Verbal Substantives in Proto-Semitic
Introduction
This blog post discusses three related verbal substantive patterns in Proto-Semitic (PS) which were associated with passive-resultative, stative-resultative, and active semantics of the verbal root. The central claim is that the quality of the vowel (either /ā/, /ī/, or /ū/) in the verbal substantive pattern C1aC2V:C3– correlates with its active/stative/resultative/passive semantics, which in turn developed into various types of verbal nouns (e.g. participles, infinitives, agent nouns) in the daughter languages. It also addresses the morphological and semantic relationship between this pattern with the common qaṭṭāl-, qaṭṭīl-, and qāṭil– nominal patterns.
Active, Passive, Resultative, and Stative in Semantics
Before jumping into the PS forms, it is important to clarify the semantic relationship among active, passive, stative, and resultative types of verbs and verbal nouns. While the active and passive categories primarily concern whether the agent or patient is foregrounded, the stative and resultative categories concern the relationship between an action and its resulting state. Specifically, while a resultative denotes a state that arises from a prior event, a stative does not necessarily invoke or imply that a prior event occurred. In some respect, therefore, the resultative functions as a bridge between passive events and states (i.e. passive → resultative → stative). In English, this is the difference between “the door is/was opened (by someone)” (passive or resultative, depending on context) and “the door is/was open” (resultative [because someone opened it] or stative [because it has always been open], depending on context).
Passive-Resultative and Resultative-Stative Verbal Substantives in Proto-Semitic: qaṭūl- and qaṭīl-
The common PS pattern for verbal substantives was C1aC2V:C3-, where the quality of long vowel (either /ā/, /ī/, or /ū/) determined, at least in part, its semantic tendancies. Although not particularly common, Akkadian words like zaḳīpu ‘stake’ (i.e., ‘an impaled thing’ < zaḳāpu ‘to fix upright’) and baūlātu ‘subjects’ (i.e., ‘ruled ones’ < bêlu < *baˁālu) suggest that all three patterns were used in PS. In Gə‘əz, the form deriving from /ū/ became a passive verbal adjective for transitive verbs, e.g. ብቱክ bətuk ‘broken’, ፍቁር fəḳur ‘beloved’, while the form from /ī/ became the infinitive, i.e. ቀቲል/ቀቲሎት qatil(ot), and, when combined with possessive pronominal suffix, a converb, ቀቲሎ qatilo ‘after his killing…’.
In Northwest Semitic, forms historically derived from C1aC2i:C3 (= qaṭīl-) and C1aC2u:C3 (= qaṭūl-) came to function as passive/resultative/stative participles in languages like Hebrew (e.g. נָבִיא nāḇī ‘prophet’ < ‘called one’; בָּנוּי bānūy ‘built’) and Aramaic (e.g. כְּתִיב kəṯīḇ ‘written’, סְנוּאֲתָא < sənūˀ ‘hated’ [TgO Gen 29:31]). Ugaritic also attests both. Pairs such as Hebrew בָּרוּךְ bārūḵ corresponding to Aramaic בְּרִיךְ bərīḵ show that, for the most part, the form with /ū/ became the productive passive participle in Hebrew, whereas the form with /ī/ became the productive one in Aramaic. The alternative forms (i.e. with /ī/ in Hebrew and /ū/ in Aramaic) are inherited lexicalized nouns, especially common for qāṭīl-nouns in Hebrew (מָשִׁיחַ māšīaḥ ‘anointed one’; נָשִׁיא nāśī’ ‘chieftain’; נָגִיד nāḡīḏ ‘ruler’; נָבִיא nāḇīˀ ‘prophet’; פָּלִיט pālīṭ ‘fugitive’; etc.). (For a helpful blog post on the Northwest Semitic forms, see here.)
The overlap in passive, resultative, and stative semantics discussed in §2 above helps explain why a language would simplify to a single morphological form to express both categories, especially in the case of a language like Aramaic with many L2 speakers in antiquity. Perhaps to be taken as further evidence for /i/ as the original vowel for stative semantics, one can consider also the form of the D-stirp stative adjective qaṭṭīl in languages like Aramaic (e.g. תקיף taqqīp̄ ‘strong’ < ‘one who is regularly strong’) as well as the stative of the conjugated verbal adjective (> West Semitic suffix conjugation), as in Hebrew כָּבֵד kāḇēḏ < *kabida ‘to be heavy’. The /i/ of the (stative) verbal adjectives in Akkadian (< *paris, e.g. damqum < *damiqum ‘good’) and in Gə‘əz (ልሂቅ ləhiḳ ‘old’, ርሒብ rəḥib ‘broad, wise’) are also possibly related.
The Active Verbal Substantive: qaṭāl- and Its qaṭṭāl- Derivative
The third pattern for verbal substantives of this type, namely C1aC2a:C3 (qaṭāl-) with */ā/, however, has a different outcome. If the qaṭūl- and qaṭīl– patterns originally marked passive/resultative and resultative/stative semantics, respectively, the qaṭāl-pattern should, we expect, primarily be used for active transitive verbs and develop into active participles. In fact, this /a-ā/ (active verbal substantive) vowel pattern does, when combined with the pluractional semantics of D-stem gemination, create an agent noun (i.e. qaṭṭāl-) used mainly to describe human professions and habitual characteristics, as in Akkadian dayyānum ‘judge’, Aramaic אַשָּׁף ‘conjurer’ and Hebrew גִּבּוֹר gibbōr ‘warrior’ (after attenuation of /i/ < /a/ and the Canaanite shift [contra the *quṭṭulum reconstruction by some scholars; cp. e.g. Syr. gabbāra]). These lexicalized nouns tend to mark professional occupations in society. Of course, this pattern is closely related to the so-called nomen professionis pattern, again related to D-stem pluractionality for one who regularly performs such actions (e.g. Hebrew טַבָּח ‘butcher, cook’).
Nevertheless, since, however, another more common noun pattern already existed for active verbal nouns, namely C1a:C2íC3– (qāṭíl-, which was inherited as the active participle in most languages), the qaṭāl– pattern filled another syntactic role: the infinitive and gerund. On the one hand, the reflex of the latter form appears as an infinitive in Akkadian parāsu(m), Ugaritic and Hebrew qatālu, e.g. Hebrew אָכֹל ˀāḵōl (after pretonic lengthening and the Canaanite shift). On the other hand, it functions as a verbal noun in Arabic (lexicalized سلام salām ‘well-being’) and an agent noun pattern in Southwest Semitic (Gə‘əz qatāli; cp. e.g. Tigrinya ከፋቲ käfati ‘opener’).[1] On the cline of nominality (namely, the spectrum in which a verbal substantive acts more like a noun or a verb), we have three outcomes: (1) the Akkadian infinitive displays both strong nominal and verbal properties, depending on syntax and context. Huehnergard (1997: 337) lists the following reasons for its nominal properties: (a) they occur in any nominal case, (b) they take adverbial endings such as -iš, (c) the occur in the bound form before a genitive noun or pronoun, (d) they can be modified by an adjective, and (e) they can be modified by ša plus a genitive or relative clause. (2) Hebrew treats qaṭāl– words towards the verbal end of the spectrum (although note also nouns like שָׁלוֹם < *salām) (3) Arabic and Southwest Semitic take them as common or agent nouns.
[1] By Southwest Semitic, I simply seek a politically neutral term for the Semitic languages of the Horn of Africa, what are usually called Ethiosemitic (or Ethiopian Semitic). I do not intend to connote any subbranch of West Semitic which includes Arabic and Ethiosemitic, often referred to as “South Semitic” by some scholars. To be clear, Arabic is Central Semitic, more closely related to Northwest Semitic than “Southwest Semitic” as used here.
qāṭil- Active Participles and qaṭāl- Active Infinitives in Northwest Semitic
Based on their usage in Northwest Semitic, the qāṭil– and qaṭāl– overlapped on the nominal-verbal continuum. On the spectrum, the qāṭíl– form was preferred for verbal substantives (with emphasis on its nominal properties, i.e. “one who does X”), whereas the qaṭāl– form was used for verbal substantives (with emphasis on the verbal properties, i.e. “the doing of X”). This was apparently the preferred situation already in PS: qāṭil– was a substantive participle (as evidenced by similar use in Akkadian), which in turn relegated the use of qaṭāl– for other types of verbal nouns (infinitives, agent nouns, etc.). Although a much later development, it is important to note that in the history of Aramaic, the qāṭíl– (active participle) form swung back to the verbal end of the spectrum, becoming a finite (generally present-tense) verb, which affected later Hebrew as well.
Finally, it is worth noting also that Hebrew has two infinitives—the so-called infinitive absolute (šāmōr) and an infinitive construct (šəmōr). The origins of the latter remain a point of contention. These are traditionally understood to derive from a separate noun pattern, *qutul (Bauer; GKC; Moscati; J. Fox; etc.), with reduction of the pre-tonic /u/ and the raising of thematic /u/ > /o/. An alternative view treats the these as also deriving from *qaṭāl, where the reason for the reduction (rather than lengthening) of the (pro-)pretonic /a/ is because these forms are always in construct with pronominal suffixes or other nouns (A. Fox 1984; Lipiński 1997; Kim 2012).
Summary by Language: Akkadian and NWS
- Akkadian: The parās– pattern was extremely productive as an infinitive and partook equally in its verbal and nominal properties, probably similar to the situation in Proto-Semitic. Parūs– and parīs– nouns are uncommon. Pāris- was the active participle pattern, but for nomen agentis nouns of profession/habitual action/etc., these were often but into the D-stem equivalent of the qatāl– pattern, namely parrās- (e.g. dayyānum ‘judge’).
- Aramaic: The qatīl pattern became the productive passive participle, while a few qatūl nouns were inherited. The infinitives develop a mi- prefix, a subject which deserves its own post. The qāṭil, originally a participle, becomes a noun. The nomen agentis qaṭṭāl pattern is also productive.
- Hebrew: The qatūl pattern became the productive passive participle, while a decent number of lexicalized qatīl nouns where inherited. The infinitive absolute comes from *qaṭāl, while the jury is still out on the infinitive construct. Originally, the qāṭil was preferred for the active participle (noun) or, when complementing the verb hāyā, used for ongoing action in the past (verb). Through contact with Aramaic, it eventually comes to be used as a present tense verb (as in Modern Hebrew). The etymologies of forms like גִּבּוֹר gibbōr and טַבָּח ṭabbāḥ, although both somehow related to the D-stirp qaṭṭāl- (and a separate qaṭṭál-?) verbal (agent) noun, remain slightly obscure.
Summary by Form: The Hypothetical Proto-Semitic Situation
To summarize, then, the forms mentioned here and their potential semantic distinctions in Proto-Semitic, we end up with the following situation:
- qaṭūl – “X having been done,” “the one/thing to whom X was done,” (passive/resultative nominals [adjectives and participles]); “to be X-ed” (passive infinitives).
- qaṭīl – “the state of (or resulting from) X,” “the one/thing whose state has changed from the result of X having happened” (stative/resultative nominals, gerunds, and infinitives).
- qaṭṭīl – “(one who is in) the state of regularly doing X” (D-stirp stative adjective).
- qaṭāl – “the doing of X” (active gerund); “to do X” (active infinitive).
- qaṭṭāl – “the one who does X over-and-over again (i.e., regularly/habitually, often professionally)” (D-stirp agent noun).
- qāṭil – “the doer of X” (participle) – formally unrelated to the previous five.
Caveat: Of course, this should be understood as a heuristic framework intended to illustrate hypothetical distinctions. The linguistic reality of Proto-Semitic as it would have been spoken “on the ground” was undoubtedly far more complex. To be clear, we do not have enough evidence to reconstruct semantic values tied to vowel quality in PS with this degree of precision, but the correlations preferred here are inferred from patterns attested in the daughter languages. As such, they provide one possible explanatory model.
Conclusion
The Proto-Semitic noun pattern C1aC2(C2)V:C3– was used for verbal substantives, the quality of the long vowel of which was associated with whether the verbal root was active (/ā/), passive-resultative (/ū/), or stative-resultative (/ī/). The active (qaṭāl-) pattern overlapped with the common qāṭil– pattern, relegating it for the most part to an infinitival (rather than active participial) use, and it could also be put into the D-stem (qaṭṭāl-) as a nomen agentis of regular (habitual and professional) actors. Similarly, the stative (qaṭīl-) pattern could be put into the D-stem (qaṭṭīl-) as a stative adjective. The reflexes of these six forms overlap and diverge in the various daughter languages, but all can generally be situated on the cline of nominality as agent nouns, adjectives, participles (active and passive), gerunds, or infinitives (active and passive)—comparable to English “writer,” “the one writing,” “written” (adj.), “the written thing” (n.), “the act of writing,” “to write,” and “to be written,” etc.
Bibliography
I’m sure there’s a lot of bibliography on noun patterns in Proto-Semitic, which I haven’t yet consulted. Of course, the obvious starting places are Fox, Semitic Noun Patterns, and Huehnergard’s work on Proto-Semitic, which I regularly consult. For the Gə‘əz, I used Tropper’s grammar. For the two infinitives in Hebrew, Kim 2012 is a helpful orientation to the previous literature. And again, the blog post by Suchard (here) provides a nice overview of some NWS forms.
About The Author
Two Cases of a Pseudo-Dual in Biblical Hebrew 
Rare Semitic Languages 
Semitic Consonant Correspondences