Shakuntala Mahanta | Indian Institute of Technology Guwahati (original) (raw)

Papers by Shakuntala Mahanta

Speech Prosody 2022, May 23, 2022

Glossa, 2019

The pitch contour of an utterance in a tone language can surface with both tonal and intonational... more The pitch contour of an utterance in a tone language can surface with both tonal and intonational f 0 features. In this paper we set out to analyze the intonational phonology of Boro, a tone language, and establish that there are three levels of prosodic constituents in Boro: Prosodic Word, ip and IP. Prosodic Word is the domain for distribution of lexical tones. Phonological processes show that the next higher level of prosodic structure is that of the intermediate phrase. Downstepping is within intermediate phrases (ip) and does not cross ips. The highest level of prosodic constituency is the IP which is marked by both initial and final boundary tones. This study shows that in Boro intonational phonology, boundary tones and their scaling and alignment in the context of their lexical tones is more important than assigning pitch accents.

Himalayan Linguistics

This paper discusses the morphological and prosodic properties of Boro tones. Tonal alignment in ... more This paper discusses the morphological and prosodic properties of Boro tones. Tonal alignment in Boro vis-à-vis word formation processes throws light on some distributional properties of tone in the language. Boro has some affixes which have their own tonal specifications. Prefixes determine the tone of the stem and lead to changes in the tonal specification of the stem. The addition of suffixes do not alter the tonal nature of the stems. Suffixes in Boro, irrespective of their lexical tonal status, belong to the recessive category whereas prefixes belong to the dominant group. In terms of prosodic properties, the smallest domain for Boro tonal assignment is a minimal word and maximally it is the prosodic word.

Lecture Notes in Computer Science, 2022

Linguistics Vanguard

Our study investigates the nature of tonogenesis in Sylheti, which was triggered by the merger of... more Our study investigates the nature of tonogenesis in Sylheti, which was triggered by the merger of aspirated and unaspirated consonants in the language. We propose that the tonogenetic factors contributed by the four-way laryngeal contrast condition a three-way tonal contrast in Sylheti depending on the voicing and syllabic position of the sound. We take into consideration both monosyllabic and disyllabic sets of words and claim that Sylheti has three lexical level tones: high, mid, and low. We have built linear mixed-effect models of f0 and duration to examine the acoustic correlates of tone contrast in the language. We conclude that Sylheti has a three-way tonal contrast distinctively affecting pitch.

Deori, a Tibeto-Burman language (henceforth TB) belonging to the Bodo- Koch language group [10] c... more Deori, a Tibeto-Burman language (henceforth TB) belonging to the Bodo- Koch language group [10] can be unambiguously called an endangered language [2]. This paper analyzes the production and perception of lexical tones in Deori. Most recently, some remnant lexical tonal contrast has been shown in Deori with acoustic evidence [11]. Following the minimal tonal contrast prevalent among the older generation [11] a production experiment was conducted to determine the tonal distinction in the speech of the younger generation. The production experiment was followed by a perception test to investigate the impact of language experience on the perception of lexical tones in Deori. The result of production test shows a trend of underlying tone reversal H>L; L>H in monosyllables and a considerable F0 variation in the disyllabic stems. The tone reversal can be described as tone swap where a high tone has a low fundamental frequency and vice-versa. The perception test result too shows a sim...

Language Documentation & Conservation, 2019

Deori, a Tibeto-Burman language, is an “endangered” language and is described as a language on th... more Deori, a Tibeto-Burman language, is an “endangered” language and is described as a language on the verge of its extinction. Recent research on Deori phonetics and phonology has shown loss of distinct pitch realization and identification in the speech of older as well as younger generation speakers. The difference in production and perception of tonal categories among the speakers of the younger age group led to an examination of language vitality of Deori. To substantiate the analyses of inter-generational language change, this study takes into account intergenerational perceptions on language use and its robustness. The findings of this study show that the language status of Deori is not completely bleak, and there is a sense of optimism for the future of the language among speakers irrespective of age. The findings also show that the language suffers from lack of support in the public domain, lack of teachers to teach Deori as a subject in schools, and absence of exposure in new m...

In this paper I have argued that the Contrast Preservation approach with Neighbours can account f... more In this paper I have argued that the Contrast Preservation approach with Neighbours can account for chain shifts with semi-iterative raising only if the attribute of ‘feature distance’ is built into the proposal. This attribute can factor out irrelevant Neighbours in chain shifts, and at the same time it can also predict the right Neighbours. The tool of feature distance evaluates distances in such a way that the minimum difference must be at least the number of unshared features. The data that are currently investigated are from Standard Colloquial Bengali and Cachar Bengali. Extant proposals within the Preserve Contrast approach fail to account for chain shifts adequately and adoption of feature distance leads to a more comprehensive analysis.

The variety described here is representative of colloquial Assamese spoken in the eastern distric... more The variety described here is representative of colloquial Assamese spoken in the eastern districts of Assam.1 Assam is a North-Eastern state of India, therefore Assamese and creoles of Assamese like Nagamese are spoken in the different North-Eastern states of Nagaland, Arunachal Pradesh, Meghalaya, and also the neighbouring country of Bhutan. Approximately 15 million people speak Assamese in India (see Ethnologue, Gordon 2005, which lists 15,374,000 speakers including those in Bhutan and Bangladesh). In the pre-British era (until 1826), the kingdom of Assam was ruled by Ahom kings and the then capital was based in the Eastern district of Sibsagar and later in Jorhat. American missionaries established the first printing press in Sibsagar and in the year 1846 published a monthly periodical Arunodoi using the variety spoken in and around Sibsagar as the point of departure. This is the immediate reason which led to the acceptance of the formal variety spoken in eastern Assam (which rou...

This dissertation establishes the pattern of prominence in Assamese and Assamese English. It show... more This dissertation establishes the pattern of prominence in Assamese and Assamese English. It shows for the first time that Assamese exemplifies a left – to – right trochaic system with an iterative binary rhythm. However, this rhythmic profile is disturbed by the occurrence of closed syllables deemed to be heavy. To support our intuitive judgements of prominence in Assamese with acoustic evidence, we conducted some experiments on PRAAT (a computer software for speech analysis). In chapter 3, we report the preliminary experiments which reveal that the acoustic correlate of primary prominence could very well be syllable duration and a low tone. Moreover, our intuitive judgements of prominence were vindicated by the surface phonetic realization of the F 0 contours in our acoustic experiments. In a sequence of two light syllables, a distinct low tone on the first light syllable indicated primary prominence. In a light and heavy sequence, the F 0 contour manifested as a plateau, instead ...

This paper examines the phonetic and phonological properties of tone and explores the way voicele... more This paper examines the phonetic and phonological properties of tone and explores the way voiceless sonorants interact with tone in a previously undocumented and lesser-known language, namely Mog (an Arakan tribe settled in Tripura in India). In this study, a total of 62 words (62 words *4 repetitions *6 subjects = 1488 tokens) with two-way and three-way meaning contrasts were examined. A repeated measures ANOVA with a GreenhouseGeisser correction confirm a significant effect of mean f0 on tone types (F (1.87, 19.9) = .17, p < 0.05) and confirms the presence of three contrastive tones in Mog, viz., high-falling, mid-rising, and lowrising. We would further argue that the voiceless sonorants significantly raise the f0 of the following vowel (18 Hz on average till the 60% of the total rhyme, when compared to the voiced counterparts, (F [1.11, 17.) = .15, p < 0.00).

Languages, 2021

This paper presents an analysis of Dimasa focus intonation. The acoustic analysis shows that narr... more This paper presents an analysis of Dimasa focus intonation. The acoustic analysis shows that narrow focus sentences undergo a jump in the pitch range irrespective of the underlying tonal value of the morpheme it attaches to. In addition to f0 expansion, the prosodic property of focus in Dimasa was found to have different (tense) phonation in morphologically marked narrow focus sentences when compared to the broad focus context. Thus, the tense phonation property of sentences bearing morphological focus is not only an acoustic property of a higher pitch range but may also be an acoustic cue of discourse-level intonation.

Interspeech 2013, 2013

Intonation and prosodic structures, apart from other functions, play a significant role in convey... more Intonation and prosodic structures, apart from other functions, play a significant role in conveying the focus in an utterance. The term 'focus' is applied to a constituent which is informationally more important or salient than other backgrounded parts of the same sentence [1] [2], and in a sentence the part that receives presentational focus which answers a wh-question. In this paper, we will try to establish that Narbaria Variety of Assamese (NVA), which is a variety of Standard Colloquial Assamese (SCA), an Indo-Aryan language [3], employs phrasing as a marker of presentational focus. In marking the focus of the focused constituent, duration plays a significant role; it is not the durational increase of the focused P-phrase only which is significant but the decrease of duration in the given constituent is equally decisive. The present work considers sentences with presentational focus on Subject and on Object in two different word orders i.e. SOV (Subject-Object-Verb) and OSV (Object-Subject-Verb). Different phonetic cues of focus or prominence like F 0-max, F 0-min, F 0-range and duration values are measured and compared against respective wide-focus baseline.

Glossa: a journal of general linguistics, 2019

The pitch contour of an utterance in a tone language can surface with both tonal and intonational... more The pitch contour of an utterance in a tone language can surface with both tonal and intonational f0 features. In this paper we set out to analyze the intonational phonology of Boro, a tone language, and establish that there are three levels of prosodic constituents in Boro: Prosodic Word, ip and IP. Prosodic Word is the domain for distribution of lexical tones. Phonological processes show that the next higher level of prosodic structure is that of the intermediate phrase. Downstepping is within intermediate phrases (ip) and does not cross ips. The highest level of prosodic constituency is the IP which is marked by both initial and final boundary tones. This study shows that in Boro intonational phonology, boundary tones and their scaling and alignment in the context of their lexical tones is more important than assigning pitch accents.

Interspeech 2018, 2018

Assamese is one of the low resource Indian languages. This paper implements both rule-based and d... more Assamese is one of the low resource Indian languages. This paper implements both rule-based and data-driven grapheme to phoneme (G2P) conversion systems for Assamese. The rule-based system is used as the baseline which yields a word error rate of 35.3%. The data-driven systems are implemented using state-of-the-art sequence learning techniques such as-i) Joint-Sequence Model (JSM), ii) Recurrent Neural Networks with LTSM cell (LSTM-RNN) and iii) bidirectional LSTM (BiLSTM). The BiLSTM yields the lowest WER i.e., 18.7%, which is an absolute 16.6% improvement on the baseline system. We additionally implement the rules of syllabification for Assamese. The surface output is generated in two forms namely i) phonemic sequence with syllable boundaries, and ii) only phonemic sequence. The output of BiLSTM is fed as an input to Hybrid system. The Hybrid system syllabifies the input phonemic sequences to apply the vowel harmony rules. It also applies the rules of schwa-deletion as well as some rules in which the consonants change their form in clusters. The accuracy of the Hybrid system is 17.3% which is an absolute 1.4% improvement over the BiLSTM based G2P.

10th International Conference on Speech Prosody 2020, 2020

This study reports data collected and analyzed from 7 native speakers from a corpus of 70 Sylheti... more This study reports data collected and analyzed from 7 native speakers from a corpus of 70 Sylheti noun words. Our work shows that the loss of [+spread glottis] feature from [-voice] and [+voice] onsets have resulted in independent tone association patterns. The [-voice, +spread glottis] onset associated to a Low tone, as opposed to the tone association pattern of the voiced onsets [1]. Our study looks into the tonal pattern of Sylheti which arose from the merger of [+spread glottis] and [-spread glottis] contrast in voiceless obstruents. We built linear mixed effect models of f0 and duration to examine the acoustic factors affecting tone. We found that disyllables trigger a three way tonal contrast depending on the historical features and positions of voiceless onsets in Sylheti. The tone of the syllable spreads throughout the word rather than the syllable of origin. Word is thus reclaimed to be the TBU in Sylheti. The Mean f0 for the intercept was about 274.345 Hz which represented the High tone. It differed from the Low tone by about 75 Hz, and the Mid tone by about 25 Hz. Tone affected pitch by (χ2 (1) = 927.07, p < 0.0001).

Language Sciences, 2018

The extant literature (Smith, 2000a,b, 2001, 2002) on differences between nouns versus verbs in p... more The extant literature (Smith, 2000a,b, 2001, 2002) on differences between nouns versus verbs in phonology has shown that nouns enjoy a privileged status in exhibiting phonological contrasts and processes at the expense of verbal domains. In this paper, we show from original work on Sylheti tones that verbs exhibit exceptionally marked tonal polarity and dominant suffixes which are not seen in nouns. This does not lead to more contrasting patterns in nouns but nouns are faithful. Noun faithfulness can be taken care of by a general faithfulness constraints. Verbs however, need a conjoined markedness and faithfulness constriant ranked higher than the general Faithfulness and Markedness constraints, showing the presence of a putative marked structure in verbs than in nouns.

Journal of South Asian Languages and Linguistics, 2020

This chapter is a short thematic introduction to the papers that appear in this special volume on... more This chapter is a short thematic introduction to the papers that appear in this special volume on the phonetics and phonology of a selective but representative set of languages from the South Asian sprachbund. The volume consists of five papers that engage with a broad set of topics, namely, acoustics of Kalasha affricates, perception of breathiness in Gujarati, syllable structure of Kuki-Chin, syllable structure and incipient vowels in Lamkang, and vowel harmony in Assamese. These papers, we hope, are representative in terms of the choice of theoretical frameworks and methodologies employed in addressing the various linguistic phenomena.

TAL2018, Sixth International Symposium on Tonal Aspects of Languages, 2018

Papers in Historical Phonology, 2017

In this paper we investigate the existence of tones in Deori, a language which is historically kn... more In this paper we investigate the existence of tones in Deori, a language which is historically known to have had tonal distinctions. In a study, 5 speakers were recorded for their production of potential lexical tones in a list of 34 words, and a list of 54 monosyllabic roots were recorded for a vowel experiment. We conducted an f0 analysis in order to examine the extent of ‘tonoexodus’ and loss of tonal properties in Deori. To the extent that experimental methods can be used to determine lexical tone, phonetic measurements of f0 and further statistical analysis reliably indicate the distribution of lexically distinct tones. The results show clear presence of tonally distinctive words but without any definitive tonal alignment. We consider the diachrony and synchronic analysis of this and conclude that the syllable is not the Tone Bearing Unit in the conventional sense in current Deori, and that there is no clinching evidence to suggest alignment and spreading of the lexically disti...

Speech Prosody 2022, May 23, 2022

Glossa, 2019

The pitch contour of an utterance in a tone language can surface with both tonal and intonational... more The pitch contour of an utterance in a tone language can surface with both tonal and intonational f 0 features. In this paper we set out to analyze the intonational phonology of Boro, a tone language, and establish that there are three levels of prosodic constituents in Boro: Prosodic Word, ip and IP. Prosodic Word is the domain for distribution of lexical tones. Phonological processes show that the next higher level of prosodic structure is that of the intermediate phrase. Downstepping is within intermediate phrases (ip) and does not cross ips. The highest level of prosodic constituency is the IP which is marked by both initial and final boundary tones. This study shows that in Boro intonational phonology, boundary tones and their scaling and alignment in the context of their lexical tones is more important than assigning pitch accents.

Himalayan Linguistics

This paper discusses the morphological and prosodic properties of Boro tones. Tonal alignment in ... more This paper discusses the morphological and prosodic properties of Boro tones. Tonal alignment in Boro vis-à-vis word formation processes throws light on some distributional properties of tone in the language. Boro has some affixes which have their own tonal specifications. Prefixes determine the tone of the stem and lead to changes in the tonal specification of the stem. The addition of suffixes do not alter the tonal nature of the stems. Suffixes in Boro, irrespective of their lexical tonal status, belong to the recessive category whereas prefixes belong to the dominant group. In terms of prosodic properties, the smallest domain for Boro tonal assignment is a minimal word and maximally it is the prosodic word.

Lecture Notes in Computer Science, 2022

Linguistics Vanguard

Our study investigates the nature of tonogenesis in Sylheti, which was triggered by the merger of... more Our study investigates the nature of tonogenesis in Sylheti, which was triggered by the merger of aspirated and unaspirated consonants in the language. We propose that the tonogenetic factors contributed by the four-way laryngeal contrast condition a three-way tonal contrast in Sylheti depending on the voicing and syllabic position of the sound. We take into consideration both monosyllabic and disyllabic sets of words and claim that Sylheti has three lexical level tones: high, mid, and low. We have built linear mixed-effect models of f0 and duration to examine the acoustic correlates of tone contrast in the language. We conclude that Sylheti has a three-way tonal contrast distinctively affecting pitch.

Deori, a Tibeto-Burman language (henceforth TB) belonging to the Bodo- Koch language group [10] c... more Deori, a Tibeto-Burman language (henceforth TB) belonging to the Bodo- Koch language group [10] can be unambiguously called an endangered language [2]. This paper analyzes the production and perception of lexical tones in Deori. Most recently, some remnant lexical tonal contrast has been shown in Deori with acoustic evidence [11]. Following the minimal tonal contrast prevalent among the older generation [11] a production experiment was conducted to determine the tonal distinction in the speech of the younger generation. The production experiment was followed by a perception test to investigate the impact of language experience on the perception of lexical tones in Deori. The result of production test shows a trend of underlying tone reversal H>L; L>H in monosyllables and a considerable F0 variation in the disyllabic stems. The tone reversal can be described as tone swap where a high tone has a low fundamental frequency and vice-versa. The perception test result too shows a sim...

Language Documentation & Conservation, 2019

Deori, a Tibeto-Burman language, is an “endangered” language and is described as a language on th... more Deori, a Tibeto-Burman language, is an “endangered” language and is described as a language on the verge of its extinction. Recent research on Deori phonetics and phonology has shown loss of distinct pitch realization and identification in the speech of older as well as younger generation speakers. The difference in production and perception of tonal categories among the speakers of the younger age group led to an examination of language vitality of Deori. To substantiate the analyses of inter-generational language change, this study takes into account intergenerational perceptions on language use and its robustness. The findings of this study show that the language status of Deori is not completely bleak, and there is a sense of optimism for the future of the language among speakers irrespective of age. The findings also show that the language suffers from lack of support in the public domain, lack of teachers to teach Deori as a subject in schools, and absence of exposure in new m...

In this paper I have argued that the Contrast Preservation approach with Neighbours can account f... more In this paper I have argued that the Contrast Preservation approach with Neighbours can account for chain shifts with semi-iterative raising only if the attribute of ‘feature distance’ is built into the proposal. This attribute can factor out irrelevant Neighbours in chain shifts, and at the same time it can also predict the right Neighbours. The tool of feature distance evaluates distances in such a way that the minimum difference must be at least the number of unshared features. The data that are currently investigated are from Standard Colloquial Bengali and Cachar Bengali. Extant proposals within the Preserve Contrast approach fail to account for chain shifts adequately and adoption of feature distance leads to a more comprehensive analysis.

The variety described here is representative of colloquial Assamese spoken in the eastern distric... more The variety described here is representative of colloquial Assamese spoken in the eastern districts of Assam.1 Assam is a North-Eastern state of India, therefore Assamese and creoles of Assamese like Nagamese are spoken in the different North-Eastern states of Nagaland, Arunachal Pradesh, Meghalaya, and also the neighbouring country of Bhutan. Approximately 15 million people speak Assamese in India (see Ethnologue, Gordon 2005, which lists 15,374,000 speakers including those in Bhutan and Bangladesh). In the pre-British era (until 1826), the kingdom of Assam was ruled by Ahom kings and the then capital was based in the Eastern district of Sibsagar and later in Jorhat. American missionaries established the first printing press in Sibsagar and in the year 1846 published a monthly periodical Arunodoi using the variety spoken in and around Sibsagar as the point of departure. This is the immediate reason which led to the acceptance of the formal variety spoken in eastern Assam (which rou...

This dissertation establishes the pattern of prominence in Assamese and Assamese English. It show... more This dissertation establishes the pattern of prominence in Assamese and Assamese English. It shows for the first time that Assamese exemplifies a left – to – right trochaic system with an iterative binary rhythm. However, this rhythmic profile is disturbed by the occurrence of closed syllables deemed to be heavy. To support our intuitive judgements of prominence in Assamese with acoustic evidence, we conducted some experiments on PRAAT (a computer software for speech analysis). In chapter 3, we report the preliminary experiments which reveal that the acoustic correlate of primary prominence could very well be syllable duration and a low tone. Moreover, our intuitive judgements of prominence were vindicated by the surface phonetic realization of the F 0 contours in our acoustic experiments. In a sequence of two light syllables, a distinct low tone on the first light syllable indicated primary prominence. In a light and heavy sequence, the F 0 contour manifested as a plateau, instead ...

This paper examines the phonetic and phonological properties of tone and explores the way voicele... more This paper examines the phonetic and phonological properties of tone and explores the way voiceless sonorants interact with tone in a previously undocumented and lesser-known language, namely Mog (an Arakan tribe settled in Tripura in India). In this study, a total of 62 words (62 words *4 repetitions *6 subjects = 1488 tokens) with two-way and three-way meaning contrasts were examined. A repeated measures ANOVA with a GreenhouseGeisser correction confirm a significant effect of mean f0 on tone types (F (1.87, 19.9) = .17, p < 0.05) and confirms the presence of three contrastive tones in Mog, viz., high-falling, mid-rising, and lowrising. We would further argue that the voiceless sonorants significantly raise the f0 of the following vowel (18 Hz on average till the 60% of the total rhyme, when compared to the voiced counterparts, (F [1.11, 17.) = .15, p < 0.00).

Languages, 2021

This paper presents an analysis of Dimasa focus intonation. The acoustic analysis shows that narr... more This paper presents an analysis of Dimasa focus intonation. The acoustic analysis shows that narrow focus sentences undergo a jump in the pitch range irrespective of the underlying tonal value of the morpheme it attaches to. In addition to f0 expansion, the prosodic property of focus in Dimasa was found to have different (tense) phonation in morphologically marked narrow focus sentences when compared to the broad focus context. Thus, the tense phonation property of sentences bearing morphological focus is not only an acoustic property of a higher pitch range but may also be an acoustic cue of discourse-level intonation.

Interspeech 2013, 2013

Intonation and prosodic structures, apart from other functions, play a significant role in convey... more Intonation and prosodic structures, apart from other functions, play a significant role in conveying the focus in an utterance. The term 'focus' is applied to a constituent which is informationally more important or salient than other backgrounded parts of the same sentence [1] [2], and in a sentence the part that receives presentational focus which answers a wh-question. In this paper, we will try to establish that Narbaria Variety of Assamese (NVA), which is a variety of Standard Colloquial Assamese (SCA), an Indo-Aryan language [3], employs phrasing as a marker of presentational focus. In marking the focus of the focused constituent, duration plays a significant role; it is not the durational increase of the focused P-phrase only which is significant but the decrease of duration in the given constituent is equally decisive. The present work considers sentences with presentational focus on Subject and on Object in two different word orders i.e. SOV (Subject-Object-Verb) and OSV (Object-Subject-Verb). Different phonetic cues of focus or prominence like F 0-max, F 0-min, F 0-range and duration values are measured and compared against respective wide-focus baseline.

Glossa: a journal of general linguistics, 2019

The pitch contour of an utterance in a tone language can surface with both tonal and intonational... more The pitch contour of an utterance in a tone language can surface with both tonal and intonational f0 features. In this paper we set out to analyze the intonational phonology of Boro, a tone language, and establish that there are three levels of prosodic constituents in Boro: Prosodic Word, ip and IP. Prosodic Word is the domain for distribution of lexical tones. Phonological processes show that the next higher level of prosodic structure is that of the intermediate phrase. Downstepping is within intermediate phrases (ip) and does not cross ips. The highest level of prosodic constituency is the IP which is marked by both initial and final boundary tones. This study shows that in Boro intonational phonology, boundary tones and their scaling and alignment in the context of their lexical tones is more important than assigning pitch accents.

Interspeech 2018, 2018

Assamese is one of the low resource Indian languages. This paper implements both rule-based and d... more Assamese is one of the low resource Indian languages. This paper implements both rule-based and data-driven grapheme to phoneme (G2P) conversion systems for Assamese. The rule-based system is used as the baseline which yields a word error rate of 35.3%. The data-driven systems are implemented using state-of-the-art sequence learning techniques such as-i) Joint-Sequence Model (JSM), ii) Recurrent Neural Networks with LTSM cell (LSTM-RNN) and iii) bidirectional LSTM (BiLSTM). The BiLSTM yields the lowest WER i.e., 18.7%, which is an absolute 16.6% improvement on the baseline system. We additionally implement the rules of syllabification for Assamese. The surface output is generated in two forms namely i) phonemic sequence with syllable boundaries, and ii) only phonemic sequence. The output of BiLSTM is fed as an input to Hybrid system. The Hybrid system syllabifies the input phonemic sequences to apply the vowel harmony rules. It also applies the rules of schwa-deletion as well as some rules in which the consonants change their form in clusters. The accuracy of the Hybrid system is 17.3% which is an absolute 1.4% improvement over the BiLSTM based G2P.

10th International Conference on Speech Prosody 2020, 2020

This study reports data collected and analyzed from 7 native speakers from a corpus of 70 Sylheti... more This study reports data collected and analyzed from 7 native speakers from a corpus of 70 Sylheti noun words. Our work shows that the loss of [+spread glottis] feature from [-voice] and [+voice] onsets have resulted in independent tone association patterns. The [-voice, +spread glottis] onset associated to a Low tone, as opposed to the tone association pattern of the voiced onsets [1]. Our study looks into the tonal pattern of Sylheti which arose from the merger of [+spread glottis] and [-spread glottis] contrast in voiceless obstruents. We built linear mixed effect models of f0 and duration to examine the acoustic factors affecting tone. We found that disyllables trigger a three way tonal contrast depending on the historical features and positions of voiceless onsets in Sylheti. The tone of the syllable spreads throughout the word rather than the syllable of origin. Word is thus reclaimed to be the TBU in Sylheti. The Mean f0 for the intercept was about 274.345 Hz which represented the High tone. It differed from the Low tone by about 75 Hz, and the Mid tone by about 25 Hz. Tone affected pitch by (χ2 (1) = 927.07, p < 0.0001).

Language Sciences, 2018

The extant literature (Smith, 2000a,b, 2001, 2002) on differences between nouns versus verbs in p... more The extant literature (Smith, 2000a,b, 2001, 2002) on differences between nouns versus verbs in phonology has shown that nouns enjoy a privileged status in exhibiting phonological contrasts and processes at the expense of verbal domains. In this paper, we show from original work on Sylheti tones that verbs exhibit exceptionally marked tonal polarity and dominant suffixes which are not seen in nouns. This does not lead to more contrasting patterns in nouns but nouns are faithful. Noun faithfulness can be taken care of by a general faithfulness constraints. Verbs however, need a conjoined markedness and faithfulness constriant ranked higher than the general Faithfulness and Markedness constraints, showing the presence of a putative marked structure in verbs than in nouns.

Journal of South Asian Languages and Linguistics, 2020

This chapter is a short thematic introduction to the papers that appear in this special volume on... more This chapter is a short thematic introduction to the papers that appear in this special volume on the phonetics and phonology of a selective but representative set of languages from the South Asian sprachbund. The volume consists of five papers that engage with a broad set of topics, namely, acoustics of Kalasha affricates, perception of breathiness in Gujarati, syllable structure of Kuki-Chin, syllable structure and incipient vowels in Lamkang, and vowel harmony in Assamese. These papers, we hope, are representative in terms of the choice of theoretical frameworks and methodologies employed in addressing the various linguistic phenomena.

TAL2018, Sixth International Symposium on Tonal Aspects of Languages, 2018

Papers in Historical Phonology, 2017

In this paper we investigate the existence of tones in Deori, a language which is historically kn... more In this paper we investigate the existence of tones in Deori, a language which is historically known to have had tonal distinctions. In a study, 5 speakers were recorded for their production of potential lexical tones in a list of 34 words, and a list of 54 monosyllabic roots were recorded for a vowel experiment. We conducted an f0 analysis in order to examine the extent of ‘tonoexodus’ and loss of tonal properties in Deori. To the extent that experimental methods can be used to determine lexical tone, phonetic measurements of f0 and further statistical analysis reliably indicate the distribution of lexically distinct tones. The results show clear presence of tonally distinctive words but without any definitive tonal alignment. We consider the diachrony and synchronic analysis of this and conclude that the syllable is not the Tone Bearing Unit in the conventional sense in current Deori, and that there is no clinching evidence to suggest alignment and spreading of the lexically disti...