engleski

Recognition of Croatian Urban Varieties

Croatian language can be typologically classified as Slavic language and consists of three main dialects Stokavian, Kaikavian and Chakavian. The main goal of the research was to examine the recognition Croatian urban varieties. Urban varieties are considered to have features of both dialect and standard variety which can be noted on segmental and suprasegmental levels. The corpus for this research was collected among Croatian speakers and included samples from 11 different Croatian cities which represent urban centers of the three above mentioned dialects. Segmental and suprasegmental features were described and analyzed in order to determine speech patterns influencing the assessment of regionalism. 203 speakers from different cities were recorded for the purpose of this research. The criteria for speaker selection included minimization of possible influence of other language varieties, therefore the speakers were selected according to their age and belonged into two groups: younger speakers (aged from 15 to 20 who were born and raised in particular city) and older speakers (older than 50 years). The recordings were edited in Cool Edit program and speech samples used in experiment were both stylistically neutral and content controlled, meaning there were no dialectically marked expressions or words and sentences which could signal origin. Speech samples (average duration of 23 seconds) were assessed by 58 students of phonetics which had to asses degree of regionalism and recognize the regional background of the speaker. Results show that Varaždin (the continental part of Croatia) and Dubrovnik (South Dalmatia) are the most recognizable while Pula and Rijeka (Istria and Primorje) were not recognized in most cases. Samples from 20 speakers, 5 from each city (originating from Varaždin, Dubrovnik, Čakovec and Šibenik) which were all assessed as regionally marked and were regionally located correctly by all participants (the recognition of urban variety was 100% correct) were analyzed. The samples were analyzed respectively and compared with the results of speech samples from other varieties which were assessed as regionally neutral and whose regional background was not recognized. Segmental analysis included comparison of vowel and consonant systems of regionally marked and regionally neutral speakers and suprasegmental analysis included the description of prosodic features. Acoustic analysis was also conducted and included vowel formant and duration analysis. The results confirm previous research conducted on Croatian language (Škarić, 2009) showing that vowels and some consonants are the most significant for regional recognition. Formant measurements show statistically significant difference between regionally neutral and regionally marked speakers. For instance, the difference in formant values found in the speech of regionally marked female speakers from Dubrovnik and regionally neutral speakers from Rijeka show statistically significant difference for first three formants of vowel [a] - F1 [a] (p<0, 01), F2 [a] (p<0, 017), F3 [a] (p<0, 00005), second and third formant of vowel [o] (p<0, 001) and second and third formant of vowel [u] (p<0, 001). The comparison results were similar for regionally marked male speakers from Dubrovnik and regionally neutral speakers from Rijeka. The difference in formant values found for regionally marked speakers from Varaždin (the continental part of Croatia) and regionally neutral speakers from Rijeka also show statistically significant difference for all formant values measured for vowel [e]. The acoustic analysis confirms that differences between urban varieties are represented in vowel realization and that regionally marked speakers of continental part of Croatia frequently have open [] while speakers from Dubrovnik are recognized for more back and rounded realization of vowel [a] which is pronounced as []. Vowel duration measurements showed that speakers from Dalmatia (Šibenik and Dubrovnik) have a tendency to lengthen stressed vowels and realization of post accentual length. Duration of stressed vowels for speakers from continental part (Varaždin and Čakovec) and regionally neutral speakers is equal to standard values (Bakran, 1996) and there is no post accentual length. Consonants significant for assessment of regional markedness were lateral [l] and trill [r] for speakers from Šibenik. Lateral approximant [l] is pronounced as [l] both in postvocalic or prevocalic position when it is surrounded by front vowels [i] or [e] and [r] has more trills than standard Croatian [r]. The analysis of suprasegmental features shows differences between Croatian regions (especially the continental part of Croatia on one side and eastern part of Croatia and Dalmatia on the other side). Standard Croatian has four pitch accents (long-rising, short-rising, long-falling, and short-falling) which are also part of accent systems of speakers originating from Eastern part of Croatia and Dalmatia. The continental part and Istria rarely pronounce all four accents (almost never short-rising) and usually have only long and short fall accents. It has to be emphasized that in assessment of regional features prosody can be considered the least influential marker. The analysis of segmental and suprasegmental level of all regionally neutral speakers whose regional origin was not recognized showed that they belong to General Croatian pronunciation (described by Škarić, 2009)

Croatian language; forensics; recognition; accent identification; urban varieties

nije evidentirano