¹ Die Sprachen der Hamiten (Abh. des Hamburgischen Kolonialinstituts) 9. Strictly speaking, Meinhof does not seem to have been the first to use polarity in a linguistic sense. It is clear, however, from his elaborate statement that he was unaware of any other secondary application of this term. Moreover, Meinhof suspects the operation of polarity in phonetic and morphological developments, whereas others would use the term in connection with (Greek) syntax and stylistics. Since the advocates of polarity in the Hamito-Semitic field, with which the present discussion is concerned, are all avowed followers of Meinhof, the term will be used here in the sense in which it was independently adopted by him.

² Ibid. 19.

³ drastisch ausgedrückt'.

⁴ For numerous unexplained exceptions in Ful see ibid. 41; in Šilh feminines take ‘not infrequently’ masculine plurals (ibid. 95); in Somali only collectives to masculine singulars are feminine (ibid. 171).

⁵ Ibid. 19-20.

⁶ For a convenient summary of the affinities of Egyptian see A. H. Gardiner, Egyptian Grammar 2-3 (1927).

⁷ H. Brunner, Das Gesetz der Polarität in der ägyptischen Sprache, Zeitschrift für ägypt. Spr. 72. 139-41 (1936).

⁸ Ibid. 140.

⁹ Prof. R. G. Kent has called my attention to the fact that the term (in Meinhof's sense) has been making headway also in Indo-European linguistics.

¹⁰ The principal advocates of the theory have been, in addition to Meinhof, W. H. Worrell and G. Bergsträsser. The three sets of examples adduced by Bergsträsser will be discussed below under sections 2-4. Worrell's principal argument is detailed under section 1; his remaining illustrations are covered, for the sake of completeness, in footnote 30. I first expressed my misgivings about Worrell's position in JAOS 49. 181 (1929).

¹¹ The Formation of Arabic Broken Plurals, Am. Jour. Se. Lang. 41. 179-82 (1925); A Study of Races in the Ancient Near East 61-4 (1927).

¹² W. Wright, Comparative Grammar of the Semitic Languages 148 (1890); Grammar of the Arabic Language³ 1. 199-234 (1896). A thoroughly satisfactory treatment of the nature of these forms was given by H. Barth, Nominalbildung 417-83 (1889).

¹³ Races 64.

¹⁴ It is based on infinitives of the type qitāl, cf. Wright, Arabic Grammar 1. 111 no. 26; 202-5.

¹⁵ See Speiser, Studies in Semitic Formatives, JAOS 56. 33-46 (1936).

¹⁶ Although by no means as rare as is generally supposed; for a recent list of collectives in Hebrew see L. H. Gray, Introduction to Semitic Comparative Linguistics 53 (1934). These forms are represented frequently by ‘segholates’ of the type qitl and qutl. A good, but unnoticed, example is kofěr ‘villages’ (type qutl), I Samuel 6.18. An older instance is sefěr ‘writing, book’ (type qitl), which Hebrew borrowed from Akkadian.

¹⁷ Here belong most of the singulars in *-at which form their plurals with -êm; for examples of such forms see H. Bauer and P. Leander, Historische Grammatik der hebräischen Sprache 1.515-6. The segholate plural q∂tāl-'êm is most likely of the same origin.

¹⁸ Thus a modern Arabic term for ‘Egyptologists’ is barāgiš, which is based on the name of the German Egyptologist Brugsch; see Meinhof, Die Entstehung flektierender Sprachen 55 (1936). I have heard many formations of the type panākik < Engl. pancake.

¹⁹ This is the number of the more common forms cited by Wright, Arabic Grammar 1. 199.

²⁰ Ibid. 233.

²¹ Ibid. 110-2, the list being only 'nearly complete'.

²² Ibid. 202-5.

²³ Ibid. 111 no. 26.

²⁴ See note 17.

²⁵ Races 62-3.

²⁶ Ibid. 63.

²⁷ Wright, Arabic Grammar 1. 225.

²⁸ Ibid. 225-6.

²⁹ Barth's statement on the subject, published nearly fifty years ago (Nominal-bildung 417-31), is still substantially correct.

³⁰ Worrell cites two other positive instances of polarity. One is the use of 'the inferiority sign t' to distinguish ‘the relatively unimportant second person and feminine from the first person and masculine respectively’ (AJSL 41.182; Races 64). The fact is that the t which is used as a pronominal element in the second person of the verb represents the personal pronoun 'an-tā/tī ‘thou’, while the t in the feminine forms qatal-at, ta-qtul, and the like, is based on the feminine ending -at. The respective formatives for person and gender are absolutely heterogeneous. The other instance concerns the use of banāt ‘daughters’ in plurals of compounds with ibn ‘son’ when these compounds are used for names of animals and other objects (Races 63-4), but not of persons (Wright, Arabic Grammar 1.196). Thus ibnu dā'irin ‘son of the lewd', ’young stallion camel’ appears in the plural as banātu dā'irin, without any regard for the actual sex of the object. But in this case banāt- is not properly a feminine plural; it is an original singular collective form of *bin ‘son’, viz., *binat- whose present form is due to the analogy of banāt- ‘daughters’. Such a use of the collective ’sons’ would have its exact parallel in Arabic 'iḫwat (sing.) ‘brothers’ and Hebrew ’āḇôṯ ‘fathers’, for which see below.

³¹ G. Bergsträsser, Einführung in die semitischen Sprachen 15 (1928); Hebbräische Grammatik, Verbum 5 (1929); Meinhof, Enstehung flekt. Sprachen 75; Bauer and Leander, Hebr. Gramm. 1. 515, explain the ending of 'ābôṯ as due to the analogy of *'immôṯ ‘mothers’.

³² See note 17, and Bauer and Leander, Hebr. Gramm. 1.515-6.

³³ Speiser, JAOS 56. 41 and the references there given.

³⁴ The exact character of the final vowel is unknown since we find it only in contractions with case-endings. These contractions in the masculine forms result in Hebrew in ê (from the genitive), in Aramaic in â (from the accusative). Arabic lost the final vowel of these stems ('ab, 'aḫ). However, various derivative forms with -w- favor an original stem-ending .

³⁵ Josh. 2.13; Ezek. 16.52; Job 42.11; I Chron. 2.16.

³⁶ JAOS 56.40.

³⁷ Arabic does precisely the same thing with 'aḫ ‘brother’, whose abstract form is specialized for the plural: 'iḫwat ‘brothers’; cf. C. Brockelmann, Grundriss der vergleichenden Grammatik der semitischen Sprachen 1.427 (1908). In this case, however, the plural ending could not be suspected, as it has been in the parallel Hebrew form 'āḇôṯ, because the stem-ending of the Arabic word assumed consonantal function, thus preventing contraction.

³⁸ See Speiser, Annual of the American Schools of Oriental Research 10.11, The semantic changes exhibited by the respective abstract forms of the two kinship-terms under discussion are illustrated in the following schematic grouping:

Basic Form Abstract Form

See also JAOS 56.40.

³⁹ Cf. Sapir, Language 197-203 (1921).

⁴⁰ For the latest literature see Bergsträsser, Verbum 9-10, and Driver, Problems of the Hebr. Verb. Syst. 155-7.

⁴¹ Verbum 10-4, 89, 92.

⁴² Ibid. 6.

⁴³ Cf. e.g., Driver, Problems 17-20 and Bergsträsser, Verbum 10-4.

⁴⁴ Problems 72-3.

⁴⁵ I can see no linguistic advantage in statements to the effect that a given form was vocalized 'with the soft or light i to express stativity or passivity and the hard or heavy a to express activity'; see ibid. 72.

⁴⁶ It is not necessary to contribute here to the rather academic speculation as to which particular verbal form (imperative, preterite, infinitive, each of which has its supporters) was the earliest in the language. Nor does it matter whether verbal and nominal forms existed side by side from the beginning, or not. For our present purposes it will suffice to go back to that late, although still prehistoric, period when the two basic sentence-types were fully established.

⁴⁷ For the terminology see L. Bloomfield, Language chap. 11 (1933).

⁴⁸ Logical considerations compel the use of the second-person forms as paradigmatic in historical discussions, instead of the third-person forms, as is commonly done. In both types of Semitic conjugation (preformative and afformative) third-person forms are demonstrably later than the other two. The same pronominal elements underly in both conjugations the forms for the first and second persons respectively, whereas the third person employs in the preformative conjugation a dissimilar pronominal element, and in the afformative conjugation no pronoun at all. Moreover, in the latter conjugation there are accentual differences between the first and second persons on the one hand, and the third person on the other (cf. Arab. qatál-tu, qatál-ta, but qátala), which are the indirect result of the disparate origin of the forms in question. Finally, the second person influenced in West Semitic the consonantal afformative of the first (t instead of k). To start with *qátala rather than *qatál-tā would, therefore, be anachronistic and generally disadvantageous.

⁴⁹ Again it would be futile to speculate how and why the vowel-sequence a-i/u came to prevail. The fact is that it did.

⁵⁰ To use ‘stative’ as a term descriptive of 'Aktionsarten’; for the contrast between objective and subjective verbal systems see Bergsträsser, Einführung 11 (based on Landsberger, Die Eigenbegrifflichkeit der babylonischen Welt, Islamica 2.355-72[1923]).

⁵¹ For the position of the dialect of Ras Shamra see A. Goetze, The Tenses of Ugaritic. JAOS 58.266-309.

⁵² The possibility of such a reversal shook Bergsträsser's confidence in polarity; cf. Verbum 13. See also F. R. Blake, The Internal Passive in Semitic, JAOS 22. 45-54 (1901), and A. Ungnad, Zeitschr. d. deutsch. morg. Ges. 59.766-8 (1905). The part which sentence-types may have played in these changes has not been considered thus far.

⁵³ Cf. Wright, Arabic Grammar 1.30; Driver, Problems 12; see also F. R. Blake, The so-called Intransitive Verbal Stems in Hebrew, JAOS 24.145-204 (1903).

⁵⁴ It is perhaps pertinent to stress once again the fact that according to the above account the origin of the second a of qatál-tā, the chief difficulty in previous discussions, was determined automatically. The same principle which led to fí-qtal- and tú-qtal- in the neuter forms was responsible for qatál-tā in the active class.

⁵⁵ Cf. Bergsträsser, Einführung 15; Driver, Problems 72.

⁵⁶ See JAOS 56.39.

⁵⁷ The priority of the (‘feminine’) forms in -at was not disputed by the older grammarians; see the summary in Sven Herner, Syntax der Zahlwörter im Alten Testament 4-12 (1893). A different view is taken by Reckendorf and approved by Brockelmann, Grundriss 2.274-5.

⁵⁸ ‘Two’ is also an adjective, but it frequently follows the analogy of ‘three’ to 'ten’ in behaving like a noun that governs a genitive; cf. Brockelmann 273.

⁵⁹ This is still often the case; cf. Herner, op. cit.

⁶⁰ See Speiser, JAOS 56.37-41.