Two scorers were used at intervals to determine inter-rater reliability. Test conditions were also replicated after an interval of six months or more to establish test-retest reliability. In a last round of testing, the test was administered at the intended presentation level to assess the appropriateness of the chosen level. Through adjustment of list intensities, inter-list reliability was improved to a 13.87% variance between list scores. Test-retest reliability showed a bias with an intra-class correlation agreement of 0.859. This was thought to be due to participants’ improved familiarity with the clinical environment over time, however longer term measures of test-retest reliability remains to be done. Inter-rater reliability was very high with an intra-class agreement of 0.999. The intended test level of 50dBHL with a signal-to-noise (SNR) ratio of +5dB, was found to be an acceptable fixed SNR. The test can be used in its current form for assessment of amplification candidacy as well as monitoring of progress during rehabilitation. Further research is indicated for the establishment of the sensitivity of the test. It is concluded that just as a clinician is expected to select the most suitable diagnostic test based on the patient’s history, signs and symptoms, so should the audiologist be expected to select the most appropriate rehabilitative measure based on the patient’s communication needs and established treatment goals. The SAV-L is a valuable contribution to the pool of speech perception tests available as rehabilitative measures, and is specifically suitable for adults with well-developed spoken language.

AFRIKAANSE OPSOMMING: Spraakoudiometrie vir diagnostiese doeleindes word algemeen gebruik deur oudioloë regoor die wêreld, maar die toepassing daarvan is besig om te skuif na rehabilitasie van mense met gehoorgestremdheid. Die behoefte vir die ontwikkeling van geskikte toetsmateriaal, insluitend spraak-in-geraastoetse, in die eerste taal van die persoon wat rehabilitasie ondergaan, het dus begin ontwikkel. Hierdie dokument beskryf ‘n studie wat die eerste fase van die ontwikkeling van a Afrikaanse spraakpersepsietoets in geraas behels. Dit is gebaseer op die City University of New York (CUNY) onderwerp-verwante sinne (Boothroyd, Hanin & Hnath, 1985). Die ontwikkelde toets is benoem Sinslyste in Afrikaans vir Volwassenes in Lawaai (SAV-L). Sewe-en-twintig sinslyste met twaalf sinne per lys is saamgestel en geëvalueer deur ag Afrikaans-sprekende persone om hulle natuurlikheid te bepaal voordat dit opgeneem is tesame met drie-spreker babbel as maskeringsgeraas. Die opgeneemde materiaal is aan ses groepe van tien deelnemers elk aangebied (met die uitsondering van slegs nege deelnemers in een van die rondtes) en aangepas in intensiteit om inter-lys betroubaarheid to verbeter. Twee tellers is van tyd tot tyd gebruik om inter-beoordelaar betroubaarheid te bepaal. Toetskondisies was ook ná ‘n periode van ses maande gereplikeer om toets-hertoets betroubaarheid to bepaal. In ‘n laaste rondte van toetsing was die toets geadministreer teen die beoogde toetsvlak om die geskiktheid van hierdie toetsvlak te bepaal. Inter-lys betroubaarheid is verbeter na ‘n 13.87% variase tussen sinslystellings deur die aanpassing van die intensiteite van die lyste. Toets-hertoets betroubaarheid het ‘n voorkeur met ‘n intra-klas korrelasie ooreenstemming van 0.859 getoon. Hierdie voorkeur was vermoedelik weens die deelnemers se toenemende bekendheid met die kliniese omgewing, alhoewel langer termyn metings nog gedoen moet word om hierdie tendens verder te ondersoek. Inter-beoordelaar betroubaarheid was goed met ‘n intra-klas korrelasie van 0.999. Die beoogde toetsvlak van 50dBGP met ‘n sein-tot-ruis ratio van +5dB is as ‘n geskikte ratio bevind. Die huidige vorm van die toets kan gebruik word vir die evaluasie van kandidaatskap van klankversterking sowel as monitering van vordering tydens rehabilitasie. Verdere navorsing is aangedui vir die bepaling van die toets se sensitiwiteit. Daar is afgelei dat net soos daar van ‘n klinikus verwag word om die mees geskikte diagnostiese toets te kies gebaseer op die pasiënt se geskiedenis, tekens en simptome, net so moet daar van die die oudioloog verwag word om die mees geskikte rehabilitasiemeting te selekteer, gebaseer op die pasiënt se kommunikasiebehoeftes en vasgestelde behandelingsdoelwitte. Die SAV-L is ‘n waardevolle hulpmiddel in die versameling spraakpersepsietoetse wat beskikbaar is vir rehabilitasiemetings en is spesifiek geskik vir volwassenes met goed-ontwikkelde gesproke taal.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/6584
This item appears in the following collections: