Building a natural sounding Text-To-Speech system for the Nepali language: research and development challenges and solutions

Text-to-Speech (TTS) synthesis has come far from its primitive synthetic monotone voices to more natural and intelligible sounding voices. One of the direct applications of a natural sounding TTS systems is the screen reader applications for the visually impaired and the blind community. The Festival Speech Synthesis System uses a concatenative speech synthesis method together with the unit selection process to generate a natural sounding voice. This work primarily gives an account of the efforts put towards developing a Natural sounding TTS system for Nepali using the Festival system. We also shed light on the issues faced and the solutions derived which can be quite overlapping across other similar under-resourced languages in the region.

Dynamics of meaning-making in Nepali-English bilingual interactions

A growing proportion of Nepali speakers settled in urban areas who were educated through English-medium boarding schools present a tendency to use numerous English expressions along with Nepali in their daily interactions. This is also the case in some specific socio-professional environments such as TV and radio entertainment broadcasts made for the latest generations. The speakers make use of a variety of discursive resources for expressive purposes. The combinatory properties of such mixed language practices allow them to construct meaning in order to fit the requirements of particularly demanding social interactions, where most speakers exhibit the knowledge of a complex linguistic repertoire. The purpose of this article is to introduce the sociolinguistic context of contemporary urban Nepalese society, based on the most recent studies which have attempted to apprehend the role of English in Nepal today. A brief analysis of several examples of interactions extracted from a media corpus is also presented to make visible the mechanisms of meaning-making in a conversation where the two participants alternate Nepali and English.

Chamling basic vowels: acoustic analysis

This study analyzes Chamling vowels acoustically. Fundamental and formant frequency values of the vowels have been measured for both the male and female speakers and their individual values have been calculated for the comparison. Then, the average values have been computed for the specification of the vowels through the acoustic measurement. Acoustic properties have been observed in terms of gender, age, conditioning environment and intrinsic values. All this has clearly been presented in tables and figures. The study covers an overall acoustic description of the Chamling vowels on the basis of sound spectrogram using the up-to-date Pratt software. On the whole, it is concerned with phonetic study of the Chamling vowels in terms of acoustic properties and articulatory strategies.

Honorification in Nepali: a psycholinguistic perspective

This research article explores the psycholinguistic processes of honorification manipulation in Nepali language. Honorification is not static but dynamic process and is affected by many factors including psychological. Data is drawn from published narratives, mostly in the Nepali media, and analyzed from psycholinguistic perspective. The range of data include politics, royalese, language of marital discord and media. Significant influencers in choice of honorifics are fluctuation in state of mind and proximality. The canonical three-level honorific system is discussed as a reference point for psychologically motivated honorification fluctuation. Three types of honorific operators are identified and discussed: psychological, royalese and neutralization. The article attempts to bring forth the complex mechanisms that work on the honorification system in Nepali language.

Nouns and noun phrase structure in Nubri

This paper is a preliminary investigation of the nominal morphology and noun phrase structures of Nubri, a Tibetan variety spoken in the northern Gorkha. Nubri shares a number of inflectional and derivational features with Tibetan languages, such as Kyirong Tibetan. Like its close Tibetan varieties, a number of modifiers such as, genitive-marked nouns, demonstratives, relative clauses etc. precede the head nouns, whereas some other modifiers, such as article, emphatic marker, numerals etc. follow them in the noun phrase.

Revisiting the morphophonology of Thangmi: a Tibeto-Burman language of Nepal

This article revisits the morphophonology of Thangmi, a Tibeto-Burman language spoken in Nepal by a community group of the same name whose grammar and lexicon I was involved in documenting from 1996 onwards. The Thangmi (Nepali Thāmī) are an ethnic group who number around 30,000 and inhabit the central eastern hills of Nepal. The Thangmi are autochthonous to the upper reaches of Dolakhā district as well as to the eastern valleys of Sindhupālcok district, and their hitherto undocumented Tibeto-Burman language has two distinctly recognisable and mutually unintelligible dialects. Morphophonology (also known as morphophonemics) explores the interaction between morphology and phonology, and is predicated on a rigorous investigation of the phonological variations within morphemes that oftentimes mark different grammatical functions. While complex, Thangmi morphophonology lends itself to transparent interpretation, and this paper offers a modified analysis that builds on and develops from my earlier work (Turin 2012, 2005). Following a brief introduction to Thangmi segmental phonology, this article covers four aspects central to Thangmi morphophonology: the remnants of what may be a defunct liquid-nasal alternation, a brief overview of assimilation, a robust review of intervocalic approximants and finally a brief note on syncope.

Retroflexion in Nepali

Nepali, an Indo-Aryan language spoken in Nepal along with India and Bhutan, and some parts of Burma, possesses three coronal stops (2 plosives and 1 affricate). Retroflexion is traditionally considered as the distinctive feature between two different types of plosives. Though retroflexion in Nepali is considered- like in the case of other Indo-Aryan languages- a fundamental distinctive articulatory parameter (Bhat 1973, Ladefoged and Bhaskararao 1983, Ladefoged and Maddieson 1996), Pokharel (1989), however, claims that there is no retroflex category in Nepali, because the “so-called” (sic.) Nepali retroflex stops are not produced with the “tongue tip curling back” as it is described in the traditional grammar. In this work, I have tried to show that this claim is just one side of the story and that the “retroflex” as a phonetic and phonological category “does exist” in Nepali. Based on two different palatographic and linguographic studies (of 9 speakers – four females and five males- of Nepal) I have presented a different scenario than that of Pokharel, without completely denying his claim.

Lexical comparison in Hayu: a lexicostatistical analysis

This article presents the lexicostatistical analysis of Hayu language based on 210 wordlist. There appear different ranges of lexical and phonetic similarities across the five different survey points. Being based on the Mudhajor, the core area of Hayu, exhibits a significant degree of lexical similarity with other points, i.e. Aadmara, Kodre, Wadi and Balingkhola. Such similarity percentages clearly indicate that Hayu spoken in five different points are mutually intelligible to each other. The lexicostatistical data, therefore, show that there is not much lexical variations across the villages where Hayu is spoken.

Cognitive archaeology: in search of the earliest syntactic language-users in Himalaya

Recent scientific studies unfold that neural structures bearing on intonation of speech have a deep evolutionary history traced to mammal-like reptiles called therapsids found in the Triassic period (∼252.17 mya, million years ago). Therefore, these structures were already present in the primates. It goes to the credit of Homo sapiens who developed it to the extent that humans are defined as symbolling animals, for language is the most articulated symbolism. Cognitive archaeology makes it clear that it took hominins millions of years to develop a syntactic language. Stratigraphically controlled and securely established artefact-bearing sites of the Middle Palaeolithic Arjun complex in the Deokhuri Valley, West Nepal, provide firm dates for the presence of the earliest syntactic language speakers in Himalaya from 100 ka to 70 ka (thousand years ago).

Puma verbs and the Proto-Kiranti verbal agreement system

The purpose of this paper is to overview the historical status of conjugations observed in the Puma language based on a comparison of Proto-Kiranti verbal agreement system. The provenance of Proto-Kiranti affixal agreement paradigms is presented by investigating the conjugations of Puma. While some Puma suffixes are identical with the Proto-Kiranti morphemes, a few are cognate with the Proto-Kiranti reflexes.