Notes on the Emergence of New Semitic Roots in the Light of Compounding

Independently of the question as to whether bior triradical roots have historical preponderance in Semitic, there are clear cases of Semitic verbal and nominal roots that have emerged through a process of compounding or integration of additional elements (verbal or nominal affixes and even prepositions). In this paper, an attempt will be made to establish a hierarchical typology of such processes of morphophonological re-analysis, in both historical and modern times.


Introduction
It has long been recognized that new roots in Semitic can emerge through the re-analysis of verbal or nominal affixes.On the one hand, there has been the theory of "matrices et étymons", propagated notably by Georges Bohas (e.g., BOHAS 2000) and a number of his pupils, which builds on the observation that roots with two common consonants and a homorganic, but different third consonant, often share a common semantics.Christopher Ehret (notably EHRET 1995) has developed this theory even further, in trying to reconstruct uniconsonantal semantic core elements at an early stage of Afroasiatic.This well-known line of thought will not be pursued here (for an overview of the pros and cons, cf.e.g., ZA-BORSKI 1991).Rather, with a focus on Arabic and Hebrew, I will give an overview of various processes that can be described as "compounding", which involve either the integration of grammatical morphemes-verbal and nominal affixes, including prepositions-or the creation of new roots by exploiting acronyms or the conjunction of clipped elements, captured by the Arabic term naḥt, literally 'sculpture'.

Integration of grammatical morphemes
Adam MEZ (1906) was among the first to present relevant examples in Arabic, an example being the root √ s-b-q in Arabic sabaqa 'to leave behind', which was explained as resulting from the compounding of the causative afformative s-(also present in the Arabic form X, together with the reflexive afformative -t-) and the root √ b-q-y in Arabic baqiya 'to remain behind'.The productivity of such processes can also be demonstrated by a modern Arabic neologism.Next to the form IV of the root √ s-l-m ʔaslama (IPF yuslimu) 'to render onesolf in security, become a Muslim', a verb ʔaslama (IPF yuʔaslimu) with the meaning 'to islamize' has emerged (in possible analogy to verbs such as ʔamraka 'to americanize'), which synchronically represents the root √ ʔ-s-l-m.
This kind of re-analysis is not specific to Arabic.HUEHNERGARD (2014: 14), based on KURYŁOWICZ (1973: 7) and others, adduces the common-Semitic verb √ š-k-n, Akkadian šakānum 'to place, put, set'.As the Central Semitic cognates of šakānum, Hebrew šāḵan/ šāḵēn, Aramaic šǝḵen, and Arabic sakana 'to dwell' are intransitive, HUEHNERGARD assumes two roots in this context: the transitive Akkadian root can be explained to derive from an š-causative form of the verb √ k-w-n 'to be firm, fixed', whereas the intransitive Central Semitic verbs with the meaning 'to dwell' derive directly from a root √ s-k-n (Ugaritic features both roots).I would suggest that "to dwell" could also be rephrased in a transitive way as "to make a living" or the like and thus could also be explained an an original causative.Another example of this kind offered by MEZ (1906) is the Arabic verb √ r-t-ʕ rataʕa 'to graze', derived from form VIII of the common Semitic root √ r-ʕ-y.The same phenomenon surfaces in the nominal form taqwā 'belief', derived from form VIII ittaqā of the verbal root √ w-q-y.A comparable Hebrew case is presented by the secondary root √ t-ḥ-l, attested in the hifʕīl-binyan as hitḥīl 'to begin', ultimately deriving from the root √ ḥ-l-l with the same meaning in the attested in the hifʕīl-binyan as hēḥēl.Diachronically, a comparable process can be observed in the re-analysis of the Arabic noun madīna 'city', which historically constitutes a noun of the pattern /maC 1 C 2 aC 3 a/ based on the root √ d-y-n, and which synchronically functions as a noun of the pattern /C 1 aC 2 īC 3 a/, based on the root √ m-d-n.Accordingly, the historical plural form is madāʔin, whereas the modern plural form is mudun.1 HUEHNERGARD (2014: 10) cites the Biblical Hebrew plural form dǝlāṯōṯ 'doors', which was generated by re-analyzing the final feminine t of the singular deleṯ 'door' as a root consonant.
In modern Hebrew, such processes are especially productive.USSISHKIN (1999: 407) lists, among others, the following examples, in which nominal affixes are integrated into new expanded roots.In most cases, quadri-radical roots emerge, which are fitted into the piʕʕel or hitpaʕʕel binyanim; in the case of the noun (tertiae infirmae) qaṣe (√ q-ṣ-h), the hifʕil binyan is used: In the case of pele(ʔ) 'wonder' + ṭelefon 'telephone' > pele(ʔ)fon 'smart phone', the firt letter pe of pele(ʔ) replaces the first letter ṭet of ṭelefon.And in the case of yadid 'friend' + zayin 'penis' > yaziz 'lover' ("penis friend"), the first letter zayin of zayin (in itself historically a "polite" form using just the first letter of the word zanav 'tail') twice replaces the letter dalet of yadid.2

Root formation via acronyms
Acronyms as new roots occur in modern Arabic, notably in religious, political, and technical context.As in many European languages, one can observe a tendency to create acronyms that simultaneously have a meaning of their own.The following two political examples are prominent: ḥarakat at-taḥrīr al-filasṭīnīya 'Palestinian liberation movement' ḥamās (<ḥ-m-ā-s>): ḥarakat al-muqāwama al-ʔislāmīya 'Islamic revolt movement' In the first example fatḥ, the order of the constituents (<ḥ-t-f>) is reversed, in order to yield the meaning "conquest" and to avoid the meaning "slow death". 3The second example ḥamās in itself means "rage, enthusiasm".
The formation of new roots through a process of clipping and blending is especially productive in modern Hebrew.Here is a selection of examples (cf., e.g., EDZARD 2006) ).Amharic also features some true blends, e.g., dämoz 'salary', representing the concatenation of däm 'blood' and wäz 'sweat' (necessary ingredients for making a living). 8

Conclusion
Compounding and blending, while being less frequent than in Indo-European languages, nevertheless have emerged as powerful mechanisms in Semitic.This observation also pertains to "compound roots" or expanded roots, i.e., newly emerging Semitic roots, in which nominal prefixes, prefixes of diatheses, or even prepositions governed by the respective verb are integrated into the new root. 9This question is touched upon here here without prejudice to the question of "original" bilateralism vs. trilateralism in Semitic roots. 107 Cf.NÖLDEKE 1898: 83f.

Root formation by exchange of a root consonant
Next to verbal and nominal affixes, prepositions also can be integrated in newly emerging roots.Among the more prominent cases figures Arabic ǧāʔa (√ ǧ-y-ʔ) 'to come' +bi-'in, with', which in Arabic dialects surfaces as ǧāb (√ ǧ-y-b) 'to bring'.A similar case already occurred in Classical Syriac, where the verbal root √ n-t-n 'to give' is typically construed with the dative marker lə-'to', resulting in a new root √ n-t-l 'to give' (cf., e.g.,EDZARD  2011).
Both new nouns, adjectives and verbs can emerge as the result of blending and (preceding) clipping (see below section 2.5).Of special interest is the case, where only one letter of one part of the blend replaces another letter in the other part, thus creating a new root.In the case of tapuax̱ 'apple' (√ t-p-x̱ ) + zahav 'gold' > tapuz 'orange' ("gold apple"), the first letter zayin of zahav replaces the last letter x̱ et of tapuax̱ , thus creating the new root √ t-p-z.