¹ This visit, undertaken from January to March 1979, was made possible by a grant of overseas research leave from the School of Oriental and African Studies. It would be impossible to thank individually all those who assisted me in Pakistan, but I should like to express particular gratitude to M. Azeem Bhatti in Islamabad, to Ayub Sabir and Javed Shah in Kohat, and to Sabir Husain Imdad, Sh. Shaukat, and Z. I. Athar in Peshawar.

² Theories current in the literary circles of Peshawar are variously advanced by Fārigh Bukhārī, ‘Hindko adab’, in Mahmūd, Fayyāẓz (ed.), Tārikh-i adabīyāt-i musalmānān-i Pāk-o-Hind, XIV, pt. II, Lahore, 1971, 210–56Google Scholar, Nayyar, Mukhtār 'Alī, Tārīkh-i zabān-i hindko, Peshawar, 1977Google Scholar, and Shaukat, Sh., Hindko zabān aur adab kā tārīkhī jāiza, I, Peshawar, 1977, 15—23Google Scholar. In the Linguistic survey of India, VIII, pt. I, Calcutta, 1919, 234Google Scholar (subsequent references to the LSI are to this volume), Grierson took ‘Hindko’ to mean ‘the language of Hindus’, a definition naturally hotly disputed in Pakistan. Grierson also understood ‘Hindkī’ to be a synonym of ‘Hindko’, but the former term is used locally only in the sense of ‘Hindko-speaker’. The slightly pejorative connotations of ‘Hindkī’ (at least to Pathans) are avoided in the modern self-definitions of Hindko-speakers by such coinages as ‘Hindkūn’, which suggest parity with ‘Pashtūn’.

³ The criteria for Grierson's, classifications are set out in LSI, 239–43, 431—2Google Scholar.

⁴ Criticisms of Grierson's, scheme are discussed in my paper ‘Problems of classification in Pakistan Panjab’, TPS, 1979, 191–210Google Scholar. The present article seeks to derelop by concrete example some of the ideas sketched in this earlier paper.

⁵ The short account in LSI, 398–403 is not very informative. Differences from the central standard of Multan, described in my grammar, The Siraiki language of central Pakistan (SLCP), London, 1976Google Scholar, are relatively slight. They include such typical northernisms as ke ‘what’ and us ‘is his’, so ke nãvūs for Multani kia nã is ‘what is his name ?’, similarly the honorific Sæhaib ‘sir’ rather than the characteristic southern SΔĩ (< svāmin-), and some archaisms, e.g. Δla- ‘speak’ (< *ālāpayati), also found in the conservative Sirāiki of Sind for the central ḇol-.

The Hindko of Bannu district, referred to in the LSI (p. 404) as being similar to the northern Sirāikī of Mianwali (very imperfectly covered in my grammar), seems hardly to have survived the departure of the non-Muslim population in 1947. A missionary stationed in Bannu was responsible for the Hindko translation of John, St., published in Yūḥannā dī injīl, Lahore, 1929Google Scholar, but the language used is, if anything, Peshawar Hindko with a marked Panjabi influence.

⁶ This is a consequence of the movement described in my article ‘Siraiki: a language movement in Pakistan’, Modern Asian Studies, XI, 3, 1977, 379–403Google Scholar.

⁷ There is a quite good account of Dhannī, itself in LSI, 542–53Google Scholar, to be supplemented by the detailed phonetic description of the sub-dialect Kahūṇī in Varma, Siddheshwar, ‘The phonetics of Lahnda’, JRASB, letters, II, 1936, 47–118Google Scholar.

⁸ LSI, 449–57, to be taken with the ‘Lahndā of the Salt Range’ described in LSI, 433–48, in an account based on material in Wilson, J., Grammar and dictionary of Western Panjabi, Lahore, 1899Google Scholar. Bahri, Hardev, Lahndi phonology (with special reference to Awáṇkárí), Allahabad, 1962Google Scholar, and Lahndi phonetics (with special reference to Awáṇkárí), Allahabad, 1963Google Scholar, are studies of outstanding quality; but the sub-titles are important, since the books describe Avāṇkārī, not ‘Lahnda’. References to Avāṇkārī here are based on an analysis of the words and texts included in Lahndi phonetics (LPh).

⁹ LSI, 468–76, briefly added to in LPh, 16–17.

¹⁰ The awkwardness is due to Grierson's artificial grouping of these dialects with Poṭhohārī, which actually is spoken in the north-east. But Poṭhohārī, is distinguished by so many features (briefly summarized in LPh, 18–19)Google Scholar that it must be classified quite separately.

¹¹ LSI, 468.

¹² Diehter, D., The North-West Frontier of West Pakistan, Oxford, 1967, 73Google Scholar provides a map of tribal distribution and linguistic allegiance.

¹³ LSI, 565—65, and the useful description of Kāgāni in Bailey, T. G., Linguistic studies from the Himalayas, London, 1915, 87–112Google Scholar.

¹⁴ LSI, 458–67, which includes an interesting local account of the distribution of languages by tribe.

¹⁵ The expression is deliberately vague, since—as the census-takers of British India repeatedly admitted in their reports—the complex patterns of diglossia and bilingualism in the NWFP make even approximate enumerations of language-speakers difficult to arrive at. On the basis of the 1961 figures (the 1972 census omits all language figures) it might be hazarded that there are upwards of 30,000 native speakers of Kohāṭī, perhaps more than ten times that number of speakers of Peshawar Hindko.

¹⁶ I have examined some of the implications of this and related developments in ‘Language, dialect, and local identity in Northern Pakistan’, a paper to be published in the proceedings of the colloquium on ‘Pakistan in its 4th decade’ organized by the Deutsches Orient-Institut. Hamburg, in May 1980.

¹⁷ LSI, 554–564.

¹⁸ Other language-names are abbreviated according to the system used in CDIAL, together with Urdu (U.). Forms of the older language of the Ādi Granth are cited as AG.: salient features of the morphology are summarized in Shackle, C., ‘“South-Western” elements in the language of the Ādi Granth’, BSOAS, XL, 1, 1977, 36–50CrossRefGoogle Scholar.

¹⁹ The only text I have seen is the pair of couplets by ‘Azīz Akhtar Vāriṣi cited in Bukhāri, Fārigh, loc. cit., 250Google Scholar.

²⁰ In order to provide a reasonable basis for comparison the transcription here follows the system used in SLCP, with the addition of the cirumflex to mark the high-falling tone. Conversion from Bahri's system is normally straightforward, except that my transcription does not distinguish ǝ as an unstressed allophone of Δ, although it does differentiate positional variants of diphthongs beginning with front vowels, e.g. accented ia and ea versus postaccentual ia and ya, cf. SLCP, 14–15.

²¹ cf. LPh, 18 and Varma, art. cit., on Dha.

²² The word is used with some hesitation, but regressive vowel-harmony from the Ap. masc. nom. sg. -u looks the likeliest point of origin.

²³ This of course makes it impossible to form an idea of the internal consistency of Ghe., but cf. LPh, 16–17.

²⁴ Bahri, describes 1 as ‘a very important and peculiar sound in Awankari’, LPh, 116–17Google Scholar, but does not transcribe it in his texts. It is certainly clear that the maintenance of ḷ, at least outside the west-central P. area, is occasional only, and that there is no need to follow Grierson in attaching much importance to this feature as a major dialectal shibboleth.

²⁵ cf. SLCP, 13, 16–17.

²⁶ It seems best to avoid disturbing the CP. rule which forbids gemination of y (and h), cf. SLCP, 27, and to accept the inelegance which results from VΛi-, VΛie versus pi-, pive (with regular -V-), since the inf. vΛyyna would conflict with the rule of simplification of geminates shown by dΛSS-. dΛsṇa. A sound phonetically similar to that in Ko. Λi, VΛi- is found in Sir. iho, an equally isolated form (SLCP, 23).

²⁷ An analysis of the texts in LPh reveals th e preference for extension in agentive an d locative uses, which would be expected on the basis of the older loc.-instr. -i, -ai, but this preference certainly does not constitute any absolute rule. Parallel use of extended and unextended forms was also observed in the speech of Ko. informants.

²⁸ But cf. note 27 for extension as agentive marker.

²⁹ Bahri's notation putrā-ẵ should perhaps be taken to represent an overlong vowel, but some cis-Indus sub-dialects may have ko with pl. nouns.

³⁰ Dialectal distribution within Av. is shown on the map in LPh, 25.

³¹ The CP. norm consists of compound derivatives of yādṛśa-, e.g. P. Ajêa, Sir. ejhã, etc., although the corresponding U. æsa is also spreading.

³² Though SΛk- is often substituted in the Ko. of Kohat City, apparently to avoid the embarrassing homophony with hΛgg- ‘defecate’ (<*hagga-, cf. CDIAL hadati).

³³ Explained in CD1AL as derivatives of *vrañjati and vrajati respectively.

³⁴ For this pte. in Sir., cf. SLCP, 85, 126, 128–9. It is certainly found in Av. also: the derivation from Pk. -iā is exemplified in Lahndi phonology, 85.

³⁵ The original case of th e inf. in this construction was not obl., but loc.-instr., cf. AG., karaṇi laggā. LSI, 459Google Scholar does cite -Λṇ as well as -ṇe with lΛgga for Ko., bu t I was unable to confirm this usage locally.

³⁶ Although not always in Av., Ghe., which distinguish transitives as pres. ptc. kΛrena versus inf kΛrna.

³⁷ It is to be explained as an instance of the generalization of post-accentual -o- (cf. note 22), i.e. as a derivative of -anti, not of -antu.

³⁸ The ambiguity, while structurally interesting, is practically more apparent than real, since it may always be easily removed by use of modal intensives, e.g. mΛr Věsã versus mar děsā

³⁹ The forms cited for Ghe. in LSI, 469 are misleading: cf. also th e Av.p. an d Av.v. sets given in LPh, 27.

⁴⁰ The original base is āh- (with āhi- > ǣ-), explained by the CDIAL as derivative from ākṣeti rather than āsīt.

⁴¹ Att. has also the fem. 2 sg. kΛrni, 3 sg. kΛrni, but Ghe. and Av. share the Ko. fern, forms.

⁴² cf. SLCP, 101 ff. for the Sir. dir. suffixes.

⁴³ Bahri cites Av. forms with 1 pi. -se, the standard Sir. suffix.

⁴⁴ SLCP, 120 ff.

⁴⁵ The illustrative passages given in LSI, 462–7 provide examples. The subject is well worth exploring, but could only be dealt with properly by one more competent in Pashto than the present writer.

⁴⁶ Dani, Ahmad Hasan, Peshawar, historic city of the Frontier, Peshawar, 1969, 164Google Scholar.

⁴⁷ LSI, 554.

⁴⁸ The claim is supported in Pe. literary circles by reference to the tradition of verse composition in Pe., attested for the last 200 years, and now being continued in the beginnings of a modern literature, besides being used in a few programmes broadcast by Peshawar radio and television stations: for details cf. Fārigh Bukhārī, loc. cit.

⁴⁹ Hamdāni, Raẓa, ed., Sāīṅ Aḥmad 'All Pishāvarī, Islamabad, 1977, 11Google Scholar: ‘While Persian-speakers are scattered throughout the Frontier province, they are found in greatest numbers in Peshawar and Kohat. They are called fārsīvān, and include the Qizilbāsh, Kābulī, Durrānī, Shahzādagān, and some Bukhārī tribes who migrated to Peshawar during the Russian revolution.’

⁵⁰ A similar influence has been operating over the last generation on the Sir. of such cities as Multan, , cf. SLCP, 2–3Google Scholar.

⁵¹ Such is the consensus of local opinion: I found it to be borne out by a visit to the village of Chuha Gujar.

⁵² These were noted on a visit to Wad Pagga and some surrounding villages. I have no information on more distant rural varieties of Pe., though Att. features might well be more prominent towards the east. Raẓā Hamdānī, loc. cit., summarizes the distribution of Pe. as being ‘in the Vale of Peshawar, Peshawar itself and a number of nearby villages, such as the Tappa Khālsā; and in Nowshera tehsil, Akbarpura, Nowshera Kalan, Akora Khattak and a number of neighbouring villages’.

⁵³ Awan, Elahi Bakhsh Akhtar, The phonology of the verbal phrase in Hindko, University of London Ph.D. thesis, 1974 (unpublished)Google Scholar, is a minute study of a restricted corpus, of interest both for its phrasal analysis and as the work of a native Pe.-speaker.

⁵⁴ Similarly in Dha., cf. Siddheshwar Varma, loc. cit. Po. follows P. having a low-rising tone, but the phonetic realization (with marked emphasis of the rising tail in the post-tonic syllable) is quite distinctive. Neither group provides a parallel to the peculiar evolution of Pe.

⁵⁵ This is the usual analysis, although the following glottal constriction obscures the loss of voicing.

⁵⁶ cf. Nayyar, Mukhtār 'Ali, Hindko naṣur dī kahāṇī, Peshawar, 1965, 34—42Google Scholar, where much is made of its importance as a marker distinguishing Pe. from P.

⁵⁷ Nayyar's adaptation of the U. alphabet to Pe., followed by several other writers, records 'as he without the subscript shosha, but does not record the tone as such, i.e. graphic b'ār represents b'âr, but bār may represent either bar or bâr.

⁵⁸ It is this structural similarity which needs to be underlined in the context of CP., where so much discussion has been devoted to the aspirate-derived tones. Phonetically, of course, it is the contrast with P. which is striking, particularly in secondary features associated with the tones such as the lengthened vowel associated with the P. low-rising tone, versus the shortening that characterizes the high-falling tone everywhere, including the special Pe. tone with glottal constriction.

⁵⁹ Although the retroflex ṇ is of course distinct from n. Arguments over how best to devise a sign in the U. script for Pe. ṇ have distracted attention from the more difficult problems of writing the tones, the main candidates being the Sir. nūn with superscript ṭoe, Nayyar's re with superscript dot and ṭoe, and the combination nūn + ṛe suggested in Imdād, Ṣābir Ḥusaīn, Hindko rasmu'l-khaṯ: hik bahīs, Peshawar, 1978Google Scholar.

⁶⁰ P. is similar to Ko.-Hko., with the short vowel y, i.e. pyo, gya, etc. It is tempting to compare the Pe. preference for final -i versus CP. -e in emphatic forms of the pronouns.

⁶¹ The back quality of Pe. Λ helps account for the alternation Λ ~ a: VΛste has also been noted for Ko.

⁶² But Kha. appears to agree with city Pe. in the other common postpositions.

⁶³ Nayyar, Mukhtār 'Ali, Hindko qavā'id [written in Pe.], Peshawar, 1976, 79Google Scholar. Such peculiar mixtures are in fact quite characteristic of educated Pe. speech, where the alternative possibility oṛiñja ‘49’ is considered vulgar, and nmanve ‘99’ is similarly preferred to noṛmve.

⁶⁴ Their origin is doubtful, though one might postulate *suhā- < śubha-, on the basis of the loss of aspirate-derived tone in the corresponding t-forms. There is no S-form for the dir.

⁶⁵ So, while the vowel is characteristic of Hko., the stem-alternation brings Pe. into line rather with P., where poss. OSda, regularly written usdā in Gurmukhi, alternates with colloquial ôda (i.e. <uhdā, ohdā).

⁶⁶ Other simple pronominals do not inflect in any distinctive fashion, e.g. hor ‘other’ has Pe. obl. pl. horã, without the characteristic CP. -n- (P., etc., hornā). For ‘all’ Pe. shares the Hko. preference for sara over P. SΛb, Sir. SΛbh (where the pronominal declension ia again typical).

⁶⁷ An interesting contrast to P., where *kitthō tō is hardly likely.

⁶⁸ -v- is also retained, as such, in Ghe., which has avṇa (versus pres. ptc. ana).

⁶⁹ Also in the fut. 3 pl., where the vowel of mārsan in the paradigm given in LSI, 555 is quite misleading.

⁷⁰ The transference of sense may be compared with the development, in a different direction, of the U. polite imperative in -ie.

⁷¹ As against th e regular appearance of -v - in the 2 sg. aorist imperatives javĩ. pivĩ, rΛvĩ. The similar P. pattern is exemplified in the paradigms in Gill, H. S. and Gleason, H. A., A reference grammar of Punjabi, 2nd ed., Patiala, 1969, 22–8Google Scholar.

⁷² Particularly 1 sg. -ã, which is also found in the 1 pl. forms of the Pe. substantive verb. The parallel with U. 1 pl. -e should perhaps be pointed out.

⁷³ In written texts, especially in verse, 2 sg. forms in -o (-õ) are occasionally met with, e.g. je tu khãdo ‘if you had eaten’. These may be compared with the similar forms cited for the P. of Wazirabad in Cummings, T. F. and Bailey, T. G., Panjabi manual and grammar, Calcutta, 1925, 378–81Google Scholar. As in this variety of P., simple past forms used in conditional clauses may also have this ending in Pe., e.g. Λgor tu jit giõ ‘if you win’. This may be the explanation of the 2 sg. geö cited as a simple past form for Ko. in LSI, 461.

⁷⁴ Such symmetry is general in groups with a sigmatic fut., e.g. Hko. -ã, -Sã, Sir. -ũ, -sũ: but P. has 1 pl. fut. kΛrãge versus pres. subj.kΛrie.

⁷⁵ Although a famous shibboleth of Pe., it is not mentioned by Grierson, whose statement in LSI, 555 that ‘there is a negative verb substantive nayyā, I am not, equivalent to the Standard nimhǖu' is quite misleading. Pe. nĩga is different in both origin and function from the Sir. negative pres. discussed in SLOP, 107.

⁷⁶ Reference is to pages of Nayyar, Mukhtār 'Ali, M'atlẵ (hindko ẓarbu'l-amṣāl), Peshawar, 1974Google Scholar. The limitations of the script (cf. note 57) make phonemic transcription more appropriate than exact transliteration here and in quotations below.

⁷⁷ There is an exact parallel in Sir., where 3 sg. fem. hṣi is often also used for masc. ha.

⁷⁸ The si-past is often felt to be ‘wrong’ in Pe., and is criticized by Nayyar, , Hindko qavā'id, 71Google Scholar, on the grounds that it leads to confusion with the fut. -si. Of course, it is just those speakers who favour the si-past who will prefer fut. forms in -ega. This is an interesting case of structural adjustment.

⁷⁹ The rule is not absolute, however. Nayyar, , op. cit., 51Google Scholar, gives -da throughout in his paradigm of the imperfect, but elsewhere also writes forms with -na, e.g. Λsi ja SΛkne æ M19.

⁸⁰ Bailey's account is still the best; see Cummings, and Bailey, , op. cit., 349–60Google Scholar.

⁸¹ As opposed to the Sir. -o-, so + 3 sg. kitos ~kiⁱtosi, + 1 pl. kl¹tose, etc. The lack of such steins seems to be a general Hko. feature, to judge from the evidence of Bahri's texts. It is not, however, characteristic of the western P. described by Bailey, who clearly states (p. 356), ‘To express the simple past with a sums we must use the -o form or the pluperfect. The ordinary past if followed by a suffix, has the force of a present perfect’.

⁸² No attempt is made to distinguish these formally, since little would be added to the scheme of transformations suggested in SLCP, 150–3.

⁸³ At least in the commonest instance of the 3 sg., although there is of course a theoretical distinction elsewhere between conditional -da and pres. -na, e.g. 1 sg. Λgor mǣ kǣdΛs ‘if I were to tell him’, versus mǣ kǣdΛs ‘I tell him’.

⁸⁴ Quoted from Nayyar, , Hindko naşur dī kahāṇi, 8, 10Google Scholar.

⁸⁵ cf. Nayyar, , Hindko qavā'id, 69–70Google Scholar.

⁸⁶ The construction is also spreading to Sir., cf. SLCP, 123.

⁸⁷ This is to be contrasted with Sir., which uses the ordinary past ptc. with kΛr- (SLCP, 132). The position in Hko. is perhaps similar.

⁸⁸ Of course Hko. itself has many P. loans, as is made admirably clear for Av. by Bahri, in Lahndi phonology, 130 ffGoogle Scholar.

⁸⁹ An unpublished dictionary of 12,000 words has been prepared by Nayyar with the title Kliazāna.

⁹⁰ The problems of classificatory references posed by historical absence of a full locally based standard for CP. have been discussed in my ‘Problems of classification’, 198 ff. These problems are of course further exacerbated in the NWFP by the fact that it is Pashto which is the chief challenger to the standard status of U.

⁹¹ The settlement of the important Tappa Khālsā area is, for instance, dated by its inhabitants to the time of the emperor Aurangzeb.

⁹² The theory has been advanced that Pe., as the first city encountered on the passage through the Khyber, actually represents the original form of U., cf. Bukhāri, Fārigh, ‘Sarḥad mē urdū’, Urdū, 01 1955, 81–138Google Scholar, which attempts a rebuttal of Sherānī's well-known thesis of the P. origin of U. The hypothesis must be regarded with some reserve until the historical status of Pc. is better understood.

⁹³ The Pathan-oriented concerns of most writers who have dealt with the NWFP mean that the former have been much the more extensively discussed, although with far from conclusive results, cf. the classic account in Caroe, O., The Pathans 550 B.C.–A.D. 1957, London, 1948Google Scholar.