^{page 310 note 1} For the purpose of this paradigm—but here only—I have ventured to construct some forms not actually noted from usage. A junction-vowel may or may not be inserted before a termination beginning with a consonant. Where it is inserted vowel-harmony may influence it: dekk u-vāyi, dekk-u-no, dekk-i-ni, dekk-(a)-na, dekk-i-ven. In the remaining cases the vowel is invariable, except that -o- often replaces -a- in the 3rd sing, present and the continuative verbal, as in kerosu, kerotu. But see also the following footnote.

^{page 311 note 1} Such forms as dekkelte (dekkeste) are perhaps best explained as orthographic variations. See p. 314, footnote 2. Vocalic roots certainly base all these verbals on the past participle as well as on the simple stem, e.g. rhiyette alongside rhātte. Consonantal stems might be expected to do the same. But, even if they did, the uncertainty of Saurāshtran orthography would obscure the distinction.

^{page 311 note 2} By analogy pronominalized passive verbals with the long -ā should be found. If such passives occur (*janāttiso, such as is known) they would of course be quite distinct from forms in -yā-, cited p. 322. But the distinction could disappear, since the y- in -yā- is omitted in some past plural forms, as avāsi for avyāsi (p. 314). Therefore an actual form might be ambiguous.

^{page 311 note 3} See p. 326, footnote 1, for Madurek. Yavetīs = aveti (subjunctive), with the emphatic suffix -s(ī) or -s(ī), which always lengthens the preceding vowel.

^{page 311 note 4} For the inserted -v- see Beames, vol. iii, p. 68 (§ 24). A form like khavāyi is common to Saurāshtran and Gujarātī.

^{page 312 note 1} The tendency to develop a systematic negative conjugation is charracteristic also of Kōṅkaṇī (see LSI., vii, p. 171). But no genuine negative base has been developed. The “negative auxiliary” supplies a few forms such as nhī or nhā, nhītte or nhātte, nhītto, nhāsto, which in a disaspirated form provide the terminations of the “negative conjugation” given here.

^{page 313 note 1} N.S. 67. Gōśi is a combination of a particle and a suffix of emphasis, gō and -si (or -śi).

^{page 313 note 2} The plural of sani noun-infinitive of sā-, to see, is used here as a concrete noun, “eyes”.

^{page 314 note 1} The language of this publication (Madras, 1924, Nāgarī character) differs considerably from Rāma Rāo's, e.g. oṇṭe for oṇṭo; amre for amro. The Urdu word jawāb (even if it is the usual word), would hardly be found in his publications; although he does include in his Nandinighaṇṭitu three or four Urdu words, e.g. nasīb (fate), yādu (memory). See also p. 326, footnote 1.

^{page 314 note 2} The contrast between the characteristic -e- of the past singular and -a- (-o-) of the 2nd and 3rd persons present singular is steadily maintained, in spíte of the orthographic (= phonetic) uncertainty of ě, ă, and ǒ'adverted to in footnote 2 to page 321. The reason may be that the contrast is here protected by a harmonized contrast in the terminal vowels (menesi; menasu or menosu)—whether the terminal sound is a full vowel or a resonance only. There is no such harmonic safeguard in the case of dekkette and dekkatte (dekkoste), and therefore it is precarious to see in these forms a morphological contrast (dekkette past, dekkatte and dekkoste present). Nevertheless, the 1st person present dekkusu is absolutely fixed in contrast with the 3rd person dekkasu or dekkosu; although the terminal sound here supplies no supporting contrast.

^{page 314 note 3} If it is permissible to suppose that the so-called subjunctive is formed with the same “participial theme” -t which forms the present participle, there will be complete parallelism of verbals based on the present stem and verbals based on the past stem:— jā-t(u) when going, jiye-t(i) when gone; jā-t-te which goes, the act of going now, jiye-t-te which went, the act of going then; jā-t-teno the person who goes, jiye-t-teno the person who went. The -st- verbals may be analogously based on the present tense jās(u) and the past tense jiyes(i).

^{page 315 note 1} The heading of any episode in the Saṃgīṭa-Rāmāyaṇu will provide an example of the -ni infinitive, e.g. Hanumaṃt hāt ghāmu lī Lamkhīṇi [sic] khāl poṇni (Laṅkini's falling down on receiving a blow at the hand of Hanumân).

^{page 315 note 2} Compare hōna (hōnajāsu, is going to be), and the infinitive rhāna implied in the desiderative rhānākkāmu. In form this infinitive is indistinguishable from the negative hōna (is not). To add to the difficulty Saurāshtran uses -nā as an enclitic of emphasis in affirmative statements.

^{page 315 note 3} In the case of infinitives, as of absohitives in a following section, numerous examples have been given with the purpose of providing a sample vocabulary of Saurāshtran verbs.

^{page 316 note 1} See, however, p. 322, footnote 1.

^{page 317 note 1} The -i may become -y when added to vocalic stems, e.g. hoy, gey, sey.

^{page 317 note 2} These are absolutives of the reflexive stem in -l- which Rāma Rāo calls the ātmanepada, and exemplifies by saṅgulusu, “I say for myself”, as contrasted with saṅqusu (BSOAS., xi, Part 1, p. 118). It is the same formation as Hindī kah l⋯gā, contrasted with kah⋯gā. In Hindī the verb lenā “to take” has a more frequent independent existence than the corresponding verb in Saurāshtran. The Saurāshtran forms noted are: līśi “she took”; liyās “they took”; lī (absolutive); lēt(u) (present participle).

^{page 317 note 3} But it is a loosely attached suffix, found also with the continuative: pusilētkan hotyās, they were asking.

^{page 317 note 4} There is no justification for the doubling of -l-, which has come about from the analogy of such forms as kelli (for ker-li) and dhelli (for dher-li).

^{page 317 note 5} Mr. Alfred Master informs me that Kōṅkaṇī uses muṇ, moṇ, muṇu, maṇto in the sense of iti, like Sau. meni, mento. He therefore connects Sau. men- with Marāṭhī mhaṇ-, “say”, which has a similar use (Bloch, La langue marathe, p. 272; where the parallel use of Dravidian verbs meaning “say” is noted).

^{page 318 note 1} Again, the doubling of -l- is unjustifiable. See penultimate footnote.

^{page 318 note 2} It has the normal participal function and is never a substantive. The sense is not passive, and special passive forms are in use. See p. 311.

^{page 319 note 1} See below (pp. 321–2) for the plural form hoyātte.

^{page 319 note 2} The three formations have been grouped separately from the obviously pronominalized verbals, which are dealt with in the next section, because the dative form -ttoku (-ttaka, -staka) has noun-inflection. But it may be that the -tte formation is pronominal. See p. 320, footnote 3.

^{page 320 note 1} It is not always easy to draw the line between verbal substantives and adjectives, e.g. in such a passage as ghyāsi poḁste sommaru, morijāste bara (dying is better than suffering: translating the Sanskrit mama maranam ēva varam). The writer would have translated mama by mī, not moro ═ I-suffering is better than I-dying. But this is in accordance with Dravidian syntax, and does not furnish any criterion. Certainly the verbal in -tte is often a substantive. For instance, cǒratte is a synonym of cōru, theft (the act of thieving). And it can be used with an inflection: kaḷānātteka hāli “through not being familiar with”. See, however, the next footnote but one.

^{page 320 note 2} Examples of consonantal roots in -ette (-este) are quite common, but it is not clear that they are based on the past participle. See page 311, footnote 1.

^{page 320 note 3} It is possible that -toku, -tak(a) in all these forms are distortions of the pronominal -teko, tek(a), and that there is no morphological distinction between them and a form like kaḷānātteka cited in the penultimate footnote. In that case -tte may be the neuter pronominalized verbal and these forms its inflections.

^{page 320 note 4} This pronominalization, whether with pronouns or adverbs of place and time, is so closely parallel with Tamil usage as to give the impression that Saurāshtran is (as it were) translating a Tamil way of speaking into Indo-Aryan words. It is a foreign idiom (but perhaps none the worse for that), completely naturalized in the language.

^{page 321 note 1} See p. 316 above, and note (1) to p. 322 below.

^{page 321 note 2} Kereste, kerasteṅko, gavosteṅko provide an exemplification of the uncertain orthography of short vowels. The spelling with e suggests that the formation is based on the past participle kere. But it would be unsafe to accept the suggestion. See above, p. 320, footnote 2.

^{page 321 note 3} It should be noted that throughout the Nīti-śambu, whenever Rāma Rāo adds the exclusive enclitic -s(i) to -teno, the form is -tenās, instead of the normal tenōs(i).

^{page 321 note 4} Corrigendum. —In the first part of this article, p. 120, line 6, hovāsi is wrongly given as the plural corresponding to the hoyesi (hoyo) form of the past. It should, of course, be hoyāsi. And in the same passage the word singular should be deleted twice (lines 3–4 and 5–6).

^{page 322 note 1} Speakers of the language might feel that hoyes- and hoyās- in the singular and plural pronominalized verbals hoyesteno and hoyāstenu are in fact hoyes(i) and hoyās(i), the singular and plural of the past tense: as if the forms meant “he-was, he” (═ he who was), and “theywere, they” (═ they who were). And it is not an impossible explanation of the -st- forms of these verbals. But it would not explain the -tt- forms, which also show the yā- base, as in hoyātte and hoyāttenu. Nor does it explain -st- forms in the negative conjugation (sānāstenu, those who do not see, etc.).

^{page 322 note 2} My impression is that the language is still in some respects “in the making”, and that in particular it is in the experimental stage in its use of these verbals, which are new ground for an Indo-Aryan language. See page 323 (sānāsthāmu).

^{page 322 note 3} All these examples have the force of a locative verbal noun. The full meaning of a (plural) pronominal locative is exemplified in: Bhovē dēśinum jivan jiyāttemāku thevuḁo jen dakṣiṇa Hindustānumu sthiravāsin hoya, “From among those who went to live in many countries certain folk settled in South India”.

^{page 323 note 1} The use of the inflected form of the -stān(u) verbal is precisely parallel. See above p. 319.

^{page 323 note 2} Veṅkaṭa Sūri's extensive work is by far the most important document in the language, both as literature and as linguistic evidence. Unfortunately I have been able to make little use of it, since my knowledge of the language has proved inadequate for the interpretation of the poem. The existing edition seems to be remarkably free from errors, but it gives no artha.

^{page 323 note 3} The print has ami hoyeāttenhi: which is no doubt the printer's conflation of two forms mī hoyettenhi and ami hoyāttenhi. At any rate mī hoyette and ami hoyātte are given as the two positive forms of the past in the bhāvārthalca-dhātu. The positive forms of the present are given as mī hōtte, ami hōtte; to both of which hōttenhi would be the corresponding negative.

^{page 325 note 1} Gīta-Gōvinda, p. 39 (Aṣṭ. viii. 1): Rūsur geye togo ravaḷḷānāsto mī mōsu geyy aparādu keresi mī, “I committed a fault in that I was deceived into not causing you to stay when you went in anger.” The long ā in ravaḷḷā-, however, calls for explanation.

^{page 325 note 2} Of the distorted forms in the gramophone records autiya seems to stand for avḁěyŏ (the long vowels must be an idiosyncrasy in the form avḁēyō above cited), and audas for avaḁesi: but autrya must represent avatu rhiye.

^{page 325 note 3} Mr. Alfred Master pointed this out to me. I was disinclined to admit a disaspiration of rh; but there is no room for doubt. The -riyo form is explained by the fact that rhiyo, hoyyo occur as variants of rhiye, hoyye, in the publications which use these disaspirated forms of rhā. See the next footnote. Similarly disaspirated forms of the negatives nhī, nhātte, etc., appear to be used as “terminations” in negative conjugation (see p. 312, footnote 1).

^{page 326 note 1} Amre is used for Rāma Rāo's amro in the Śikṣāvaḷi. Amra is found in Tamil prints. Oṇte is similarly a common alternative for oṇṭo, in some publications. It seems that a replacement of -o of nominal bases by -e is one of the symptoms of dialectical varieties of Saurāshtran. Thus a popular Naṭana-Gōpāla-caritu uses such forms as siṣruse [sic] (obedience), parīkṣe (examination), poṭkke (the hood of a snake), Madurek khanvas pōkum (in the eastern quarter of Madura). The normal forms of these words are -o (nominative) -ā (inflected base). On the other hand there is a tendency for -o to replace -e in such past participles as hoyyo and rhiyo (for the normal hoyye, rhiye). There are, however, communities calling themselves Saurāṣṭra Brāhṃans whose language may prove to be quite distinct from that studied in the present article. S. M. Katre (The Formation of Koṅkaṇī, 1942, pp. 159–160) states that “the so-called Saurāṣṭra Brahmins of Āndhradeśa who are also found in parts of H.E.H. the Nizam's territory, and who pursue the goldsmith's profession, generally speak a dialect of Koṅkaṇi”.

^{page 326 note 2} Idiosyncrasies in the language (1880) of a writer of text-books, Hāḷivi Caturvīda Lakṣmaṇācārya of the Madras Free Church College, may derive from Oriyā, through Halbī, if his name indicates connection with Bastar. He uses gōṭa meaning “one”, and an occasional plural in -mānun or -māma, e.g. yugamānun as the plural of yuga. The locative in -r(u) might perhaps be due to Oriyā influence; it is not merely a local variation, however, but established in the standard speech. See also the preceding footnote. — One variation, regularly used by C. Lakṣmaṇācārya, and occasionally by other writers, is the use of kari as the “genitive postposition”.

^{page 327 note 1} In correction of Bloch's statement cited in the earlier part of this article (p. 114), Professor R. L. Turner points out that Sinhalese also has practically lost the ya- relative. Apparently it became unnecessary owing to the development of “relative” verbals (under Dravidian influence). For a similar development in Kōṅkaṇī see LSI., vii, p. 171. Bloch himself sums up the facts in L'Indo-aryen, p. 200: “seules les langues extérieures ont perdu le relatif: le groupe du Nord-Ouest, sauf le kaśmiri le remplace par l'interrogatif.” See the following footnote also.

^{page 327 note 2} Professor F. W. Thomas draws my attention to the occasional use of ka- as a relative in the Central-Asian Prākrit documents. See Burrow, T: The language of the Kharoṣṭhi documents from Chinese Turkestan (1937), sections 83–5.Google Scholar.

^{page 327 note 3} Rāo's, RāmaSaurāṣṭra-nandi-nighaṇṭu (1908, jn Saurashtran script)Google Scholar, corresponds in content with the first book of the Amara-kōśa, consisting of twelve chapters on the gods, space, time, music and drama, water, and so on. No more has been published, unfortunately. It is a valuable source for the vocabulary. Some hundreds of words are to be found in word-lists with Tamil, Telugu, and sometimes English meanings included in two or three publications, notably the first (1902) edition of Rāma Rāo's Saurāṣṭra-nīti-śambu. (This edition prints the text twice; first in Saurāshtran script with Saurāshtran explanations, then in Telugu script with Telugu explanations. Since it gives Tamil as well as Telugu meanings in its word-lists it must be supposed that literate Saurāshtrans' read both the Telugu and Tamil scripts and use both languages. There are some unusual words in Saurāshtran. What, for instance, is the source of the common word for the East, khaunas(u), khanvas(u)? The word is also given in the form khōnu. The Nandi-nighaṇṭu gives learned as well as common words; and it is not to be supposed that (for instance) the Greek word hēḷi, which is given among names for the Sun, was ever used outside an astronomical work. But the case is perhaps different with saul, another name for the sun given in the same passage. Sauḷi-vaṃśa is used of Daśaratha several times in the Saurāṣṭra-Rāmāyaṇu, apparently in the sense of “solar”. (The “elders”, the second class in the traditional organization of the Saurashtran community were called “Saulins”, but the meaning of this name is obscure. The word forms part of T. M. Rāmarāo's own name, Doppe Munisauḷi Rāmarāyi.)

^{page 327 note 4} Rāma Rāo, in the English preface to his Saurāṣṭra-vyākaraṇu, would derive it from Śaurasenī.