Maria Cecilia Bevilacqua* Marcos R. Banhara* Everardo Andrade Da Costa § Adriana Braga Vignoly § Ka´tia F. Alvarenga* *Universidade de Sa˜o Paulo (USP), Bauru, Sa˜o Paulo, Brazil § Universidade Estadual de Campinas (UNICAMP), Campinas, Sa˜o Paulo, Brazil Short Report International Journal of Audiology 2008; 47:364365 The Brazilian Portuguese Hearing in Noise Test Introduction The hearing in noise test (HINT) is currently available in several languages (Nilsson et al., 1994). The similarity of the procedures and the resulting similarity of the test materials makes the measures obtained with the HINT comparable across languages. Individuals from both genders, in the age range of 1850 years, with hearing thresholds 20 dB HL or better at the octave frequencies from 250 to 8000 Hz, for both ears, who were native speakers of the Portuguese spoken in Brazil participated in this study. Method and Results Preparation of test materials Initially, 1700 sentences were compiled via three different methods: (1) A survey of the corpus of Brazilian Portuguese available in the CHILDES (child language data exchange system) database was completed (Murari, 2004). The 50 000- word list was assessed by speech pathologists from various states of Brazil, and the words agreed upon by all participants were considered*regional vocabulary that might have a confusing meaning in some regions was excluded (Costa, 1997). (2) The Latin American Spanish HINT materials were analysed. (3) Spontaneous dialogue with daily phrases and vocabulary for subjects whose native language is Brazilian Portuguese, regard- less of their background or education level, was recorded. After the sentences were selected, 10 subjects were asked to assess the familiarity of the sentences, grading them on a scale of 0 to 7, scoring the most familiar and natural as 7 and the most artificial as 0. Finally, of the 1700 sentences, the sentences with scores of 6 and 7 were further evaluated, and 800 were selected for use in the study. Two centers participated in the research: Universidade de Sao Paulo, Bauru (USP) and Universidade Estadual de Campinas, Campinas (UNICAMP). The selected sentences were recorded at House Ear Institute, Los Angeles by a professional Brazilian voice actor. Estimation of the performance-intensity function Twelve subjects were evaluated (six at USP and six at UNI- CAMP) in order to estimate the performance-intensity (PI) function. From the existing speech material, six lists of 50 sentences, totaling 300 sentences (three lists for each center), and a masking noise with the same sentence frequency spectrum were produced. Noise intensity was constant at 65 dBA, with the following variations of signal to noise ratios (S/N ratios): 7 dB, 4 dB, and 2 dB, for all lists tested. The correct words were counted, and this number resulted in an intelligibility percentage, for each list, at every S/N ratio. The average slope of the PI function for the two centers was 11.4% per dB change in S/N ratio. Equalization of sentence difficulty Sentence equalization from the S/N ratio corresponding to 70% of speech intelligibility was subdivided into three rounds with 20 subjects for each round (10 from USP and 10 from UNICAMP). For each round, 16 lists containing 50 sentences each, totaling 800 sentences (eight lists for each center), were used. On the first round, the noise intensity was fixed at 65 dBA, and the S/N ratio for all sentences presented was 6.5 dB. Sentences were scaled to obtain approximately equal intelligibility in the presence of the noise. Final S/N ratios varied over a 6-dB range, from 3.5 dB to 9.5 dB. Upon the completion of the third round, 74% of the sentences had been scaled within 92.5 dB of the mean S/N ratio to achieve intelligibility scores between 55% and 85% correct, and 94% of the sentences scaled within 93.0 dB of the mean exhibited intelligibility scores within this range. Formation of sentence lists From the data of the equalization phase, 24 lists of phoneti- cally balanced sentences were created. Coded phonemic tran- scriptions were entered into a computer program that automatically counted the phonemes in each sentence of each list. The final result was the creation of 24 lists of 10 ISSN 1499-2027 print/ISSN 1708-8186 online DOI: 10.1080/14992020701870205 # 2008 British Society of Audiology, International Society of Audiology, and Nordic Audiological Society Received: December 12, 2007 Accepted: December 17, 2007 Maria Cecilia Bevilacqua Centro de Pesquisas Audiolo´gicas, Universidade de Sa˜o Paulo, Rua Silvio Marchione, 320 Bauru, Sa˜o Paulo, Cep: 17.012-900, Brazil. E-mail: cecilia@implantecoclear.com.br