نقش عوامل صوتی در ایجاد تمایز معنایی بین افعال ماضی ساده و ماضی نقلی فارسی

در این مقاله عوامل صوتی موثر در ایجاد تمایز آوایی بین افعال ماضی ساده و ماضی نقلی فارسی را بررسی کردیم. صورت آوایی افعال ماضی ساده و نقلی فارسی در گفتار محاوره‌‌ای به‌لحاظ ساختِ زنجیری کاملاً شبیه یکدیگر است و فقط از نظر محل وقوع تکیه با یکدیگر متفاوتند. در مطالعه‌ای آزمایشگاهی الگوی توزیع مقادیر پارامترهای صوتی مهمِ نوای گفتار مانند دیرش، شدت انرژی و فرکانس پایه را در صورت آوایی افعال ماضی ساده و نقلی در پیکره‌‌ای شامل 24 جمله بررسی کردیم. نتایج به‌دست‌آمده نشان داد هیچ‌کدام از پارامترهای انتخاب‌شده نمی‌‌تواند به‌‌‌طور منظم و پایدار الگوی نوایی افعال ماضی ساده و نقلی را از یکدیگر متمایز کند. پس از هنجارسازی مقادیر فرکانس پایه و محاسبۀ میانگین تغییرات زیروبمی در سطح تمامی داده‌‌ها به ازای هر گوینده دریافتیم آنچه اساساً باعث تمایز صوتی صورت آوایی افعال ماضی ساده و نقلی می‌‌شود مقادیر محلی قله و دره f0 در سطح هجاهای هدف این افعال است. نتایج تحلیل‌‌های آماری با تایید این واقعیت نشان داد فرکانس پایه‌ محلی (ارتفاع قله) پارامتر صوتی معتبر و باثباتی است که صورت آوایی افعال ماضی ساده و ماضی نقلی را بر روی هر دو محور جانشینی و هم‌نشینی از یکدیگر متمایز می‌‌کند.

The Role of Acoustic Parameters in Distinguishing Persian Simple Past and Present Perfect Tenses

AbstractThis paper addressed the acoustic factors involved in distinguishing simple past and present perfect tenses in Persian. The pronunciation of Persian simple past and present perfect tenses in colloquial speech are the same segmentally but different in terms of the position of stress. In an experimental study, the pattern of distribution of some important prosodic parameters, including F0, intensity, and duration, was investigated in a speech corpus consisting of 24 sentences. Results suggested that none of the study parameters could differentiate simple past and present perfect tenses reliably and consistently. After normalizing F0 and computing the average pitch for all acoustic data per speaker, it was found that it is the value of F0 peaks and valleys in the target syllables that makes a fundamental distinction between simple past and present perfect. Results of statistical tests confirmed this finding, suggesting that the local F0 value is a reliable and consistent parameter that distinguishes simple past from present perfect in both the syntagmatic and paradigmatic axes.Keywords: Simple past, Present perfect, Local F0, Duration, Intensity IntroductionThis paper addressed the acoustic factors involved in distinguishing simple past and present perfect tenses in Persian. The pronunciation of Persian simple past and present perfect tenses in colloquial speech are the same segmentally but different in terms of the position of stress.Stress is a linguistic property of a word, specifying which syllable is stronger in the word compared to any others. Early studies, such as Fry (1955, 1958), Lieberman (1960), Beckman (1986), Harrington, Beckman, and Palethorpe (1998) (see also Laver, 1994 for an overview) have shown that there are clear acoustic differences between stressed and unstressed syllables: stressed syllables are realized with a higher pitch, higher intensity, longer duration, and more peripheral vowel quality than unstressed syllables. Studies in many stress accent languages show that the realization of a stressed syllable differs from the unstressed syllable due to having a higher pitch. Also, results have shown that speakers consistently used duration to distinguish between open and central vowels having contrastive stress at the word level, while contrastive stress between open vowels at the phrase level was mainly accompanied by the intonational prominence contrast. Compared to F0 and duration, the relation of intensity variation needs to be discussed further in the speech signal to word stress.On the one hand, previous work has generally emphasized that intensity manipulations prove much weaker cues than duration in perceived stress (Fry, 1955, 1958; Turk Sawusch, 1996, for English; Mol Uhlenbeck, 1956, for Dutch). Furthermore, several different operationalizations of intensity, such as intensity summed over time (Beckman, 1986) and spectral tilt (i.e., the degree to which intensity changes as frequency increases) (Sluijter van Heuven, 1996a; Sluijter, van Heuven, Pacilly, 1997), have been shown to be consistent correlates of stress. For example, Sluijter and van Heuven (1996) argue that previous study has demonstrated that, on the one hand, loudness variation virtually inconsequential for perceived stress is typically based on analyses that do not distinguish between word stress and, on the other hand, pitch accentinduced prominence. They note that if a more accurate measure of intensity is used, the traditional account of stress as a local increase in loudness seems justified. Materials and MethodsIn an experimental study, the pattern of distribution of some important prosodic parameters, such as F0, intensity, and duration, was investigated in a speech corpus consisting of 24 sentences. The target words in such sentences were simple past and present perfect tenses that would be produced with a flat pitch melody characterizing postfocal accent, and that the phraselevel accent was far away from the target word so that no effects of the accent would be observed there.All acoustic measurements were taken using the speech software Praat (Boersma Weenink, 2020). In each word, syllable and vowel boundaries were manually identified and annotated as the measurement intervals. Segmentation criteria were based on both waveform and spectrogram cues, as suggested by Peterson and Lehiste (1960). Measurements of all acoustic variables were made automatically using ProsodyPro (Xu, 2020). Discussion of Results and ConclusionsResults suggested that none of the parameters selected could differentiate simple past and present perfect tenses reliably and consistently. After normalizing F0 and computing the average pitch for all acoustic data per speaker, it was found that it is the value of F0 peaks and valleys in the target syllables that makes a fundamental distinction between simple past and present perfect. Results of statistical tests confirmed this finding, suggesting that the local F0 value is a reliable and consistent parameter that distinguishes simple past from present perfect in both the syntagmatic and paradigmatic axes.