Received:05,Apr,2016 Journal of Multidisciplinary Scientific Research, 2016,4(2):04-08 ISS: 2307-6976 Available Online: http:jmsr.rstpublishers.com PHOETIC-ISTRUMETATIO OF BALA ASPIRATIO: A SPECTRORAPHIC AALYSIS. Linguistics & Contemporary English,The English and Foreign Languages University, Hyderabad, India. Email: moumita.singha90@gmail.com Accepted:27,Apr,2016 Abstract The aim of this experiment is to prove the phonemic nature of the phenomenon called aspiration in Bangla, which is a distin ctive feature when compared to English, where aspiration is allophonic. r. Binoy Barman had written an article on istinctiveness of Aspiratio n in Bangla, [http:dspace.library.daffodilvarsity.edu.bd:8080] where he had listed 11 minimal pairs but hadn t proved the di stinctiveness through phonetic instrumentation (spectrographic analysis) or by giving the phonetic correlates of the same. Therefore I have taken this up to prove the same by giving the phonetic correlates through instrumental phonetics and laboratory phonetics. By comparing the minimal pairs in Bangla, depending upon the place of articulation, position of the unaspiratedaspirated segments, length and duration of the precedingfollowing segments (vowels, nasals or other consonants), etc, I wish to prove that there are differences between aspirated & unaspirated segments. On the basis of these differences, I will try to prove that these differences lead to the different measurements in the length, duration of unaspir atedaspirated sounds; their preceding & following sounds, vowel length & formant types, and therefore they are distinct phonemes in Bangla. As we know, in a minimal pair, in the same environment (parallel distribution), substitution of one sound with the other if creates difference in mean ing, then those 2 sounds are phonemes. Similarly, I will try to prove with examples & analysis that in a minimal pair, in the same environment the differences between an aspirated & unaspirated sound (in terms of their individual length, duration & length-duration of their preceding & following sounds) makes a contrastive distribution between the two, thereby rendering them as two distinct phonemes. Keywords: Aspiration, istinctiveness, Minimal Pair, Phonemes, Spectrogram ITROUCTIO The phonetic correlates of the distinctive feature aspiration in Bangla and to prove the reality of distinctiveness of the phenomenon in our target language. The fact that aspiration is phonemic in Bangla and not allophonic like English will be proved in this experiment. I wish to give the phonetic instrumentation of the minimal pairs and analyze them through sound spectrogram and hence make an attempt to prove the distinctiveness. Instrumentation Since the aim is to prove distinctiveness of the phonemes (unaspiratedaspirated), the length and duration of the segments are necessary for this establishment. Sound Spectrograph is the convenient instrument for the above mentioned purpose. Therefore, I have chosen Sound Spectrograph for the instrumentation of my experiment. This is because Sound Spectrograph gives a clear image in the form of spectrums, where vertical striations are clearly shown. This helps us to get the information about accurate temporal or accurate duration point view. From the spectrum its convenient to measure the length in a graduated scale in cms then converting it into mms, then comparing with the total duration (seconds converted to milliseconds) that we obtain from the spectrogram, we calculate the duration and length of individual soundssegments. Subjects and text Since the sample spectrograms obtained by my own voice recording gave clear, satisfactory images, according to my respected instructor r. B. A. Prabhakar Babu, our department and instructor advised me to be the subject of my experiment. Apart from myself other 6 native speakers of Bangla belonging to several districts of Bengal, comprising of 3 male and 3 female were the subjects too. Once the subjects, who after being seated at a table, indicated readiness to begin recording, the microphone was set up and adjusted as necessary and specific verbal instructions similar to the following were given: They had to read this list of word pairs three times with a normal speaking rate and in a clear manner. You will not be evaluated on your performance; the researcher is simply interested in how you normallynaturally speak. In order to get a clear recording, subjects needed to speak loudly and clearly to make sure that they were close to the microphone, and speaking in direct line with it. Once the technical assistant was satisfied with these various details, recording was begun and all the subjects were requested to read through the list of word pairs one time, and then to re-read them necessary until it was decided that the data elicited was satisfactory. At this point, the recording was stopped, and the subjects were given any necessary feedback regarding the performance of the task. This same process was used for the second, third, and any subsequent recordings as well. Almost all of the recordings were done in a moderately quiet Phonetics Laboratory of The English & foreign Languages University, in Hyderabad, Andhra Pradesh, using a professional, unidirectional, dynamic microphone plugged into the microphone jack of the Creative SoundBlaster sound card of a desktop computer. The microphone was either held by the subject or positioned on a microphone stand, such that it was 1 3 inches from the lips and in direct line with the airflow from the mouth. Professional microphones are generally designed with a low sensitivity or signal level that assumes the use of an amplifier, whereas computer sound cards are normally designed according to the consumer standard that is used for accessories like computer microphones that are usually plugged directly into other electronic devices. The desktop was connected to Sound Spectrograph, which broke down the recorded sounds into its components (spectrum) and gave us the picture (spectrogram). I took the Broad Band Spectrogram for the recorded sounds. It gave me the information about accurate temporal or duration point of view.
Journal of Multidisciplinary Scientific Research, 2015,4(2):04-08 5 The vertical striations were very prominent. All of the recordings were saved and retained in the WAV audio format for all archival and analysis purposes. Text I had 11 minimal pairs (a total of 22 words), where the contrasting phonemes were categorized according to their place of articulation, manner of articulation and voicing. The words in individual pairs contrasted with each other according to the type: unaspirated or aspirated. Also the unaspiration or aspiration varied according to their specific positions (initial or medial) in individual word pairs. The subjects read out each minimal pair out of the 11 minimal pairs in a single go, keeping in mind to maintain a pause between words of individual pairs. For eg: The word d a:n and d ʰa:n were uttered together maintaining a pause of 1 sec between both the words. This was done intentionally to maintain a specific pattern of utterance, thereby retaining the clarity of spectrograms so that neither of the two spectrograms overlaps each other. If a pause isn t maintained between two words of a minimal pair, then there will be confusion in interpreting the overlapped spectrums of the spectrogram of both words. This will lead to faulty measurements of length and calculation of duration of individual phonemes of each word. Only after every perfect utterance, the technical assistant recorded and did the spectral analysis, converted into WAV file, saved it for further investigations (if necessary), and took the print out of individual word spectrograms. As soon as I got each print out, I wrote my name, marked the total duration and denoted the word beneath the spectrogram and the same task was followed by my other subjects too who denoted the word beneath the spectrogram, wrote their names and duration so that all the spectrograms had unique labeling and no confusion. This was done immediately, so that neither of the spectrograms were mistook for the other. Labeling the words with its duration is one of the most important tasks that an analyzer should do, which is again something my instructor taught me. ATA It is listed similar words used by r. Binoy Barman in his article istinctiveness of Aspiration in Bangla [http:dspace.library.daffodilvarsity.edu.bd:8080] having 11 pairs of sounds of Bangla, which were categorized like: 1. Voicelessvoiced 2. Unaspiratedaspirated 3. Initialmedial positioned Bangla Aspiration Consonant type according to Voiceless Voiced manner & place of articulation Aspirated Aspirated BILABIAL PLOSIVES pʰ bʰ ETAL PLOSIVES t ʰ d ʰ ALVEOLAR PLOSIVES tʰ dʰ PALATAL PLOSIVES tʃʰ dʒʰ RETROFLEX ---- ɽʰ VELAR PLOSIVES kʰ gʰ This listing is done to classify the entire list of plosives in Bangla phonology and to examine and evaluate them to make sure that aspiration plays an important role to create meaning difference in the minimal pairs and categorize each pair as different phoneme bearing words. In the experiment two positions of the aspiratedunaspirated sounds have been considered, viz., initial and medial. In these two positions, the length and duration have been examined. Apart from these sounds, their preceding and following vowel length and duration has also been examined for proper analysis. Coming to the diacritics, for aspirated sounds, a superscripted h is used, eg. Aspirated - p h and for unaspirated sounds - p is used. The sounds are closed by in order to denote them as two different phonemes. Before evaluating the length and duration of segments (aspiratedunaspirated, precedingfollowing vowels) in initialmedial positions, a brief discussion on aspiration in six different types of plosives is given: Bilabial plosives: In the utterance of bilabial plosives of Bangla, the two lips come in close contact, press each other for a complete closure and there is sudden release with plosion of air. The voiceless and voiced bilabial plosives are: p, pʰ, and b bʰ respectively. We can notice in the following examples, how aspiration in each pair creates semantic difference [pak] mud h [p ak] gap [ba:n] flood [bʰa:n] pretension [ripu] enemy h [rip u] cloth hole repair [nɔbο] new [nɔbʰο] sky ental plosives: In the utterance of dental plosives, the tip of the tongue comes in contact with the back of upper teeth thereby there is sudden air release with plosion. The voiceless and voiced dental plosives are: t, t ʰ, and d d ʰ respectively. We can notice in the following [t ala] lock [mat a] mother [t ʰala] plate [mat ʰa] head [d a:n] donation [ad a] ginger [d ʰa:n] paddy [ad ʰa] half Alveolar plosives In the utterance of alveolar plosives the tip of the tongue comes in contact with the alveolar ridge and thereby there is sudden air release with plosion. The voiceless and voiced alveolar plosives are: t, t ʰ, and d d ʰ respectively. We can notice in the following Medial position [tɔk] sour [pitα] hit (Imperative) [tʰɔk] cheat [pitʰa] cake [da:k] callpost [dʰa:k] drum Velar plosives In the utterance of velar plosives the back of the tongue comes in contact with the velum or soft palate and thereby sudden release of air with plosion occur. The voiceless and voiced velar plosives are: k, kʰ, and g gʰ respectively. We can notice in the following
6 Initialposition: [kɔr] tax [pa:ka] brick-built [kʰɔr] straw [pa:kʰa] fan [gɔnο] people [ʃɔngο] companion [gʰɔnο] thick [ʃɔngʰο] association Palatal plosives: In the utterance of palatal plosives the mid portion of the tongue and the hard palate comes in close contact, and thereby there is a sudden release of air with a plosion. The voiceless and voiced palatal plosives are: tʃ, tʃʰ, and dʒ dʒʰ respectively. We can notice in the following examples, how aspiration in each pair creates semantic difference [tʃαi] Iwe want [mutʃi] shoe-maker [tʃʰαi] ash [mutʃʰi] Iwe rub [dʒαl] netforged [bodʒa] closed [dʒʰαl] hot tasting [bodʒʰa] load Retroflex plosives: In the utterance of retroflex plosive the hard palate is in contact with the curled back tongue, thereby a sudden release of air with plosion occur. In Bangla only one pair of retroflex voiced plosive: ɽ and ɽʰ is found which occurs on medial position. [ga:ɽo] to dig [ga:ɽʰo] deepdense Interpretation of the data: Table (1)Lengths of the unaspiratedaspirated segments in initial & medial positions UASPIRATE ASPIRATE UASPIRATE ASPIRATE IITIAL length IITIAL length MEIAL length MEIAL length p 10 pʰ 48 p 46 pʰ 67 t 10 tʰ 37 t 44 tʰ 44 k 10 kʰ 45 k 36 kʰ 61 b 45 bʰ 52 b 30 bʰ 48 d 43 dʰ 60 d ---- dʰ ---- g 41 gʰ 62 g 7 gʰ 26 t 10 t ʰ 62 t 30 t ʰ 57 d 41 d ʰ 58 d 25 d ʰ 63 tʃ 30 tʃʰ 59 tʃ 56 tʃʰ 64 dʒ 59 dʒʰ 68 dʒ 28 dʒʰ 52 ɽ ----- ɽʰ ----- ɽ 4 ɽʰ 28 From the above data it can be observed that out of the 11 aspirated sounds, t ʰ in the initial position (62mm) shows the greatest difference in length with its unaspirated pair t in initial position (10mm). When we consider aspiration in the medial position, then difference of length between the unaspirated and aspirated sounds in each pair is comparatively lesser than the difference of length between unaspirated and aspirated sound in the initial position. Moreover, from the above list, we find out t ʰ, a voiceless, dental plosive, to be the lengthiest aspirated segment in initial position of a stressed syllable. Therefore, it is clear from this observation that voiceless plosives are the most aspirated in the initial position of a stressed syllable. Aspirations in voiceless plosives are stronger than aspiration in voiced plosives. Also, we find that the range of aspiration in voiced, palatal plosive is not much effective. The lengths of the aspirated segment dʒʰ in both initial (68mm) and medial positions (52mm) doesn t vary much in length when compared to the unaspirated segment dʒ in initial (59mm) or medial (28mm) positions. It can be observed from the above table that the unaspiration or aspiration in the initial position of a syllable supersede than that in the medial position. When we go by the measurements, the length of all the unaspirated & aspirated segments is comparatively more in the initial position than in medial position. When the position of the unaspiratedaspirated segment changes from initial to medial, the length of the unaspiratedaspirated segments gradually decreases. Table (2) uration of the unaspirated & aspirated segments in initial & medial positions UASPIRATE ASPIRATE UASPIRATE ASPIRATE IITIAL URATI O IITI AL ME IA L ME IAL p 45.683 pʰ 210.769 p 152.406 pʰ 268.748 t 13.375 tʰ 168.25 t 126.933 tʰ 148.609 k 29.125 kʰ 140.567 k 91.469 kʰ 195.577 b 163.5 bʰ 208.282 b 98.804 bʰ 178.666 d 158 dʰ 227.167 d ---- dʰ ---- g 134.681 gʰ 220 g 24.942 gʰ 115.857 t 25.649 t ʰ 302.903 t 106.648 t ʰ 226.380 d 172.803 d ʰ 265.738 d 93.35 d ʰ 233.181 tʃ 100.285 tʃʰ 225.061 tʃ 167.363 tʃʰ 282.256 dʒ 222.156 dʒʰ 290.214 dʒ 93.177 dʒʰ 194.360 ɽ ---- ɽʰ ---- ɽ 13.704 ɽʰ 16.134 From the above table, it can be observed that the duration of every aspirated segment is more than that of the unaspirated segment in each pair, irrespective of its position (initialmedial) in a word. In initial position, aspiration is preferable higher than that of medial position, as we can see the range of aspiration doubles in terms of duration when unaspiration-aspiration occurs in initial position than unaspiration-aspiration medially. Table(3)length of following vowels when unaspiratedaspirated segments occur in the initial position UASPIRATE FOLLOWI LETH ASPIRATE FOLLOWI LETH p a 81 pʰ a 58 t ɔ: 75 tʰ ɔ: 43 k a: 118 kʰ a: 72 b a: 73 bʰ a: 52 d a: 77 dʰ a: 58 g o 78 gʰ o 52 t a 59 t ʰ a 30 d a: 82 d ʰ a: 55 tʃ a 59 tʃʰ a 33 dʒ a: 70 dʒʰ a: 45 ɽ ---- ---- ɽʰ ---- ---- Aspiration for kʰ in the initial position shows the greatest affect in the following vowel length, when compared to unaspirated k in the initial position. From the above data, it can be observed that the length of the following vowel a: which occurs after unaspirated, voiceless, velar plosive k is 118mm, whereas when a: is followed by voiceless, aspirated, velar plosive kʰ, the length is reduced to 72 mm, i.e. about 46 mm difference. Therefore it can be said that aspiration does affect the following vowel length to a great extent, thereby reducing the vowel length. This is because, in aspirated segment, the original unaspirated segment is aspirated, thereby length of aspirated segment is lengthened, so the following vowel length has to be compensated by reducing its length. I will explain you from the above data: In the word ka:n, the length of unaspirated k in initial position is 10 mm, and that of a: is 118 mm. But when k is aspirated to
Journal of Multidisciplinary Scientific Research, 2015,4(2):04-08 7 kʰ in the initial position, then length of k is lengthened due to the effect of h which is the aspirated segment that leads of aspiration in k. The length becomes ( k + h ) = (10+35) mm = 45 mm, i.e. 35 mm more than unaspirated k. Therefore, the following vowel length is compensated by reducing it to 72 mm from 118 mm. This happens as the duration of the entire segment has to be maintained. For the word with unaspirated k in the initial position, the duration is 533 ms, and for word with aspirated kʰ in the initial position, the duration is 606 ms. The difference in duration is just about 100 ms. Therefore, we find that a difference of 100 ms is not much. When duration of the aspirated and unaspirated segments are taken into consideration, it is found that aspirated segments have more duration than the latter. Therefore more time is covered by the aspirated segment. Consequently, the length and therefore the duration of the following vowel naturally get reduced. Table(4)uration of following vowels when unaspiratedaspirated segments occur in the initial position. UASPIRATE FOLLO WI ASPIRATE FOLLOWI VOWE L p a 370.035 pʰ a 254.679 t ɔ: 250.796 tʰ ɔ: 195.533 k a: 343.683 kʰ a: 224.907 b a: 265.233 bʰ a: 208.282 d a: 282.930 dʰ a: 219.595 g o 197.094 gʰ o 184.516 t a 151.333 t ʰ a 146.566 d a: 345.607 d ʰ a: 251.993 tʃ a 197.228 tʃʰ a 125.882 dʒ a: 263.575 dʒʰ a: 192.053 ɽ ---- ---- ɽʰ ---- ---- Again, from the above data it can be observed that, like the length of the following vowel, even duration of the following vowel decreases in case of a vowel preceded by an aspirated segment (initial aspiration), when compared to the duration of a vowel preceded by an unaspirated segment (initial unaspiration). Table (5) length of preceding & following segments when unaspirated or aspirated segment occurs in medial position UASPIRA TE LETH ASPIRATE LETH PRECEI FOLLOWI PRECEI FOLLOWI p i - 50 u - 63 pʰ i - 34 u - 58 t i - 44 a - 60 tʰ i - 32 a - 32 k a - 50 a - 100 kʰ a - 33 a - 90 b ɒ - 51 o - 68 bʰ ɒ - 44 o - 60 d ---- ---- dʰ ---- ---- g ŋ - 40 o - 55 gʰ ŋ - 30 o - 52 t a - 43 a - 70 t ʰ a - 29 a - 65 d a - 53 a - 65 d ʰ a - 41 a - 50 tʃ u - 30 i - 57 tʃʰ u - 16 i - 51 dʒ o - 41 a - 78 dʒʰ o - 36 a - 72 ɽ a - 80 o - 65 ɽʰ a - 62 o - 62 From the above data, it can be observed that both preceding and following segments (vowels, nasals, etc) are reduced in their length even when aspirated segments occur in the medial position of the word. If we compare the lengths of the preceding and following segments of the aspirated versus unaspirated sounds, the latter (unaspirated) has more length and the former (aspirated sounds) has reduced lengths. When g h occurs in medial position, the preceding nasal length & the following vowel length doesn t vary much when compared to the preceding nasal length & following vowel length of g in medial position. The length of ŋ preceding g is 33mm & length of o following g is 55mm, whereas, the length of ŋ preceding g h is 30mm & length of o following g h is 52mm. As already mentioned, the length of the aspirated segment is more than unaspirated segment irrespective of its position (initial, medial) in the word. But the length of the preceding & following segments of the aspirated segment gradually decreases when compared to the length of the preceding & following segments of an unaspirated segment. Table (6) duration of the preceding & following segments of the unaspirated & aspirated segments in medial position UASP IRATE ASPIRA TE PRECEI FOLLOWI PRECEI FOLLOWI p i - 165.659 u - 208.730 pʰ i -136.379 u - 232.648 t i - 126.933 a - 201.939 tʰ i - 108.079 a - 219.536 k a - 127.040 a - 254.081 kʰ a - 105.804 a - 288.556 b ɒ - 167.967 o - 223.956 bʰ ɒ - o - 223.333 163.777 d ---- ---- dʰ ---- ---- g ŋ - 142.526 o - 195.973 gʰ ŋ - 133.681 o - 231.714 t a - 152.862 a - 248.846 t ʰ a - 115.176 a - 258.153 d a - 197.902 a - 242.71 d ʰ a - 151.753 a - 185.064 tʃ u - 89.659 i - 170.352 tʃʰ u - 70.564 i - 224.923 dʒ o - 136.438 a - 259.566 dʒʰ o - 134.557 a - 269.144 ɽ a - 274.098 o - 222.704 ɽʰ a - 250.089 o - 250.089 When unaspirated & aspirated segments are compared, we can find that during their occurrence in word medial positions effect the duration of their preceding & following segments. The duration of a preceding segment of an unaspirated segment is more than the duration of preceding segment of an aspirated segment. But, the duration of the following segment of an unaspirated segment is lesser than the duration of following segment of an aspirated segment in most cases. Even, when we carefully look into the formants of vowels in the spectrogram, formants of following vowels of an aspirated segment is less darker (2 formants, f1 & f2 visible) than the formants of following vowels of an unaspirated segment (3 formants, f1, f2, f3, visible, darker). COCLUSIO Therefore, the above experiment stands to prove the fact that aspiration is a distinctive feature in our target language Bangla (Standard Colloquial Bangla). In English aspiration also occur but it is not a distinctive feature. Aspiration in English is a phonetic characteristic, which is determined by sound context. There is no influence on meaning in English aspirates. The environment for English aspirates is mutually exclusive, i.e., where one sound occurs, the other sound never occurs in the similar position. Eg: [pin] and [p h in], where [p] and [p h ] are allophones of the same phoneme [p] and are in complementary distribution. But in Bangla, we find that the sounds are in parallel distribution and in the same environment the sound creates meaning difference. The length and duration of the aspiratedunaspirated segments from the above tables the length of the preceding & following segments of the aspirated segment gradually decreases when compared to the length of the preceding &
8 following segments of an unaspirated segment. It is a distinctive feature in Bangla and influences meaning in a definite way. It is also observed that the duration of every aspirated segment is more than that of the unaspirated segment in each pair, irrespective of its position (initialmedial) in a word. In initial position, aspiration is preferable higher than that of medial position, as we can see the range of aspiration doubles in terms of duration when unaspiration-aspiration occurs in initial position than unaspirationaspiration medially. In English, the pairs of aspirated and unaspirated consonants are same phonemes but in Bangla they are different phonemes. Aspiration is a great phonetic phenomenon for Bangla. Factually, it is one of the very few languages of the world to feature the unique property of aspiration and enriches the language inventory globally. Reference 1)Abercrombie, avid.(1967). Elements of eneral Phonetics. Edinburgh: Edinburgh University Press. 2)Chatterji, Sunitikumar. (1988). Bhasha-Prakash Bangala Vyakaran. Kolkata: Rupa and Company. 3)Crystal, avid. (2003). A ictionary of Linguistics and Phonetics. Massachusetts: Blackwell Publishing. 4)Islam, R. (2002). Bhasha Tatta. haka: Shikha Prakashani. 5)Biswas, Sailendra. (Compiled by). (1959). Samsad English- Bengali ictionary. Calcutta: Sahitya Samsad (5 th edition, 1980). 6)Halhed, athaniel B. (1969). A rammar of Bengal Language, 1778. Menston, Yorkshire: Scholar Press. 7)Fudge, E. (1969). Syllables. Journal of Linguistics 5, 253-286. 8)Bender, Ernest and T. Riccardi. (1978). An advanced course in Bengali. Philadelphia. South- Asia Regional Studies, University of Pennsylvania. 9)O Connor, J.. (1973). Phonetics. London: Penguin Books. 10)Shahidullah, Muhammad. (2000). Bangala Bayakaron. haka: Mowla Brothers.