1 The curious case of Kannada ‘maaDu’ Srivatsa B R Indian Heritage and Language Computing Group Centre for Development of Advanced Computing Bengaluru, Karnataka, India srivatsa@cdac.in Annarao Kulkarni Indian Heritage and Language Computing Group Centre for Development of Advanced Computing Bengaluru, Karnataka, India kulkarni@cdac.in AbstractKannada is one of the major languages of India. It is one of the four major languages of the Dravidian family, the other three being Tamil, Telugu and Malayalam. Kannada belongs to the Southern branch of the Dravidian family (Burrow and Emeneau, 1984) and is spoken primarily in the state of Karnataka (formerly Mysore), South India. It is the official language of the state and one of the 22 languages listed in the Eighth Schedule of the Indian Constitution. It is one of the six classical languages of India. Kannada has a rich, continuous literary tradition and a fine grammatical tradition. In this paper, we discuss the curious and odd case of the Kannada verb ‘maaDu’, address the problems it poses to Natural Language Processing (NLP) and Machine Translation (MT) systems and offer possible solutions. KeywordsNatural Language Processing, Machine Translation, Kannada verbs I. INTRODUCTION A Verb is one of the most important lexical items (word classes or lexical classes/categories). It is a Part of Speech (PoS), which, in Syntax, conveys an action (tinnu, kuDi, oodu) or occurrence (jarugu, aagu, saMbhavisu) or a state of being (iru, nillu, kuLituko). A verb has tense, aspect and modality/mood (together known as TAM features). A verb may also agree with the person, number and gender (together known as PNG features) of some of its arguments, such as its subject or object. Further, the verb is an obligatory constituent of a sentence, except in copular and dative subject constructions, where it is optional. Kannada is rich in inflectional and derivative morphology and the same applies to the verbs in the language as well. Kannada is a di-glossic language and is agglutinative in nature. The language has different dialects, spoken in different regions. In this paper, we illustrate with examples, using the written form, which is widely accepted in the language as it affects Natural Language Processing and Machine Translation, if it is written in a different format. Because of the agglutinative nature of Kannada language, many of the grammatical features of a lexical item in the language become a part and parcel (sometimes, inseparable) of that lexical item. Verb is no exception to this phenomenon. It is a well-known and widely acknowledged fact that PNG and TAM features too become a part of the verb in Kannada. Like many other Indian Languages, Kannada is an SOV language, in general and the verb occurs in the sentence-final position, under normal circumstances, though it is possible that the Scrambling and Topicalization effects may affect the word order. II. THE CURIOSITY AND PECULIARITY SURROUNDING MAADU ‘maaDu’ is the simplest and the base form of a verb in Kannada which translates to ‘do’ or ‘make’ in English. It is an independent verb by itself and behaves like other verbs in the language. But, in addition to this, this particular verb is distinctively, exclusively and widely used for another purpose in the language. It is this purpose that generates curiosity and the paper mainly focuses on this aspect of the verb. Let us look at a few examples from the language to understand this phenomenon better. We shall illustrate with examples, the different sentential constructions with the verb ‘maaDu’. (i) kuMbaara maDike maaDuttaane potter pot do/make-Simple Present The potter makes a pot. (i) is a simple SOV construction in the language, with ‘kuMbaara’ as the Subject, ‘maDike’ as the object and ‘maaDu’ as the verb. The object of the sentence ‘maDike’ does not have an explicit case marker as the accusative case marker is optional in Kannada language and the sentence is grammatical and meaningful. Such sentences, with ‘maaDu’ as a verb meaning do/make, are possible in the language. (i) a. kuMbaara maDikeyannu maaDuttaane potter pot-Acc do/make-Simple Present The potter makes a pot. (i) can also be written as (i) a. with an explicit Accusative case marker for the object ‘maDike’ in the sentence. The sentence remains grammatical, meaningful and acceptable. We shall look at a few more examples and see what is happening in the language. (ii) ? naanu uuTa maaDutteene I meal do/make-Simple Present I do/make meal.