¹ Vienna, 1905 (SBAW Wien, Phil.-hist. Kl., CL, 6). The reference is to section (i), p. 73.

² ‘Four strange tongues from South Arabia: the Hadara group’, Proceedings of the British Academy, XXIII, 1937, 231–331Google Scholar. The reference is to p. 243 of the Proceedings and to p. 17 of the separate offprint.

³ On the first occasion in a paper ‘Non-Arabic place names in central South Arabia’ given at the 24th International Congress of Orientalists, Munich, 1957. I have been lent a corrected printed version of the text published in the Akten (Wiesbaden, 1959)Google Scholar. A longer and revised paper was given at the 25th International Congress of Orientalists in Moscow, 1960, entitled ‘Again on non-Arabic place names in central Southern Arabia’. This appears in vol. i of the ‘Proceedings’ of this Congress published in Moscow in 1962 on pp. 548–55. However, I am working from a copy amended by the author which is on loan to me. Matthews also mentions this point briefly in his El article.

His later publications on this subject include ‘On the border of the sands. Third installment. Part 11, 1. Noun determination in South Arabian’, University of South Florida Language Quarterly, VII, 1–2, 1968, 41–8Google Scholar, and ‘Modern South Arabian determination—a clue thereto from Shahri’, Journal of the American Oriental Society, LXXXIX, 1, 1969, 22–7Google Scholar. These more recent papers were not available to me at the time when this article was first written, but one interesting suggestion in the last is discussed below, p. 305, n. 37. To the best of my judgement, however, Matthews does nothing in these later two papers to demonstrate the function of the prefixes he is discussing.

⁴ Thus for example the belief that ‘Suqutrī…is basically Mahrī’ leads him to suggest an etymology for the name of the Haggier (mountains) in Socotra which is quite wrong. The Socotri term is ɧájhεr or ɧághεr, which is cognate with the Arabic and is not Jabal ha-G^ehēr ‘The weak-sighted mountain’, or any of the amendments suggested. Similarly the rejection of the name ‘Skhawri’ is only partly correct. One of the names for this language is Śḥeri (ǰɧέri) deriving from the Śḥeri word əǰɧέr)Er (əǰɧayr, əǰɧár), the usual word for ‘mountains, highlands’, with which may be compared the (Nejdi) Mehri ǰɧayr with the same meaning.

⁵ From October 1968 to March 1969 on a study leave from SOAS. My thanks are due to the Sultan of Muscat and Oman for permitting me to work on Harsiisi during most of this time, and to Petroleum Development (Oman) Ltd. for their generous permission to use their facilities. The examples are from the speech of Ḥāmid b. Musallam al-Ḥarsūsi who gave me much of my material on his language and for whose patience and help I am grateful.

⁶ The system of transcription is based on that of the IPA but the following may be specifically noted. Є, h voiced and voiceless pharyngals; ? glottal stop; ǯ,ǰ voiced and voiceless laterals; t', k', s' voiceless ejectives; 8 is a not wholly voiced dental fricative in Mehri and Ḥarsūsi, enunciated with the blade of the tongue on the upper teeth, whereas 5 is enunciated with the point of the tongue on the upper teeth. It can tentatively be classed with the ejectives. In Śḥeri it is usually unvoiced and ejective [θ'].

The short vowel a represents a centralized variety [ä], and ə a centralized variety of e [ë] as well as an anaptyctio [ə].

A long vowel is indicated by doubling (as aa, ee, etc.), and prominence by an acute accent (as -áa, -ée, etc.). The vowel of a final open prominent syllable (as -ú, -ó) is half-long to long.

A characteristic feature of Śḥeri is that all syllables, the vowel of which is not anaptyctic, tend to be equally prominent, the vowel of an open syllable being half-long to long, as for example έmέs ['ε'mε's] ‘her mother’.

⁷ Miiller, D. H. (ed.), Die Mehri- und Soqoṭri-Sprache, iii, Šhauri-Texte (Kaiserliche Akademie der Wissenschaften [in Wien]. Südarabische Expedition, vii), 1907, 40Google Scholar. Note the occurrence of the 3 f. dual in Ḥarsüsi. On 1 c. dual forms cf. p. 302, n. 33.

⁸ The speaker compared the Arabic ‘to get possession of’.

⁹ The drawing of lots refers to the custom followed by Bedouin so that food is divided fairly. In the form ərəwáat, the initial vowel is almost certainly a prosthetic vowel. These occur regularly in many interconsonantal positions, less regularly at word juncture. On the final word aráwah, compare the discussion of status pronominalis on p. 304.

¹⁰ Compare also example 13 below.

¹¹ On the prosthetic vowel of əlá cf. p. 297, n. 9. The prosthetic vowel occurs so regularly before la as to be found even in postvocalic positions.

¹² This dialect is that of Nejd in Dhofar and represents a dialect of a type more conservative in a number of respects than those studied by the Sadarabische Expedition. Mehris from this area are often encountered in the Giddat al-Ḥarāsīs, but this work was done mostly in Dubai with a Mehri speaker called 'īlī b. Musallam.

¹³ Parts of the body can have unaffixed forms in Mehri, but not apparently in Ḥarsūsi, namely iid, ʡʡ⋯éen, uwbíib, and ru(h). Usage seems to be confined to partitive constructions, but the incidence of occurrence is very low.

¹⁴ If this is correctly to be compared with Heb. . Bittner, however, compares (Studien zur Laut- und Formenlehre der Mehri-Sprache, IV (SBAW Wien, Phil.-hist. Kl., CLXXIV, 4), 1914, 58)Google Scholar.

¹⁵ Ḥ yiim, M yuum ‘days’ (in enumeration).

¹⁶ Viz. 1 >Ḥ> W. This is common in this dialect. Cf. Thomas, art. cit., s.v. ‘heart’ (>owbīb).

¹⁷ Jahn, A., Die Mehri-Sprache in Südarabien (Kaiserliche Akademie der Wissenschaften [in Wien]. Südarabische Expedition, III), 1902, 161–242Google Scholar.

¹⁸ Probably to be identified with the root īb, the –uu(–) of this form being a collective termination. If this is so (cf. Soc. ʔəʔəfo ‘parents, people’, Śḥeri ⋯Υ⋯ ‘people’), then the Mehri form without a final b is the more conservative. Cf. Leslau, , Lexique soqoṭri, Paris, 1938, 69Google Scholar.

¹⁹ Bittner, M., Studien zur Šḫauri-Sprache, I (SBAW Wien, Phil.-hist, Kl., CLXXIX, 2), 1916, 15Google Scholar. On the root cf. Soc. ʔ⋯εrər ‘to light’, Akk. urru as against √ʔwr in Heb., Ugar., etc.

²⁰ Ugar. irby, Heb. , Akk. a/eribu. Bauer-Leander (Historischie Gr. der hebräischen Sprache (repr.), Hildesheim, 1965, 7) consider this item as, amongst others, typical of North and East Semitic, though Brown-Driver-Briggs quotes the Mehri s.v. There are in fact a number of items where MSA correlates with North Semitic rather than with Arabic.

²¹ Usually compared with ‘a road marker’.

²² cf. p. 302, n. 30.

²³ cf. Leslau, , Lexique, 75Google Scholar.

²⁴ The root would seem to be ʔgr > wgr. Cf. Landberg, , Glossaire daṱdῖnois, i, Leiden, 1920, 361Google Scholar.

²⁵ The root of уəbiit/beer, bуaar would seem to be ʔbl. On the final t of. ESA 'bit. The pl. may be related to bƹr > bʔr but cf. Śḥeri yit/⋯y⋯l.

²⁶ Śḥeri wdín.

²⁷ There are a few nouns with prefixed yə– which apparently is not detachable, as e.g. M yəriiz (Ś írəz ) ‘rice’, M yəɣaráyb (Ś iɣaréb) ‘crow’. Whether this prefix is comparable, however, is not established.

²⁸ It may be of interest, however, to consider the possibility that certain of these nouns, apparently not with initial ʔ, w, y, do have such initial consonants. Thus cf. for example Ś ⋯n⋯f ‘self’.

²⁹ The Śḥeri information was collected during a number of sessions with Dhofari policemen and soldiers in Trucial Oman. This work showed that there is more than one dialect in Śḥeri, and it seems possible that the prefix i– is characteristic of one dialect type, and ε– of another.

³⁰ Since the Socotri form has initial ʔ (< W ?) and the Mehri initial W cannot derive directly (on comparative grounds) from initial Ƹ, Śḥeri is the only language of the group to have (preserved the) initial Ƹ. It may be doubted therefore whether the usual comparison with is the correct one for all of these forms unless the change Ƹ < ʔ took place at an early date in M and S.

³¹ Viz. ~ Ar. ; cf. also M smii, Ḥ səm⋯.

³² Examples in which ə– could be interpreted as an anaptyctic at form juncture have been avoided as far as possible. No affixed forms have been noted of lexical items with initial x, ɧ, or ʔ < ƹ (cf. n. 30, above), though affixation can occur where the initial consonant is ɣ.

³³ The fact that these are 1 c. dual forms may be noted. The existence of such forms was postulated for Ḥarsūsi by E. Wagner, on the very scanty evidence given by Thomas, , in his ‘Die erste Person Dualis im Semitischen’, ZDMG, CII, 2, 1952, 231Google Scholar.

³⁴ cf. Müller, , op. cit., p. 36Google Scholar, paras. 11–12.

³⁵ It seems to me at the time of writing that the gemination in the final word of this is a function of stress. Cf., however, Ullendorff, E., ‘The form of the definite article in Arabic other Semitic languages’, in Makdisi, G. (ed.), Arabic and Islamic studies in honor of Hamilton A. B. Gibb, Leiden, 1965, 636Google Scholar. Compare also Ḥ tuwi, әt^təwi ‘meat’.

³⁶ baƹíir is the Ar. equivalent of Ḥ bʔayr. Ar. words occur freely in Ḥ, as do Ar. pronunciations of certain consonants, as e.g. garaʃ, k'araʃ in example 32.

³⁷ That a form which has initial nasalization implies prefixation of ε– was pointed out by Matthews, , JAOS, LXXXIX, 1, 1969, 22–7Google Scholar. Thus m⋯sgíd ‘a mosque’, but (ε + m⋯sgíd) ἕsgíd ‘the mosque’, músmər ‘a nail’, ṓsmər ‘the nail’, etc. It may be added that a comparable phenomenon is observable in but ‘a house’ and (ε + but) [w]út ‘the house’, etc. Certain other categorical statements in this article are incorrect, however. Thus is a palatalized allophone of g, and ƹ occurs freely in Śḥeri (though it may be replaced by pharyngalization of the vowel). It may be added of the in examples 53 and 54 that it is an allophone of 1, occurring in the contiguity of i.

³⁸ Viz. the final consonant is a retroflex d.

³⁸ In status pronominalis such items do not have nunation, thus ɧadúutiyε ‘my hands’.

⁴⁰ Although, as in the case of mimation in ESA, nunation in MSA does not ‘per se correspond the English indefinite article a/an’ (cf. Beeston, A. F. L., A descriptive grammar of Epigraphic South Arabian, London, 1962, 30)Google Scholar.

⁴¹ Ḥarsūsi and Nejdi Mehri are clearly more conservative in certain respects than these dialects. Thus for example Ḥarsūsi has conserved dual forms of the verb and they are freely used, though plural forms may be used in their place. In Nejdi Mehri almost as many dual forms occur, but plural forms seem to be preferred in ordinary speech. Comparable forms do not occur in the Mehri dialects of the Südarabische Expedition texts, as e.g. Bittner, , Studien zur Šḫauri Sprache, II (SBAW Wien, Phil.-hist. KJ., CLXXIX, 4), 1916, 14Google Scholar, ‘(als) das Mehri im Bereiche des Zeitwortes keinen Dual besitzt’.