Prepoznavanje reči u govornoj audiometriji

Abstract

It was noticed that the standard set of words used for speech audiometry contained some archaic words, as well as words which were much more difficult to understand out of context. The first aim of this paper is to determine the words which are significantly easier or more difficult to recognize than the rest in speech audiometry at the ENT Clinic in Novi Sad (we have dedicated more attention to the incorrectly recognized words), as well as their distribution across sets containing 10 words which are used during one measurement. The second aim of the paper is to account for the errors from the point of view of linguistics and medicine. The results that we have analyzed belong to different intensity levels (5-80 dB and 25-40 dB). The research participants were 66 patients suffering from multiple sclerosis. The study has shown that there are 14 words (out of 160) whose recognition accuracy is significantly worse than that of the other words in their 10-word group. Most poorly recognized words constitute minimal pairs with some other words, and most of these words contains plosives. Even though consonants cause a higher number of errors, hearing-impaired patients sometimes misunderstand and therefore mispronounced vowel segments as well, e.g. the vowel /i/ is replaced with the vowel /u/. Another important factor which influences perception is the part of speech – nouns, adjectives and adverbs are identified more easily that other parts of speech.

Uočeno je da se u standardnom setu reči za govornu audiometriju nalaze neke arhaične reči, kao i reči koje je mnogo teže razumeti bez konteksta. Prvi cilj ovog rada jeste da odredi reči koje se znatno lak- še ili teže prepoznaju od ostalih u govornoj audiometriji na ORL klinici u Novom Sadu (veću pažnju posvetili smo pogrešno prepozna- vanim rečima), kao i njihovu raspodelu po setovima od po 10 reči koje se koriste pri jednom merenju. Drugi cilj rada je da objasni greške sa lingvističkog i medicinskog aspekta. Rezultati koje smo analizira- li pripadaju različitim nivoima intenziteta (5-80 dB i 25-40 dB). U istraživanju je učestvovalo 66 pacijenata obolelih od multiple skle- roze. Istraživanje je pokazalo da postoji 14 reči (od ukupno 160) čija je tačnost prepoznavanja znatno lošija od tačnosti drugih reči u nji- hovoj grupi od 10 reči. Većina loše prepoznavanih reči čini minimalne parove sa nekim drugim rečima, a većina ovih reči sadrži plozive. Iako konsonanti uzrokuju veći broj grešaka, ispitanici sa oštećenjem slu- ha mogu pogrešno razumeti pa izgovoriti i vokalske segmente, npr. vokal /i/ zamenjuju vokalom /u/. Još jedan bitan faktor koji utiče na percepciju jeste vrsta reči – imenice, pridevi i prilozi identifikuju se lakše od drugih vrsta reči.