¹ Shackle, C., ‘“South-Western” elements in the language of the Ādi Granth’, BSOAS, XL, 1, 1977, 36–50CrossRefGoogle Scholar [hereafter referred to as ‘S-W elements’].

² Arabic elements are supposed, for present purposes, to have been transmitted exclusively through Persian and are transliterated accordingly.

³ ‘Ibād Allāh Gyānī, Gurū Granth aur urdū, Lahore, 1966, 108–209Google Scholar. This catalogue, which provides a useful starting-point, has been checked in detail against the indispensable AG concordance, Singh, Gurcaran, Ādi Granth śabad-anukramṇikā, 2 vols., Patiala, 1971Google Scholar. Passages and words from the AG are quoted from the text of the Śabdārath Srī Gurū Granth Sāhib jī, fourth ed., Amritsar, 1969Google Scholar, whose standard pagination is followed in references here.

⁴ e.g. andari (CDIAL 357Google Scholar), dūri (6495), raṅg- (10560), māha (9993); similarly bābā (9209, rather than Turkish).

⁵ e.g. kāinu ‘marriage(?)’ (p. 1167) < Ar. qayna, pl. qiyān ‘slave-girl’! This and other etymologies were clearly drawn by ‘Ibād Allāh from the classic guide to the AG, Singh, Kānh, Mahān koś (second ed., Patiala, 1960)Google Scholar. Typical of his time in his firm grasp of Muslim learning, Kānh Singh certainly erred on the side of generosity in suggesting Persian or Arabic etymologies for AG words, an error largely corrected in the commentary in the Śabdārath.

⁶ Except in such isolated cases as dast-giri > dastagīrī with subscript t (p. 724), versus the normal dasatagīrī (p. 727).

⁷ i.e. as in Platts, J. T., A dictionary of Urdū, Classical Hindī, and English, London, 1884.Google Scholar

⁸ No Persian words with appear as loans.

⁹ In the interest of brevity, meanings (of the AG word) are given only in unusual or unpredictable cases.

¹⁰ Indicates a word only appearing in a work attributed to Kabīr, whose language, given the distance of Benares from the Panjab, naturally shows many peculiarities. Words appearing only in the other major Bhagats are indicated as Nāmdev (N), Ravidās (R), Farid (F).

¹¹ Shackle, C., The Siraiki language of central Pakistan, London, 1976, 36–40Google Scholar, describes the characteristic changes noted in the current colloquial of south-western Panjab. These are, however, for the most part so widespread in the Panjab as a whole that their correspondence with the AG changes to be described here cannot be used to suggest distinctions in sixteenth-century dialectal realizations.

¹² -p- also results from the preference for the less usual member of Persian pairs, thus tupak > tupaka ‘cannon’; supaid > supedu ‘white’ (never with -ph-). Such pairs are, however, too infrequent to suggest possible dialectal origins in the Persian from which loans were made.

¹³ nadar alone survives today of these archaic -d- forms, with technical sense of ‘grace’ in Sikh religious usage, as opposed to najar/nazar ‘look’: there is no such differentiation of meaning m AG nadari.

¹⁴ But there is a peculiar change of s > h in the unique mahajidi (N).

¹⁵ The derivation is given by Platts, whose parallel etymology from ardana- (CDIAL 642Google Scholar) seems unlikely, however.

¹⁹ A similar reduction of a labial (through /-w-/?) is perhaps suggested by the unique qalam > kalaü.

¹⁷ A final syllable is truncated in the unique doza > doja (for usual dojaku); and apparently in alṯān > galatu ‘wallowing’ (also galatāna).

¹⁸ The -ai- of raibarī is interesting for its possible suggestion of the fronting of a before h (to /æh/): cf. rahbar > raibāriā. The spelling -aha- is normal in the AG for what would correspond in modern Hindi-Urdu to phonetic /æh/, in central Panjabi to // with high tone, but in modern Siraiki to /Ah/: cf. The Siraiki language, 32–3Google Scholar. Whether or not this can be used as an argument for the westward extension of fronting before /h/ to cover the west-central area of Gurū Nānak's native area since his death is a nice point. AG loans from Persian, at all events, regularly have written -aha-, except pahlvān > pahilavānaṛā; and pinhān > pahināmīā (both from Gurū Nānak, versus pinhān > pahanāma, in Gurū Arjan); also mana' > manahi.

¹⁹ It is very doubtful if the separation of bad 'amal > bada amala indicates breaking of a compound by a glottal stop rather than two distinct words, although compounds whose second member begins with a vowel are often run together, like sar-anjam > sarañjāma. The obvious rhyme-form masolā < mas'ala does not suggest a phonetic status for the glottal stop in this word. Finally, the realizations 'aṣa > asa, and 'afū > āphū, are probably to be taken as instances of the frequent metrical lengthening of tonic a, rather than as a reflex of initially realized 'ain.

²⁰ The reduction of ai > e, standard in khemā ‘tent’, supedu ‘white’, sekhu ‘Shaikh’, is perhaps too commonplace to be recorded separately: forms with -e- are given for all these words in Platts, although usually marked as ‘vulg.’.

²¹ Perhaps suggesting a fellow-faqir's derivation from σοφα! Contrast the usual ṣūf > sūphu (F).

²² The alternations are nicely illustrated by the parallel lines of Gurū Nanak: pīra paikāmara sālaka sādaka [ < ṣādiq] (p. 53), versus Dās, Gurū Rām: pīra paik bara sālika sādika (p. 358).Google Scholar

²³ The list could be extended by the inclusion of words that substitute native sir- for Persian ear-: sardār > siradāru; sarkār > sirakāra; saropā > sirapāu ‘robe of honour’. But these are better regarded as examples of the mixture of elements in loan-compounds to be discussed in section 2.

²⁴ The -u- of asulū suggests a parallel with some varieties of modern Siraiki: cf. The Siraiki language, 43.Google Scholar

²⁵ A special case is formed by loans with original -āh, which have ah in the AG, particularly the common daragaha ‘court’; kulaha ‘cap’; gunaha ‘sin’: the exclusive preference for ah may go back to the source of borrowing (cf. p. 75, n. 12). But several pairs with long or short vowels probably reflect metrical choices: these are allāh > alahu or alāhu; panāh > panaha (also panāhi); girah > giraha (also girāha). One is tempted to compare the confusion of /Ah/ and /ah/ in modern Siraiki: cf. The Siraiki language, 32–3.Google Scholar

²⁶ As still in modern Sindhi. The point is only worth making in view of the semi-official scheme currently being followed in Sikh transliterations of the AG, which is an uneasy com promise between the original script and the modern reading pronunciation: this involves omission of final -a and bracketing of final -i -u, thus sg. dir. ghar(u), obl. ghar, loc. ghar(i).

²⁷ Thus Kabir eha rāha calāī, f. sg. (p. 477), versus Nānak ehi calāe rāha m. pl. (p. 144).

²⁸ Perhaps also silk ‘thread’, f. in Urdu (and perhaps in Panjabi?), versus AG silaku, captive's halter', m.

²⁹ AG declensions are summarized in ‘S-W elements’, 40Google Scholar. Adjectives ending in a consonant in Persian usually add -a, with m. sg. -u.

³⁰ Other exceptions are very rare. Thus qudrat appears once as kudarata, but 82 times as kudarati.

³¹ Still regularly maintained in modern Siraiki, which has kept this declension: cf. The Siraiki language, 47–8Google Scholar. Thus AG sajāi, pl. sajāī, is paralleled by Sir. sazā, pi. sazā , versus Panj. sazā, pl. sazāv.

³² Which once yields the splendid parallel form turakū (N), for the usual turk > twraku ‘Muslim’.

³³ Also tejaṇi and tejanaṛī (with diminutive suffix), whose use in parallel contexts supports the suggested influence of tazī on tejī put forward in Smith, J. D. (ed.), The Vīsaḷadevarāsa, Cambridge, 1976, 312.Google Scholar

³⁴ Tilaṅg Kabīr, 1 (p. 727)Google Scholar; Tilaṅg M5, 1 (p. 723)Google Scholar; Tilaṅg Nāmdev, 1–2 (p. 727)Google Scholar; Tilaṅg M1, 1 (p. 721)Google Scholar; Mājh vār, 13Google Scholar, salok M1, 1–6 (pp. 143–4)Google Scholar; Malār vār, 27, salok M1, 2 (p. 1291)Google Scholar. The reason for the assignment of most of these hymns to rāg Tilaṅg is obscure, but this rag was clearly felt to have a special appropriateness for poems associated with Islam. It is, for instance, used in the apocryphal Tilaṅg kī vār attributed to Nānak, Gurū (Mohān koś, p. 444)Google Scholar, which is included in the Nasīhat-nāmā that forms the latter part of IOL MS Panj. B41 (see Shackle, C., Catalogue of the Panjabi and Sindhi manuscripts in the India Office Library, London, 1977, no. 27, II).Google Scholar

³⁵ In ‘S-W elements’, 36.Google Scholar

³⁶ cf. Singh, Piār (ed.), Janampatrī bābe Nānak jī kā, Amritsar, 1969, p. 105Google Scholar: kichu torkī bhī paṛiā. Odd sentences in a similar kind of language do appear in the prose hagiographies, as in the disputation with Shaikh Sharaf described in Singh, Vir (ed.), Purātan Janamsākhī, Amritsar, 1971, 54–6.Google Scholar

³⁷ The nasalized ending of the repeated karadã in Tilaṅg M5, 1 might be supposed to indicate nmdhi with the following words, i.e. as representing kardam or kardan.

³⁸ Though pākhāka (elsewhere paikhāka) is less true Persian than a caique on such compounds as caraṇadhūṛi: bakhīla (4) is rendered ‘envious’, rather than ‘stingy’ on the basis of the similar meaning regularly attached to the noun bakhīlī in the AG.

³⁹ cf. the use of āmada as 2 sg. in kujā āmada kujā raphatī kujā meravī (Tilaṅg Nāmdev, 2).Google Scholar

⁴⁰ Nor is the basic marker -rā used, except in janu turā (4), and in hamārā (p. 144) < hama-rā.

⁴¹ cf. such phrases as gusala karadana būda ‘you ought to have bathed’ (Tilaṅg Kabīr, 1Google Scholar), recalling Urdu usl karnā thā.

⁴² Listed in ‘S-W elements’, 37 (and n. 5).Google Scholar

⁴³ Usually applied to the washing of clothes in the AG (cf. pp. 4, 914, 1294)Google Scholar, but once by Gurū Nānak to the washing of hair (Āsā poṭī M1, p. 432Google Scholar). Perhaps one should also include masakala (< miṣqal) ‘burnisher’: interestingly, the usual words for ‘rusty’, jaṅgālī, jaṅgīlā (< zang), and ‘rust’, moracā (< morca), are also of Persian origin.

⁴⁴ This is regularly used as a noun in the AG, but as an adjective ‘happy’ in the prose janamsākhīs, whose Persian loans are quite frequently more distorted from their original, both orthographically and semantically, than those of the AG.

⁴⁵ As might be expected, there is an interestingly high rate of obsolescence among such terms of disparagement. AG phādilu corresponds to modern fazūl, and phāvā is obsolete, like jandāiu (< jandāl) ‘wretched, rude’ (applied only to Death), khāiku (< ? āyak ‘locust’) ‘loudmouth’: others, it is true, have maintained their popularity, like the perennial cugala ‘backbiter’, although not Iditabdru (< Id-i'tibār), regularly used by Gurū Nānak in the same sense, i.e., of ‘destroyer of trust’.

⁴⁶ The financial sense is established by its regular use in the AG with such words as pjī, rāsi ‘capital’.

⁴⁷ Also jañjīrī, a typical by-form belonging to the extended f. declension, which is used in the same sense (versus Persian zanjīrī ‘one in chains’).

⁴⁸ Combined with the former by Gurū Nānak to yield the unique pātisāhibu (Japu, 27, p. 6).Google Scholar

⁴⁹ On pp. 25, 464, 465, 1241.

⁵⁰ On pp. 5, 14 (in an unusual sense), 27, 59, 940, 1021, 1022.

⁵¹ Gaüṛī M5, 146 (p. 195).Google Scholar

⁵² Dhanāsarī M1, 7 (p. 662).Google Scholar

⁵³ Malār M1, 8 (p. 1257).Google Scholar

⁵⁴ As in Ravidās's remarkable evocation of the Heavenly City, termed begamapurā (< begẖam) ‘city without grief’ (Gaüṛī Ravidās, 2, p. 345).Google Scholar

⁵⁵ Mārū vār M5, salok ḍakhaṇe M5 (pp. 1094–1105).Google Scholar

⁵⁶ These strictures are not, of course, to be regarded as applying to the balanced discussion in McLeod, W. H., Gurū Nānak and the Sikh religion, Oxford, 1968, 158–61.Google Scholar

⁵⁷ The canonical status of the AO as a whole for Sikhs certainly encourages this composite view of the book, as in Kohli, S. S., A critical study of Adi Granth, New Delhi, 1961, 240–4Google Scholar, where a useful selection of verses relating to Islam is given, but without much attempt to differentiate between the different writers.

⁵⁸ The authenticity of many of these attributions has been called into question in Vaudeville, C., Kabīr, I, Oxford, 1974, 16–17, 58, 70–9Google Scholar, so ‘Kabir’ is here to be regarded as being used in the sense of ‘AG Kabir’. Since no textual study comparable to Vaudeville's yet exists for Nāmdev or Ravidās, their hymns (although sometimes relevant to the discussion) are hence forward disregarded.

⁵⁹ Mārū M1 sol., 12 (p. 1032).Google Scholar

⁶⁰ M¯rū vār M5, 17 (p. 1100).Google Scholar

⁶¹ cf. McLeod, , Gurū Nāmak, 187–9Google Scholar; Vaudeville, , Kabīr, 245–53.Google Scholar

⁶² Who does, however, receive the alliterating loan-epithets jamu jandāru (< jandāl) ‘wretched death’ (first used by Gurū Nānak), and jamu jāgātī (< zakẉātī) ‘death the poll-tax gatherer’ (first used by Gurū Amar Dās): both epithets become literary clichés in the later Gurūs.

⁶³ Cited only once, with typical learning, by Gurū Arjan: dojaki paüdā kiu rahai j¯ citi na hoi rasūli ‘falling into hell, how should one escape, when one does not remember the Prophet?’ Gaüṛī var 2, 5, salok M5, 2, p. 319–20).Google Scholar

⁶⁴ It still receives powerfully fresh treatment as late as the period around 1800, as in the Siraiki sīḥarfī by Muḥammad Jamāl Multāni: cf. Haq, Mahar Abdul (ed.), Nur-e-Jamal, tr. Shackle, C., Multan, 1977.Google Scholar

⁶⁵ Salok Farīd, 1 (p. 1377).Google Scholar

⁶⁶ Gaüṛī vār 1, 27 (p. 315).Google Scholar

⁶⁷ e.g. salok Farīd, 33, 44–6 (pp. 1379–80).Google Scholar

⁶⁸ Āsā, M1 a., 12 (p. 417)Google Scholar: notable among other handlings of this theme by Gurū Nānak is the hymn Sirī M1, 6 (p. 16).Google Scholar

⁶⁹ Salok Farīd, 47 (p. 1380).Google Scholar

⁷⁰ Malār vār, 19, salok M1, 1 (p. 1286)Google Scholar, following an interesting reference to the Ṣūfī custom of bestowing caps (kulah ) on disciples.

⁷¹ Sirī M1, 28 (p. 24).Google Scholar The typically Muslim word gora ‘grave’ is used on only one other occasion by Nānak, Gurū, in a ‘Torkī’ verse (p. 1291)Google Scholar, but six times by Farīd.

⁷² Japu, 21 (p. 4).Google Scholar

⁷³ Soraṭhi Kabīr, 1 (p. 654).Google Scholar

⁷⁴ Āsā Kabīr, 8 (p. 477).Google Scholar

⁷⁵ Rāmakalī vār 1, 11, salok M1, 1 (p. 951).Google Scholar

⁷⁶ Among a gallery of such types, there is, for instance, a memorable reappearance of the corrupt qāḍī at the beginning of Vāris Shāh's Hīr, the great eighteenth-century classic of Panjabi Muslim literature (see Hasan, Mumtaz (ed.), The adventures of Hir and Ranjha, tr. Usborne, C. F., Karachi, 1966, 30).Google Scholar

⁷⁷ Sāraṅg var, 22, salok M1, 1 (p. 1245)Google Scholar: cf. the similar juxtaposition of the false qāḍī, Brahman, , and Yogī, , in Dhanāsarī M1, 7 (p. 662).Google Scholar

⁷⁸ Singh, Vir (ed.), Purātan janamsākhī, Amritsar, 1971, 43.Google Scholar

⁷⁹ It is, of course, the Hindus who are associated with white clothes. Elsewhere in the AG, too, blue (n¯lā) clothes are associated with Muslims by Nānak, Gurū, as in Basani M1 a., 8 (p. 1191)Google Scholar, while their servile adoption by Hindus to curry favour is condemned in Āsā vār, 16, salok M1, 2 (p. 472).Google Scholar

⁸⁰ Rāmakatī M5, 9 (p. 885).Google Scholar

⁸¹ Āsā Kabīr, 13 (p. 478)Google Scholar, translated in full and discussed in Vaudeville, , 93.Google Scholar

⁸² Āsā Kabīr, 17 (p. 480).Google Scholar

⁸³ Cf. the peculiar association of terms in: manu mandaru tanu vesa kalandaru ghaṭa hī tīrathi nāvā ‘my mind is a temple, my body a qalandar's clothes, I do pilgrimage in the pool of the heart’ (Bilāval M1, 2, p. 795).Google Scholar

⁸⁴ M¯jh vār, 7, salok M1, 1 (pp. 140–1)Google Scholar; cf. the following salok also.

⁸⁵ i.e. sarīati, hakīkati, māraphati, hakīkati, which, like many other learned Islamic terms, occur in the AG only in this hymn, Mārū M5 sol., 12 (pp. 1083–4).Google Scholar

⁸⁶ Two loan-words in this line present problems. Most commentaries derive kadā from qad' ‘restraining’, some from qadḥ ‘reviling’, but both words are too unusual to be likely. Semantically, the commoner qat' seems a more probable source, although this etymology involves a unique voicing of -t-: yet a further possibility, involving no exceptional sound-change, is qaẓā in the sense of ‘settling, eliminating’. The word is used only here, as is pahiraṇu, whose derivation by the commentaries from Pers. pair¯han is perhaps possible, given a confusion with pahir- ‘wear’: peraṇa is used in the same sense elsewhere (p. 418), and there is one case of the unproblematic pirāhanu (p. 943)—both in hymns of Gurū Nānak.

⁸⁷ Sirī M1, 7 (p. 16).Google Scholar

⁸⁸ The equation does appear in Kabīr, though not in the AG: cf. Vaudeville, , 263Google Scholar. A typical modern exposition of the syncretic theory can be found, e.g., in Ahmad, Aziz, Studies in Islamic culture in the Indian environment, Oxford, 1964, 143 ff.Google Scholar, which is largely derived, like so many other rimilar accounts, from the unreliable Husain, Yusuf, L'Inde mystique au moyen Age, Paris, 1929.Google Scholar

⁸⁹ Prabhātī Kabīr, 2 (p. 1349).Google Scholar

⁹⁰ Sirī M1 a., 17 (p. 64).Google Scholar

⁹¹ The point is well made in McLeod, , 159Google Scholar, in relation to the words phakīru and daravesu. Cf. also the peculiar Rāmakalī vār 3 (pp. 966–8)Google Scholar, a bardic poem in praise of Gurū Aṅgad, in which a unique use is made of a number of Islamic terms in reference to Sikh institutions, e.g. umati for the Sikh community, but hardly suggesting any genuine syncretism.

⁹² But Islamic words which reflect this emphasis are usually rejected in the AG. Thus the word rabu (< Ar. rabb) is frequently used by Farīd (cf. ‘S-W elements’, 50Google Scholar), but very sparingly by the Gurūs, usually in special contexts where a Muslim word would be appropriate.

⁹³ Salok Farīd, 60 (p. 1381).Google Scholar

⁹⁴ Salok Farīd, 98 (p. 1383).Google Scholar

⁹⁵ Sūhī Kabīr, 3 (p. 792).Google Scholar

⁹⁶ But see further the discussion of some of these terms, especially hukamu and nadari, in McLeod, , 199–207.Google Scholar

⁹⁷ It has not been thought necessary to cite etymologies, since the more unusual ones have been explained in section I above: phonetic variants, metrical by-forms, and extensions have been grouped together for the purposes of enumeration here.

⁹⁸ The proportions have been determined by an approximate page-count of the AG, rather than by the usual method of counting the total number of hymns and couplets, which is less reliable. The adjustments have been made as follows (with fractions rounded up or down as appropriate):

The few couplets of Gurū Aṅgad (M2) have been disregarded.

⁹⁹ Also occasionally in Gurū Nānak, amaru (< amr) ‘order’, not readily to be distinguished in the concordance from amaru ‘immortal’.

¹⁰⁰ Including nadarī ‘God the Gracious’: this is not always to be distinguished, in either concordance or context, from inflexions and expansions of nadari. A similar difficulty attaches to the determination of the frequency of the less usual synonym karamu (< Ar. Karam), fatally liable to confusion with the very common native karamu ‘deed’ (CDIAL 2892).Google Scholar