The markers of person, gender, and number in the prefixes of G-preformative conjugations in Semitic.

• Author: Hasselbach, Rebecca

As is well known, the preformative conjugations in the individual Semitic languages all operate according to the same basic principles in marking person, gender, and number. Person is usually indicated by a prefix that is attached to the verbal base, while gender and number are marked by the presence or absence of a suffix. The Classical Arabic verbal form yaktub "he shall write," for example, can be analyzed as [TEXT NOT REPRODUCIBLE IN ASCII], where ya- marks the 3rd person masculine and [TEXT NOT REPRODUCIBLE IN ASCII] the singular. In Arabic ya-ktub-u "they shall write" the same prefix ya- is used for indicating the third person, and final -u marks the (masculine) plural. In certain forms in which the plural prefix has a different form than the singular, as in the 1st persons, no special suffix is required to mark plurality; compare [TEXT NOT REPRODUCIBLE IN ASCII] "I shall write" with [TEXT NOT REPRODUCIBLE IN ASCII] "we shall write." Similarly, masculine and feminine only have to be distinguished by suffixes when the prefixes are ambiguous, as [TEXT NOT REPRODUCIBLE IN ASCII] "you (ms) shall write" versus taktub-i "you (fs) shall write." In this case the final -i in the feminine form explicitly distinguishes the 2fs from the 2ms, as opposed to the 3rd person singular forms [TEXT NOT REPRODUCIBLE IN ASCII] (ms) and [TEXT NOT REPRODUCIBLE IN ASCII] (fs), in which the use of two different prefixes, ya- and ta-, is sufficient to distinguish masculine and feminine.

Verbal prefixes usually consist of a consonant and a vowel. (1) While the consonant indicates the person and is consistent through all paradigms, the prefix vowel is dependent on verbal stem and voice. (2) The Proto-Semitic form of the prefix vowels is still a matter of dispute. In this article I will suggest a new reconstruction for these vowels in the G active stem based on their derivation from the respective independent pronouns.

1. EVIDENCE

   The quality of the prefix vowel of the active G-stem differs in each of the descendant languages. In West Semitic (WS) languages, prefix conjugations contain a homogeneous vowel, that is, the same vowel is used for all persons within the same paradigm. The following table provides an overview of the most important WS variations:

   Person Cl. Arabic Ge'ez Hebrew Syriac Ugaritic 3ms ya- y[partial derivative]- yi-/ya- ne- ya-/yi- 3fs ta- t[partial derivative]- ti-/ta- te- ta-/ti- 2ms ta- t[partial derivative]- ti-/ta- te- ta-/ti- 2fs ta- t[partial derivative]- ti-/ta- te- ta-/ti- 1cs 'a- '[partial derivative]- 'e-/'a- 'e- 'a-/'i- 3md ya- -- -- -- y/ta-/yi- 3fd ta- -- -- -- ta-/ti- 2md ta- -- -- -- ta-/ti- 2fd ta- -- -- -- ta-/ti- 3mp ya- y[partial derivative]- yi-/ya- ne- ta-/ti- 3fp ya- y[partial derivative]- ti-/ta- ne- ta-/ti- 2mp ta- t[partial derivative]- ti-/ta- te- ta-/ti- 2fp ta- t[partial derivative]- ti-/ta- te- ta-/ti- 1cp na- n[partial derivative]- ni-/na- ne- na-/ti- The prefix vowel in the Classical Arabic basic stem is uniformly [TEXT NOT REPRODUCIBLE IN ASCII]

   The consonants in Hebrew and Ugaritic differ from those of the other West Semitic languages in the 3fp, which has prefixed [TEXT NOT REPRODUCIBLE IN ASCII], as opposed to [TEXT NOT REPRODUCIBLE IN ASCII] in Arabic, Ge'ez, and Syriac. (5) Furthermore, Ugaritic differs in that it also has a t- prefix for the masculine plural. (6) This t-prefix for the 3mp, which is vocalized as [TEXT NOT REPRODUCIBLE IN ASCII] in syllabically written cuneiform documents, is also found in the canaanizing Amarna letters, (7) in Mari documents from the sakkanakku period, (8) and in Ebla. (9)

   It is possible to explain the different attestations for the prefix consonant of the 3mp as a result of leveling. On the basis of Arabic it has often been assumed that the original form of the 3mp prefix was *y-. (10) For the 3rd feminine plural we find t- in Hebrew, Ugaritic, and the Amarna letters. If we assume that the distribution in Hebrew is original, that is, *y- for the 3mp and *t- for the 3fp, all other distributions can easily be explained by the leveling of one of the prefix consonants. In those cases in which we find t- for the 3mp, the feminine form was leveled, supported by the 2p, which also has the prefix t-. (11) In those cases where we have y- for both the 3mp and 3fp the masculine form became dominant.

   In Akkadian, the prefix vowels differ significantly from the WS system described above. Here, the prefix vowels of the G- and N-stems depend on person and number; the 3rd persons and the 1cp in Babylonian have [TEXT NOT REPRODUCIBLE IN ASCII], the 2nd persons and the 1cs have [TEXT NOT REPRODUCIBLE IN ASCII] as prefix vowel:

   3ms i- 3mp i- 3fs i-/ta- (12) 3fp i- 2ms ta- 2cp ta- 2fs ta- 1cs a- 1cp ni- Striking exceptions to this East Semitic pattern are the already mentioned ti- prefixes for the 3mp in Ebla and the occurrence of the prefix ti- in an Old Akkadian incantation from Kish. (13) The perplexing feature of these attestations is not so much the consonant, but the [TEXT NOT REPRODUCIBLE IN ASCII] vowel, because in Ebla and in the Old Akkadian text this prefix alternates with the 3fs prefix ta-. (14) Since Ebla and Kish have been proven to share certain linguistic and cultural features, (15) and since the prefix ti- is otherwise unknown in Old Akkadian, the attestations in Ebla and Old Akkadian can be considered the result of a single phenomenon. We therefore have to assume an origin outside the classical Old Akkadian writing for the attestation in MAD 5:8.

   It has been suggested that these prefixes might be the result of West Semitic influence, but such an influence cannot be proven with certainty. (16) Suggesting another explanation for the attestations of ti- in the Amarna letters, Izre'el proposes that the consonant t- might be a reflection of the substrate language and the vowel [TEXT NOT REPRODUCIBLE IN ASCII] the common vowel of the 3mp prefix of Akkadian. (17) Although it may not be necessary to assume a distinct substrate language in the case of Ebla to account for the [TEXT NOT REPRODUCIBLE IN ASCII], the suggestion that the [TEXT NOT REPRODUCIBLE IN ASCII] vowel in the 3mp prefixes is a result of an analogy with the regular vowel [TEXT NOT REPRODUCIBLE IN ASCII] in other 3rd masculine prefixes, that is, the singular and the dual, is quite appealing. This kind of analogy can explain the vowel alternation found between the 3mp prefix ti- and the 3fs prefix ta- in Ebla, Old Akkadian, and Mari. (18)

   In general, the vowels of the prefixes of the basic stem in Semitic languages reflect two main vowel qualities, *i and *a. The vowel [TEXT NOT REPRODUCIBLE IN ASCII] underlies the prefixes in Ge'ez, Syriac, and most roots in Hebrew, while [TEXT NOT REPRODUCIBLE IN ASCII] is uniformly found in Arabic and in some weak roots in Hebrew. Only Akkadian differs in that it regularly has both vowels within one paradigm, depending on person. Thus we find two different systems regarding the distribution of prefix vowels in Semitic, a homogeneous WS system and a heterogeneous Akkadian system.

2. PREVIOUS RECONSTRUCTIONS OF THE PREFIX VOWELS

   Several theories for a Proto-Semitic reconstruction of the prefix vowels have been suggested. One common view is that Arabic reflects the original situation, that is, the PS vowel of G active prefixes was [TEXT NOT REPRODUCIBLE IN ASCII]. (19) Scholars who support this theory assume that the 3ms prefix i- in Akkadian reflects the sound change *ya- > *yi- and that an analogy of the 1cp with the [TEXT NOT REPRODUCIBLE IN ASCII] of the 3ms is responsible for the 1cp prefix ni-. (20) The sound change *ya > *yi /#_ is well attested in Akkadian, as in *yasarum > *yisarum > isarum "just." (21) Why the assumed analogy only took place between the 3ms and 1cp and did not spread to the 2nd person forms (in Assyrian also the 3fs) remains unexplained.

   The derivation of all prefix vowels from original *a is not only problematic in Akkadian. It has already been noted by Barth (1894) that a simple derivation from *a is not possible because of Ethiopian; as previously mentioned, the [TEXT NOT REPRODUCIBLE IN ASCII] of the prefixes in Ge'ez cannot be derived from [TEXT NOT REPRODUCIBLE IN ASCII], but only from either *i or *u:

   Dass beide Arten aus einem und demselben fur alle Personen durchgehenden ursemit. Grundtypus a stammen sollen, hat schon wegen der erwahnten Schwierigkeit des Aeth. seine Bedenken. (22) Basing his argument on evidence from I-guttural verbs in Hebrew and I-aleph verbs in Aramaic, (23) Barth suggested, in what has become known as Barth's Law, that the vowel of the prefix was originally dependent on the theme vowel of the imperfect. Imperfects with the theme vowel [TEXT NOT REPRODUCIBLE IN ASCII] and [TEXT NOT REPRODUCIBLE IN ASCII] had the prefix Ca-, while imperfects in [TEXT NOT REPRODUCIBLE IN ASCII] had the prefix Ci-. (24) In Hebrew these verbs still show Barth's distinction, as in [TEXT NOT REPRODUCIBLE IN ASCII]

   The existence of Barth's Law in West Semitic has been further confirmed in numerous studies. In 1939, Ginsberg showed that the distribution of prefix vowels suggested by Barth existed in Ugaritic. (27) Bloch argued that Barth's Law was operative in Eastern Arabic dialects around the eighth century A.D., basing his argument on the description of these dialects by the Arab grammarian Sibawaih. (28) Reflexes of Barth's Law can still be found in some Modern Arabic dialects, such as Modern Syrian, where the prefix vowel is [TEXT NOT REPRODUCIBLE IN ASCII]

   An interesting situation was discovered by Schub in his investigation of the dialect of the Tamim tribe, which seems to reflect an intermediate stage between Barth's Law and later leveling. In this dialect the prefix is usually pronounced with the vowel [TEXT NOT REPRODUCIBLE IN ASCII] before the theme vowel [TEXT NOT REPRODUCIBLE IN ASCII], but with [TEXT NOT REPRODUCIBLE IN ASCII] before the theme vowels [TEXT NOT REPRODUCIBLE IN ASCII] and...