Tamil At first glance, Tamil, a Dravidian language spoken by more than seventy million people in India, Sri Lanka, Malaysia, and Singapore, appears to be largely devoid of loanwords from Arabic, especially if compared to other South Asian languages. Even with reference to Muslim religious practice, Tamil words often replace Arabic ones, such as Tamil toḻukai instead of Arabic ṣalāt. Nevertheless, Arabic words are used by Tamil-speaking Muslims in both everyday conversation and Islamic literary texts dating back to the 16th century, and even among non-Muslims, Arabic terms can be found in administrative and nautical vocabulary.

Tamil came into contact with Arabic through two different routes roughly corresponding to the routes by which Islam was transmitted to South India, although the spread of Arabic vocabulary at times occurred independently of processes of Islamization. One route involved the Arab and Persian merchants who frequented the port cities of Southeast India and Ceylon since at least the beginning of the 2nd millennium C.E., thereby giving rise to many Muslim communities along the coasts of these regions. Arabic was transmitted to Tamil-speaking areas through a second route, the Persianate culture of states in central and northern India, which repeatedly conquered the region from the 14th century onward (cf. generally Fanselow 1989; Nainar 1942).

Little research has been undertaken on the topic. Most contributions are descriptive, with little phonetical, morphological, or semantic analysis (Nainar 1941; Mukamatu 1996:125–154; Uwise 1976:355–405, 1983; Vaidyanathan 1958; cf. also Bausani 1971). The focus of most studies is on Arabic loanwords in Islamic Tamil literature, with examples from non-Islamic literature and contemporary spoken Tamil mentioned only occasionally. Studies of contemporary spoken Muslim dialects are confined to a few, generally not easily accessible, dissertations (cf. nos. 2 and 1349 in Agesthialingom and Sakthivel 1973). Examples from the Muslim dialect of the Kanniyakumari district given below are therefore taken from a well-known novel (Tōppil Muhammatu 2004).

Script and phonological assimilation

Given the dearth of research on Arabic loanwords in spoken varieties of Tamil, most of our information regarding the assimilation of Arabic sounds to Tamil phonology has to depend on written sources. These employ both the Tamil and a modified version of the Arabic script which shows similarities to versions in use in Southeast Asia (cf. Shuʾayb 1993:95–99; Tschacher 2001:6–18; Vinson 1895). Obviously, while the use of the Arabic script often facilitates reading, it usually obscures sound changes. Yet, use of the Tamil script does not necessarily facilitate the identification of such changes, either. The distinction between voiceless and voiced consonants and between plosives and fricatives is not phonemic in Tamil nor are there separate letters to write these phones (cf. Andronov 2003:27–31). For example, the phoneme /c/ is pronounced [c], [ɟ], [s], or [ɕ] in different contexts, but it is written with just one letter, transliterated as c. This often makes it difficult to glean the actual pronunciation of Arabic words from the orthographic representation. Furthermore, there is substantial orthographic variation. While earlier spellings often used epenthetic letters to indicate actual pronunciation, there has been a tendency to drop such letters in recent times and to come closer to a transliteration of Arabic words with the help of so-called grantha letters originally used solely to write Sanskrit (cf. Das's [1981:346] distinction of ‘scientific’ and ‘vulgar’ systems of transliterating Arabic in Tamil).

There is no space to discuss all phonological assimilations and orthographic conventions here; for a short overview of the latter, see Das (1981). It will suffice to indicate some of the more common and interesting ones. Vowels remain largely unchanged; final ī/-ū are usually shortened. Initial wa-, ya- commonly change to o-, e-, e.g. oli ‘saint; guardian’ (< walī), ekūti ‘Jew’ (< yahūdī). /ʾ/ tends to disappear in syllable-final position but lengthens a preceding vowel, e.g. mūmiṉ ‘believer’ (< muʾmin). /ʾ/, /ʿ/, and /ḥ/ tend to become /y/ in intervocalic position preceding i/ī, e.g. kāyip ‘hidden’ (< ġāʾib), cayītu ‘fortunate’ (a personal name, < saʿīd), cāyapu ‘companion’ (also an honorific title; < ṣāḥib). In words which are in common spoken use, more radical changes occur, e.g. mōtiṉ ‘muezzin’ (< muʾaḏḏin) or the personal name Meytīṉ (< Muḥyī d-Dīn). Intervocalic voiceless stops and fricatives are occasionally voiced, e.g. āġir ‘last’ (< āxir; cf. Vinson 1895:154, 159).

Most interesting are the reflexes of Arabic /ḍ/ and /ḏ̣/, as these often allow one to distinguish loans borrowed through Indian Ocean networks from those borrowed from northern India. The most widespread reflex of both phonemes, common in earlier Islamic Tamil literature but also widespread in spoken Tamil among Muslims in both India and Sri Lanka, is a lateral, either dental /l/ or retroflex /Ḥ/, e.g. paṟulu/paṟuḷu ‘duty’ (< farḍ). A lateral reflex of /ḍ/ is found also in several Southeast Asian and West African languages ( ḍād ), and in the Tamil context obviously reflects borrowing through Indian Ocean networks. More surprising is that /ḍ/ and /ḏ̣/ have identical reflexes, as most languages exhibiting a lateral reflex of /ḍ/ treat the two phonemes differently. Yet, it is possible that there were originally different reflexes of /ḍ/ and /ḏ̣/, one represented by /l/ or /ḷ/ and one by the retroflex approximant /ḻ/ [ɻ]. From the 13th century onward, the phoneme /l/ disappeared from spoken Tamil, merging most commonly with /ḷ/, /l/, or /y/ (cf. Andronov 2003:39, 86), thus obscuring the distinction between the reflexes of /ḍ/ and /ḏ̣/. Occasional use of /l/ for both /ḍ/ and /ḏ̣/, e.g. in the divine name Kapīlu ‘Guardian’ (< (al-)ḥafīḏ̣), and in the Kanniyakumari dialect word hālir ‘presence’ (< ḥāḍir), may corroborate this scenario. Whatever the case, it is probable that Tamil is the source of those Arabic loanwords in Malay that exhibit /l/ for /ḏ̣/ (Indonesian/Malay).

In contrast to these lateral reflexes, words borrowed through Persian or Urdu tend to be written with the grantha letter j, e.g. kāji ‘judge’ (< qāḍī). This reflex is especially common in administrative terms, most of which derive from Indo-Persian vocabulary. In rare cases, one also encounters /t/ for /ḍ/, obviously influenced by contemporary Arabic pronunciation.

Morphological assimilation

The vast majority of Arabic loanwords in Tamil are nouns. Rather than borrowing verbs, nouns are combined with a Tamil verb to produce a new verb phrase, e.g. in Kanniyakumari dialect hālir ā- ‘to become present, appear’. Words are usually borrowed in the pausal form without article. The common word-final -u (rarely -i) is not a reflex of Arabic case endings but rather a paragogic vowel added to avoid phonotactically restricted final consonants, e.g. napucu ‘desire, lust; soul’ (< nafs ‘self, soul’), oki ‘revelation’ (< waḥy). This phenomenon shows great similarities with comparable forms in Malay and other Southeast Asian languages (cf. Versteegh 2003; Acehnese; Indonesian/Malay). That there is no obvious explanation for this phenomenon in Southeast Asian languages, while it is clearly phonotactically motivated in Tamil, makes it likely that Tamil was the source of the respective Arabic loanwords in Malay, as was already suggested by Bausani (1971:475, 477, n. 11) with regard to the same phenomenon in Persian loans. Tamil has both -ā and -attu as reflexes of the Arabic feminine ending -a(t). Which of the two reflexes is used seems to be tied to the particular loanword, e.g. kalimā ‘the profession of faith’ (< kalima ‘word’) vs. cūrattu ‘chapter of the Qurʾān’ (< sūra). There are few semantic doublets of the kind seen in Persian, and those that exist seem to have been borrowed from either Persian or Urdu.

While Arabic plurals, personal suffixes, or whole phrases are occasionally encountered, these borrowings are either of a scholarly nature or lexicalized, with the plurals often being treated as singulars to which the Tamil plural suffix -(k)kaḷ is added, e.g. ulamākkaḷ ‘religious scholars’ (< ʿulamāʾ). Tamil suffixes are freely added to Arabic loanwords, e.g. cūmaṉ ‘one who causes misfortune, the Devil’ (< šūm ‘misfortune’ + 3rd pers. masc. sg. suffix -aṉ), akatāy ‘as one’ (< aḥad ‘one’ + suffixed verbal participle -āy ‘having become’), Kanniyakumari dialect kāpiricci ‘infidel woman’ (< kāfir ‘infidel’ + colloquial 3rd pers. fem. sg. suffix -(i)cci). Borrowed nouns in common use which end in -m change this to ttu in the oblique case just like native nouns, e.g. iculām ‘Islam’ > iculāttu ‘Islam's’.

Semantic domains

Most Arabic loanwords in Tamil belong to just a few semantic domains. Probably the single most important one is the domain of Islamic practice and thought, as many of the examples given above attest to. There is also a common tendency for Arabic loanwords to be more religiously circumscribed in meaning in Tamil than they are in Arabic, e.g. kitāb ‘book’ > kittāpu ‘Islamic religious book; book using Arabic characters’ (cf. Thurston 1909:IV, 205). Even though many common institutions and practices of Islam are referred to by Tamil terms in everyday speech by Muslims and non-Muslims alike, often Arabic loans exist side by side with these Tamil words, e.g. Tamil paḷḷivācal ‘mosque’ and its Arabic-derived equivalent macūti (< masjid).

Another important semantic domain in which Arabic loanwords are not uncommon is administration and bureaucracy. The largest number of Arabic loans in common use by non-Muslim Tamils may belong to this category. Most of this vocabulary is borrowed either directly from Indo-Persian or through Urdu or English, and thus shows reflexes of Persian or Urdu pronunciation, e.g. jillā ‘district’ (< ḍilʿa ‘side’). While terms for specific administrative divisions, such as tālu(k)kā ‘subdistrict’ (< taʿalluqa ‘connection’), are generally known only in those coun-tries where the term is used in administration, other loans in this domain are more widespread, e.g. okkīl, vakkīl ‘advocate’ (< wakīl); pākki ‘rest, remainder’ (< (al-)bāqī).

A third domain is constituted by what Bausani (1971:477, n. 2) has called the “sailors’ international vocabulary of the Indian Ocean coasts”, which includes loans from Persian and Malay in addition to Arabic. Among the Arabic loans are mālimi, mālumi ‘captain, sailor’ (< muʿallim ‘teacher, master’) and campōkku ‘boat’ (< sunbūq). Interestingly, many of these terms show signs of being borrowed through Persian, e.g. cukkāṉ ‘rudder, helm’ (< sukkān), cukkāṉkiri ‘helmsman’ (< Persian sukkān-gīr).

Not all Arabic loanwords in Tamil fit into such neat semantic categories. Especially in literary texts, words often seem to be borrowed ad hoc. Colloquial Tamil, especially as spoken by Muslims, contains many Arabic loanwords beyond the confines of religious, administrative, and nautical vocabulary, such as capar ‘journey’ (< safar) or mauttu ‘death’ (< mawt). This also includes kinship terminology, such as the Shafiʿi Muslim kinship term ummā ‘mother’, a curious mixture of the respective Arabic and Tamil equivalents ʾumm and ammā. Like many other aspects of Arabic loans in Tamil, the semantic domains covered by Arabic vocabulary require further research (for Arabic loanwords in other Dravidian languages, Malayalam; Telugu).

Torsten Tschacher (University of Heidelberg)

  • Recommend Us