PART 2: DYNAMIZATION OF SYNCHRONY - RELATED LANGUAGES ACROSS TIME

 

Serial Verb Constructions in North-West Semitic languages: From a synchronic radiation back to the 'Big Bang'

 

 

Alexander Andrason

Faculty of Humanities, University of Cape Town, South Africa E-mail: aleksand@hi.is

 

 

---

ABSTRACT

The present article argues that Serial Verb Constructions (SVCs) in North-West Semitic (NWS) languages have emerged from clause fusion. The analysis of the synchronic profiles of SVCs in four of the oldest attested languages of this branch, i.e., Canaano-Akkadian, Ugaritic, Biblical Hebrew, and Biblical Aramaic, reveals an evolutionary path from less cohesive non-canonical serializing patterns of a pseudo-coordinated character to increasingly more cohesive and canonical serializing patterns. The ultimate source of this path and verbal serialization is reconstructed as conjunctive coordination with two clauses being linked by the predecessor of a coordinator that surfaces as u/w in the four analyzed languages.

Keywords: Serial verb constructions; North-West Semitic (Canaano-Akkadian, Ugaritic, Biblical Hebrew, Biblical Aramaic); prototype/canon; dynamization of typology

---

 

 

1. Introduction

The present article is concerned with the emergence of Serial Verb Constructions (SVCs) in North-West Semitic (NWS) languages. By analyzing the synchronic profiles of SVCs in the oldest attested NWS languages, i.e., Canaano-Akkadian, Ugaritic, Biblical Hebrew, and Biblical Aramaic, I aim to determine the most plausible source of verbal serialization in this linguistic branch.¹

In my study, I will make a syncretic use of three clusters of closely related theories. In the description of SVCs in the four languages, I will draw on (1) prototype theory combined with canonical typology, as well as on (2) (part of) grammaticalization theory. The interpretation of this grammatical evidence itself will be developed within (3) dynamic approaches to synchronic variation: semantic maps and dynamization of typology. That is, by observing the radiation of SVCs across the most ancient NWS languages - in terms of their compliance with the prototype (or canon) of an SVC and the extent of their overall grammaticalization - I will propose a developmental path along which these constructions travel. The knowledge of the mechanisms operating on this path will, in turn, allow me to reconstruct its most initial unattested part and propose the original 'Big-Bang' structure from which the attested serializing patterns have emerged in all their diversity.

In order to achieve its objective, the article is organized in the following manner: in Section 2, I explain my eclectic framework - the critical element of the present study - and design my research strategy. In section 3, I introduce my evidence. I describe the various properties of SVCs in the oldest NWS languages with regard to their canonicity and grammaticalization. In section 4, I evaluate this evidence within my dynamic framework and answer the research question. In section 5, I conclude my study.

 

2. Theoretical background

2.1 Framework

Following my previous research on SVCs in North-West Semitic (Andrason & Vita 2020; Andrason & Koo 2020) and Indo-European languages (see Andrason 2018a; 2018b; 2019a), I will use a syncretic methodology that combines a semantic-map approach and an approach referred to as dynamization of typology with prototype theory and canonical typology.

According to the semantic-map approach (Haspelmath 2003; Andrason 2016; Georgakopoulos & Polis 2018) and dynamization of typology (Croft 2003), a synchronic variation of properties - whether those exhibited by a construction attested in a single language at a unique point in time or those exhibited by typologically related constructions that are attested across different languages, each spoken in a different period of time - is interpreted in diachronic terms. That is, the various synchronic profiles of a construction² (e.g., profile a, b, and c) - each such profile comprising a set of properties - can be arranged into a sequence. This sequence yields a vector or a network of vectors, the so-called vectored map, that both reflects and reveals an evolutionary path (or a fragment of such a path) along which a particular construction has been developing (see Figure 1 below where the profiles a, b, and c are organized into three consecutive stages on a shared evolutionary path).³ As a result, synchrony becomes dynamic instead of being static - it tells us something (usually very plausible) about the evolution of a construction (see Heine, Claudi & Hünnemeyer 1991; Heine 1997; Croft 2003; Andrason 2016).

 

 

In this vectorial representation, I make extensive use of the closely related concepts of prototype and canon: the former being preferred in prototype theory (Evans & Green 2006; Janda 2015; Hamawand 2016) while the latter in canonical typology (Brown & Chumakina 2013; Bond 2019; Round & Corbett 2020).⁴ A prototype (or a canon) epitomizes a given category to the fullest extent by constituting its most representative, sometimes even ideal, exemplar. It is defined cumulatively as a set of features that are crosslinguistically common and/or cognitively salient (i.e., they differentiate this prototype from the prototypes of other categories most radically and/or efficiently). In my method, such prototypes (or canons) are employed as heuristic devices - the most relevant mileposts on the evolutionary path, from which the conceptual and developmental distance of all the profiles exhibited by a construction is measured. That is, I associate the various synchronic profiles exhibited by a construction with dissimilar degrees of compliance with the prototype (or canon) of a category (or a number of such prototypes and canons) located on the evolutionary path which that construction travels. The profiles that comply with all or most features associated with the prototype, thus instantiating it fully, are canonical. The profiles that only comply with some (sometimes, very few) features are (increasingly more) non-canonical. Crucially, when interpreted in dynamic terms, such canonical and non-canonical profiles (necessarily) attest to different stages of the development. A canonical profile corresponds to the stage of the prototype - it is dynamically concurrent with the prototype itself and thus 'con-canonical'. In contrast, non-canonical profiles result from either an incomplete or an excessive advancement along the path - they are pre-canonical (i.e., they correspond to stages that precede the stage of the prototype) and post-canonical (i.e., they correspond to stages that follow the stage of the prototype), respectively. The most radical non-canonical profiles, whether pre- or post-canonical, relate the construction to other categories, namely, those that are diachronically prior (i.e., those from which the construction has derived) and posterior (i.e., those to which the construction is subsequently developing). (Figure 2 below illustrates this graphically.) My model thus explains the categorial flexibility of a single construction and its transition across various categories -such fluidity typifies constructions exhibiting large semantic/functional potential and is also inevitable if a broad, typological perspective is adopted (i.e., when a large set of constructions of a certain type are studied).

Drawing on the seminal works authored by Aikhenvald (2006; 2011; 2018; forthcoming), as well as no less important studies presented by Muysken & Veenstra (1994), Crowley (2002), Dixon (2006), and Bisang (2009), and complementing them with the results of my own research (Andrason 2018a; 2018b; 2019a; 2019b; 2019c; Andrason & Vita 2020; Andrason & Koo 2020), I view the prototype (or a canon) of an asymmetrical SVC as an integrated mixture of the non-cohesive and cohesive features.⁵ Three features attest to the non-cohesive facet of SVCs: (a) the construction makes use of (at least) two verbs; (b) both verbs are marked by inflections instead of one of them being a non-finite category, e.g., infinitive, participle, or gerund; and (c) the minor verb can be used outside an SVC pattern as a full verb, i.e., as a semantically robust, content, referential lexeme. Six features attest to the cohesive facet of SVCs: (a) the verbs used are not connected by overt clause combining markers, e.g., subordinators, complementizers, or coordinators; (b) the two verbs are not separated prosodically by means of contouring, comma intonation, pause, or any other types of bi-clausal phrasing; (c) the polarity value of an SVC is unitary, i.e., either positive or negative, even if encoded only once in the construction - inversely, the two verbs cannot be interpreted with two distinct polarity values; (d) the argument structure of an SVC is also unitary - this means that the two verbs cannot govern different arguments, especially, the external ones, i.e., subjects; (e) the operators of time, place, and manner/instrument/means operate jointly over the two verbs - thus, the construction disallows operators whose scope is limited to one verb only; (f) similarly, the construction exhibits a unitary TAM value such that the two verbs are marked by the same TAM category or, at least, by categories that are not semantically conflicting. Overall, the cohesive facet implies and necessitates the mono-eventhood of an SVC prototype (i.e., an SVC expresses a single event) as well as its mono-clausality and mono-predicativity (i.e., an SVC belongs to a single clause in which it constitutes a single predicate).⁶

Additionally, I will draw on grammaticalization theory (Hopper & Traugott 2003). In particular, I will associate the development towards a more successful grammatical strategy and, thus, an increase in the grammaticalization status of a construction, with that construction's productivity, entrenchment, and generalization, i.e., higher token and type frequency and, related to it, compatibility with a growing qualitative variety of contexts of use (this may include roots/bases, tenses/moods/voices, text types and genres, (subject) types of referents, and polarity values; see Hopper & Traugott 2003:63-69, 126-130; Traugott 2011; Gisborne & Patten 2011).

2.2 Research strategy

It has recently been suggested that SVCs in ancient NWS languages developed from bi-event, bi-clausal, and bi-predicative structures (Andrason 2019c:124; Andrason & Vita 2020:28-30; Andrason & Koo 2020:26-27). Accordingly, NWS SVCs would attest to a clause-fusion origin, one of the three typologically possible scenarios of the emergence of verbal serialization - the two others being verbal modification and concurrent grammaticalization (Aikhenvald 2018). In the case of clause-fusion origin, the prototype of SVCs - which, as explained in section 2.1 above, blends non-cohesive and cohesive properties - could be understood as a stage on the development from fully non-cohesive structures (i.e., bi-verbal, bi-clausal, bi-predicative, and bi-event) to fully cohesive structures (i.e., mono-verbal, mono-clausal, mono-predicative, and mono-event). What differentiates a prototypical SVC from the genuine non-cohesive and cohesive structures is the non-cohesive and cohesive part of its profile, respectively.⁷Therefore, within my dynamic model, insufficient cohesiveness is associated with pre-canonicity, while excessive cohesiveness is associated with post-canonicity.

By drawing on the dynamic understanding of SVCs in the oldest attested NWS languages, i.e., Canaano-Akkadian, Ugaritic, Biblical Hebrew, and Biblical Aramaic, in terms of superposed vectored semantic maps the critical reference point of which is the prototype of an SVC, I aim to verify whether clause fusion, as recently suggested in scholarly literature, constitutes the most plausible origin of SVCs in this linguistic branch. If clause combining is indeed the most likely source of NWS verbal serialization, I aim to determine the specific types of clauses from which SVCs have emerged. In other words, by observing the radiation of SVCs in the most ancient NWS languages - in terms of both their compliance with an SVC prototype (or canon) and their grammaticalization viewed as a combination of entrenchment, productivity, and generalization - and locating the various radiation profiles on a shared diachronic path, I will reconstruct the most initial, unattested part of the path and thus the original 'Big Bang' from which the attested SVCs have developed.

 

3. Evidence

In the present section, I describe the various profiles exhibited by SVCs in Canaano-Akkadian (3.1), Ugaritic (3.2), Biblical Hebrew (3.3), and Biblical Aramaic (3.4). My evidence largely draws on the previous research that I have conducted on verbal serialization in NWS languages (see Andrason 2019b; 2019c; Andrason & Vita 2020; Andrason & Koo 2020) and the databases of SVCs that I have compiled during that work. With regard to Biblical Hebrew, I will additionally draw on the data presented in (or inferable from) works that have been published by other scholars, in particular, Dobbs-Allsopp (1995), Eskhult (1998), and Chrzanowski (2011).⁸

3.1 Canaano-Akkadian

Canaano-Akkadian is the oldest NWS language attested. It is dated from the 14^th c. BCE and emerges from approximately 280 letters - a part of the larger El-Amarna corpus composed of 382 letters (Rainey 2015) - that were written in the Mesopotamian cuneiform script and sent from Canaan to Egypt (Andrason & Vita 2020).⁹ Strictly speaking, Canaano-Akkadian is a blended language - hybrid, mixed, jargon, and/or inter-language - that contains and combines both East and North-West Semitic components (Andrason & Vita 2014:165). This NWS component is visible most clearly through glosses and, albeit less directly, hybridized forms (Izre'el 2005; 2012; Andrason & Vita 2014; Wilson-Wright 2019).

According to the available evidence, the grammaticalization level of verbal serialization in Canaano-Akkadian is low (Andrason 2019b:31). Canaano-Akkadian only seems to have one serializing pattern in its repertoire, built around the motion verb alàku 'go' (Andrason 2019b; see also Dellaire 2014:176, who analyzes the verb alàku 'go' in some sequences with other verbs as "an adverb"). When used in an SVC, always as its first verb, i.e., V1, alàku expresses "certainty, [...] reaffirmation, [and] completeness" or "emphasis, urgency, and intensity", if the verbs are inflected in the imperative (Andrason 2019b:31). Certainly, there are two further possible serialized constructions in El-Amarna letters, namely those built around the verbs târu 'return' and gamàru 'finish'. These examples are, however, disputed and interpreted in both a serializing (CAD G:28; CAD T:261; Moran 1992:16, 87) and non-serializing manner (Rainey 2015:91, 285). More importantly, they do not reflect a genuine Canaano-Akkadian usage. The two letters in which the constructions with târu and gamàru are found, i.e., EA 8 and 27, are written not by Canaanite rulers but rather the kings of Babylon and Mitanni, respectively.¹⁰Overall, there are only four instances of the serializing uses with alàku 'go' (EA 102:15, 114:28-29, 294:32-33, 306:13; Andrason 2019b:18). Nevertheless, even these scarce four cases attest to the compatibility of verbal serialization with diverse categories. Specifically, the construction tolerates semantically diverse major verbs (e.g., activity, perception, and postural verbs) and diverse TAM categories (e.g., the preterite yaqtul, the imperfect (i.e., the "present-future" iparras; Rainey 1996a:51), and the imperative), as well as both 2^nd and 1^st person subjects. Significantly, however, the alàku SVC is unattested with 3^rd-person subjects, plural subjects, and inanimate subjects, nor is it found with a negative polarity value (Andrason 2019b:32).

As far as the canonicity of the alàku SVC is concerned, the following should be noted: in no case does the serializing construction exhibit a canonical profile, thus fully complying with the SVC prototype outlined in section 2 above. Instead, certain important prototypical features tend to be violated (Andrason 2019b:31). The most relevant of such violations involve: (a) the common use of the linker u (e.g., i-ti-lik / ^ru ^rahⁿ(?)-ta-ni 'I went and I urged'; EA 114:2829) - an element that is (apparently) homophonous with the clause-combining marker u, itself often used as a conjunctive coordinator (Rainey 1996b:97-100; Izre'el 2005; Tropper & Vita 2010)¹¹ (inversely, only once, with the imperative, the linker is absent: a-lik-mi i-zi-iz 'Go stay' (EA 102:15)); (b) the possibility of bi-event readings¹² and thus the interpretation of an example as "a mere combination of two clauses" in terms of (syndetic or asyndetic) conjunctive coordination (Andrason 2019b:31) - apart from an equally possible mono-event interpretation (in other words, there are no cases that could only be interpreted in terms of mono-eventhood); and (c) "the non-unitary treatment [of the alàku SVC] in [...] replies", specifically the use of only alàku instead of the entire bi-verbal sequence (ibid. 31, see also p. 20). With regard to contiguity, the contiguous arrangement of V1 and V2 predominates if the element separating the verbs is analyzed as an empty linker. Only once does a particle occur between V1 and V2; in no case is the interrupting element an argument or adjunct (ibid. 19, 23, 26, 29).

Consequently, verbal serialization in Canaano-Akkadian attests to a significantly more non-canonical than canonical profile. In all instances, non-canonicity stems from insufficient cohesiveness - the three non-canonical features presented above being typical of less evolved serializing patterns (Andrason 2019b) - and thus constitutes the expression of pre-canonicity. This observation is fully consistent with the low grammaticalization degree of verbal serialization in the language.¹³

3.2 Ugaritic

Ugaritic is diachronically posterior to Canaano-Akkadian. The attested Ugaritic texts date from the 13^th c. BCE until the very moment of the destruction of the city of Ugarit in the early 12^thc. BCE, somewhere between 1200 and 1175 BCE (more precisely between 1193/1192 and 1190/1186; Pardee 2011:460; Kaniewski et al. 2011; Knapp & Manning 2016). Ugaritic is, however, the first fully-fledged native NWS language with robust corpora preserved to our times.

The grammaticalization level of serializing strategies in Ugaritic is certainly higher than was the case of Canaano-Akkadian: Ugaritic SVCs constitute a well visible and "relatively productive" morpho-syntactic device in the language (Andrason & Vita 2020:28). Their visibility and productivity are, however, lower than what is typical of genuine serializing language systems. To begin with, the available Ugaritic corpora (KTU³ 2013) attest to eight types of SVCs that (tautologically) draw on eight different roots as their respective minor verbs - invariably used as the initial verbs in the construction, i.e., V1. These roots are: qwm, ns', ndd, hlk, ytb/twb, hws, w/yd', bky, and ngr (Andrason & Vita 2020:27).¹⁴ All these constructions yield a rather meagre set of 19 instances of verbal serialization (ibid.). Nevertheless, the existing examples attest to a relative diversity of minor verbs used. To be exact, most minor verbs present in SVCs are, in their lexical meaning, motion and/or postural verbs, for example, hlk and ndd 'go, move' (the latter also meaning 'hurry, rush, launch oneself), twb 'return',ytb 'sit down', and qwm and ns' 'rise, stand up, go up'. Few remaining minor verbs draw on roots that are associated with other types of semantic domains: modality (w/yd' 'know, be able'), manner (hws 'hurry up, hasten'), and strong emotions (bky 'cry, weep'; ibid. 27-29). In the respective SVCs, these minor verbs contribute to the aspectual (ingression/inception, continuity, duration, iterativity) and modal (emphasis, insistence, urge(ncy), ability) interpretation of the major verb, as well as the manner with which the action conveyed by the major verb is performed (e.g., rapidness, haste, immediacy, sadness, i.e., crying; ibid. 27).¹⁵ Similarly, if envisaged in their totality, SVCs are compatible with semantically diverse types of major verbs, which specify the lexical type of action or activity involved in the event (ibid. 29). These major verbs - always used as V2 (ibid. 27) - express activity (whether telic or atelic), motion, posture, and state. SVCs are also compatible with different TAM categories (i.e., the suffix conjugation qatala, the prefix conjugationyaqtul, the long prefix conjugation yaqtulu, the short modal prefix conjugation yaqtul, the imperative, and the infinitive absolute) and most types of subjects (person, gender, number) and subject inflectional markers (i.e., 1^st, 2^nd, or 3^rd, singular or plural, and masculine, feminine, or common gender; ibid. 29). The TAM categories typically express the ideas of a narrative (perfective) past and directive modality. The most common subject is 3^rd masculine singular and, in the imperative, 2^nd masculine singular. Furthermore, SVCs are found in the two principal registers of Ugaritic corpora, namely, those that imitate (or reflect) a spoken or more colloquial language (e.g., letters) and those that tend to be more conservative from a grammatical and stylistic perspective (i.e., narrative and epic poetry; ibid.). All SVCs attested exhibit a positive polarity value (ibid.). This means that similar to Canaano-Akkadian, a negative type of verbal serialization is absent. The use of non-human subject referents is also unattested (ibid.).

Contrary to Canaano-Akkadian, the majority of the examples attest to relative canonicity (Andrason & Vita 2020:27-28), thus complying with all or most features of the SVC prototype. Crucially, in most cases, an element that is homophonous with a clause-combining marker or "multi-functional connector", i.e., w, is absent. The absence of w is regular if an SVC involves verbs inflected in the imperative. Furthermore, nearly all verbs that appear with the linker w, also attest to linker-free structures (ibid.). Similarly, in most cases, the verbal sequence is contiguous, especially if the dummy element w is treated not as a constituent (i.e., a conjunctive coordinator) but an empty linker - a morpho-syntactic marker of the SVC itself (see Aikhenvald forthcoming). Equally important is the fact that in some cases, a mono-event interpretation is not only fully plausible but also significantly more likely than a bi-event reading (Andrason & Vita 2020) - in Canaano-Akkadian, the analysis in terms of mono-eventhood and bi-eventhood were equally plausible (see section 2.1 above).¹⁶ In contrast, non-canonical examples are less common (ibid. 28). The typical signs of non-canonicity are: the presence of the empty or dummy element w homophonous with the clause-combining marker w, most commonly used as a conjunctive coordinator (Tropper 2012; Tropper & Vita 2020); the use of a concordant object marking on both V1 and V2; and, perhaps, the non-contiguous placement of the two verbs, especially, their separation by an object argument (Andrason & Vita 2020:27-29).¹⁷ Lastly, in one ambiguous case, non-canonicity may be related to a potential interpretation of the minor verb as an "invariable particle", a "discourse marker", or an "interjection" (Andrason & Vita 2020:21, 27-28).¹⁸

To conclude, verbal serialization in Ugaritic tends to exhibit a more canonical than non-canonical profile. In most cases, non-canonicity stems from insufficient cohesiveness and thus implies pre-canonicity (see the use of an empty linker, concordant object marking, and non-contiguity, which are all characteristic of less developed serializing patterns). This prevalent relative canonicity is consistent with a considerable - although not too high -grammaticalization degree of verbal serialization in the Ugaritic language in general. In one ambiguous case, non-canonicity may stem from the excessive cohesion or further development of V1 and thus imply post-canonicity.

3.3 Biblical-Hebrew

Biblical Hebrew is a significantly more complex type of a corporalect than Canaano-Akkadian and Ugaritic. Although the main bulk of the Hebrew Bible most likely reflects a variety (or varieties) that was used between the 10^th and 6^th c. BCE (Standard Biblical Hebrew), in its totality, the BH language covers almost ten centuries (Rubin 2010:17-16). It spans from the 12^th/11^th c. BCE (the variety of this period is referred to as Archaic Biblical Hebrew) until the 3^rd/2^nd c. BCE (this variety is referred to as Late Biblical Hebrew) (Steiner 1997:146; Rendsburg 1997:66; Rubin 2010:16; Edzard 2011:481; Hornkohl 2014).

Verbal serialization constitutes an important component of the Biblical Hebrew verbal system, being "central" (Andrason & Vita 2020:29), "productive" (ibid.), and "robust" (Andrason & Koo 2020:8; see also Andrason 2019c).¹⁹ Overall, at least fourteen roots can be used as minor verbs in SVCs always used as V1: hlk, qwm, swb,ysp, mhr, ná', ntn, rwm (in Hiph'il) skm (in Hiph'il), rbh 'be many' (in Hip'hil), y'l (in Hiph'il), sbb, hll (in Hip'hil), andykl (Gesenius 1910:386-387; Lambdin 1971:238-239; Dobbs-Allsopp 1996; Eskhult 1998; Chrzanowski 2011; Andrason 2019c). Although, to my knowledge, the total number of examples of SVCs has not been calculated, serializing patterns are certainly a common feature of the language. This may be inferred from the high number of SVCs that draw on a few minor verbs for which such frequency data exist, namely, qwm, swb, ysp, and mhr. To be exact, there are 152 cases of SVCs with qwm (Andrason 2019c), 57x with swb, 34x with mhr, and 14x with ysp (cf. Chrzanowski 2011). The verbs that are used as minor verbs in SVCs may draw on roots (or stems) that, in their lexical meaning, express: motion (hlk 'go, walk', swb 'return', sbb 'surround, turn'; see also mhr 'go quickly, hurry, hasten'), posture (qwm 'rise, stand up'), activity - addition (ysp 'add, increase', rbh (Hiph'il) 'do a lot, increase') and giving-taking (ná ' 'take, lift', ntn 'give', rwm (Hiph'il) 'lift up'), among others (see also hll 'open, start') - and semantically more complex and/or mixed (semi-stative) verbs (see y'l (Hiph'il) 'be willing, please; do willingly' and ykl 'be able, prevail'). In the respective SVCs, the above minor verbs express a variety of aspectual and modal senses: repetition/iteration (swb and ysp), inception (hll, sbb), cessation of performing an action, i.e., 'no longer' (ysp), ingression, immediacy, urgency (qwm), rapidness (mhr), an early performance of an action (skm), volition (y'l; also contributing to politeness), and ability/possibility (ykl) (Gesenius 1910; Lambdin 1971:238239; Dobbs-Allsopp 1995; Eskhult 1998; Chrzanowski 2011; Andrason 2019c:115). In their totality, SVCs can be used in both affirmative and negative contexts (e.g., ysp) although this ability varies considerably for different types of minor verbs used (e.g., swb, mhr, and qwm (see below) are only used in positive constructions; Chrzanowski 2011; Andrason 2019c). Among the various types of SVCs, only the pattern built around the minor verb qwm has been analyzed to the extent that allows the determination of its grammaticalization and canonicity levels (see Andrason 2019c). Below, I present the results of this analysis.

The grammaticalization of the qwm SVC is advanced although not completed. On the one hand, several features signal a considerable extent of grammaticalization. First, the token frequency of the qwm SVC in the biblical corpus - or the number of the attested examples of this construction - is high, ascending, as mentioned above, to 152 cases (Andrason 2019c:106). Second, the qwm SVC is fully productive as the set of major verbs compatible with this construction is open. That is, various semantic types of major verbs are tolerated, e.g., verbs of motion, posture, activity, or state, as well as, more generally, both telic and atelic verbs. The most common are the major verbs conveying the idea of motion (ibid. 120-121). Third, the qwm SVC is compatible with most types of TAM categories: yiqtol, wayyiqtol, yiqtolah, weqatal, and the imperative. The immense majority of cases involve the wayyiqtol form (a narrative, mostly perfective past) and the imperative (ibid. 111-112). Fourth, the qwm SVC is used in all types of texts, specifically, both prose (i.e., genuine narrative, personal narrative, and discourse) and poetry (ibid. 121). Fifth, the verbs can be inflected in all persons (1^st, 2^nd, 3^rd), genders (masculine, feminine, common), and numbers (singular and plural) (ibid. 116117). The most common are 3^rd singular masculine and 2^nd singular masculine (typical of the imperative). On the other hand, certain properties indicate that the grammaticalization of the qwm SVC is not complete(d). First, the maintenance of a postural component in the qwm SVCs (and thus a more lexical reading of the verb qwm) is not uncommon (Andrason 2019c:120). Second, there are no examples in which the subject referent of the qwm SVC is non-human or inanimate (ibid. 121). Third, the qwm SVC is only attested with a positive polarity value. Examples of its use in negative contexts are thus absent (ibid. 122). Fourth, despite the fact that the qwm SVC can be inflected in all person, gender, and number categories, their particular combination into 3^rd person feminine plural is unattested. Fifth, although the qwm SVC is attested in poetry, such examples are extremely rare. This may suggest that the construction is well entrenched in genres and registers that are closer to colloquial language (where grammaticalization processes tend to originate), e.g., discourse; as well as in genres and registers where grammaticalization processes occur immediately after they took place in discourse, i.e., personal narrative and, subsequently, genuine narrative. In contrast, the qwm SVC has not been generalized in poetry, which is "a genre that is more resistant to innovations" (ibid. 121).

With regard to the compliance with the SVC prototype, the qwm SVC often exhibits a (nearly) canonical profile although instances of lesser canonicity are also well attested (Andrason 2019c). One of the signs of lesser canonicity is the presence of w. However, the use of an overt linker w is very rare. Much more commonly, the linker w is either absent or constitutes an inseparable part of a given verbal gram itself, i.e., wayyiqtol and weqatal. The absence of the linker w is particularly common with verbs inflected in the imperative (ibid. 106-108; see also Lambdin 1971:239). In contrast, with yiqtol and cohortative forms, the overt linker w is regularly employed (Andrason 2019c:107). The qwm SVC attests to both conjunctive and disjunctive accentuation in relatively equal proportions. The former suggests a more mono-clausal intonation, while the latter suggests a less mono-clausal (or more bi-clausal) intonation. Nevertheless, conjunctive accents tend to be strong, whereas disjunctive accents are usually weak (ibid. 109-111).²⁰ Although V1 and V2 are typically inflected in the same TAM categories, in certain instances, such TAM categories may differ. In all such cases, however, the TAM categories used are semantically compatible, i.e., yiqtol + weqatal and imperative + weqatal/yiqtolah. This means that the qwm SVC does not violate the principle of shared TAM marking (ibid. 111-113). Similarly, although in most cases, V1 and V2 share their subject referent and are inflected in the same person, number and gender, the presence of different subject markers is also attested. This (very sporadic) phenomenon emerges in cases where V1 and V2 are inflected in the singular and the plural respectively. This mismatch, however, does not constitute a genuine violation of the same-subject prototypical feature. It usually stems from the particular rules governing agreement between the heading verb and its complex subject (i.e., coordination of two noun phrases). The above-mentioned mismatch thus corresponds to an intermediate stage between the compliance and violation of the shared-subject property (ibid. 116-118). Lastly, although a vast majority of examples attest to a (more or less evident) mono-eventhood interpretation, some of them allow for ambiguous readings with patent motion/posture-related undertones (ibid. 115-116). Fourth, in more than half of the cases, the construction attests to the contiguous order of V1 and V2. In cases of discontinuity, V1 and V2 tend to be separated only by the subject argument (ibid. 119). Lastly, some cases of the qwm verb inflected in the imperative are ambiguous with qwm allowing for an interpretation in terms of both a minor verb and a modal particle or discourse marker (ibid. 120).

Overall, the qwm SVC is predominantly (nearly) canonical. Nevertheless, a non-canonical profile is also visible and surfaces in several cases. This non-canonicity can usually be attributed to a lesser extent of cohesiveness, thus being of a pre-canonical type. In a few cases, non-canonicity stems from an excessive advancement along the evolutionary path and corresponds to its post-canonical stages. The canonicity profile explained above concords with the grammaticalization degree of the qwm SVC - and the grammaticalization degree of verbal serialization in general - which is high, although not completed.

3.4 Biblical Aramaic

Biblical Aramaic is the Aramaic language attested in the Hebrew Bible. The largest portions of it come from the younger, post-classical books, namely Ezra (4.8-6.18; 7.12-26) and Daniel (2.4b-7.28; Folmer 2012:130).²¹ These two sub-corpora, i.e., Ezra and Daniel, differ chronologically. The Aramaic of Ezra most likely dates from the 6^th c. BCE, while the Aramaic of Daniel probably reflects the language of the 4^th c. BCE (Kaufman 2005; Lipinski 2001:6170; Gzella 2004:41-45; 2011b:583; Folmer 2012). Nevertheless, these two varieties tend to be considered jointly and viewed as a single Aramaic dialect - itself closely related to Official or Imperial Aramaic (Gzella 2011a:431; 2011b; Folmer 2012:130).²²

Verbal serialization is a well grammaticalized strategy in Biblical Aramaic and constitutes "an integral part of the verbal system" of this language (Andrason & Koo 2020:25). This extent of grammaticalization is relatively evident despite the very limited size of the BA corpus. To begin with, there are twenty cases of possible SVCs in which twelve different verbal roots are used as minor verbs. This demonstrates significant productivity of the serializing pattern (ibid. 27). These minor verbs, which always appear as V1, belong to several semantic types. They express, in their lexical meanings, the following semantic domains: motion ('th 'come', qrb 'approach, come near', 'll 'enter', 'zl 'go (off, away)', and npq 'go/come out'), posture (qwm 'rise, stand up' and npl 'fall'), activity (ns' 'lift, take', Haph'el of twb 'bring back', and slh 'send'), and state (bns 'to be angry' and rbh 'to be(come) big(ger)') (ibid. 10-25). Thus, minor verbs found in SVCs in Biblical Aramaic make use not only of domains that are located high on the hierarchy of serialization, i.e., motion, posture, take-bring; there are also those that correspond to its lower part, i.e., statives and verbs of sending (cf. Aikhenvald 2006; 2018). Nevertheless, the majority of the cases attested do involve motion and postural verbs, which are also the most productive attesting to greater morphosyntactic diversity and higher canonicity (see next paragraph) than the other types of roots (Andrason & Koo 2020:26-28). Furthermore, verbal serialization is compatible with both affirmative and negative contexts (ibid. 27). It also tolerates diverse TAM categories, namely, the suffix conjugation q^etal, the prefix conjugation (modal-future) yiqtul, the imperative, and the active participle qâtel (ibid.). The most common is the narrative (perfective) past q^etal. Although the active voice constitutes the most common type of diathesis exhibited by SVCs, serializing constructions are also used in the passive voice (i.e., Hoph'al and P^e'il) and the impersonal passive (ibid. 21-22, 27). Similarly, all types of subject inflections are attested in SVCs whether this concerns person, number, or gender subject inflections. The 3^rd person, the masculine, and the singular are the most common categories - with the 3^rd masculine singular being the most frequent (ibid. 27). SVCs can be found in both main and dependent clauses, including those introduced by the relativizer dî (ibid. 27-28). Lastly, SVCs appear in direct speech and narrative, whether genuine or personal (i.e., narrative discourse; ibid. 28).

SVCs in Biblical Aramaic attest to a considerable degree of canonicity. Although only two examples are fully canonical, a number of other instances are nearly canonical. Such nearly canonical cases diverge from the prototype only with respect to phonology by exhibiting disjunctive accents. Nevertheless, the majority of cases are non-canonical, and the violations concern not only phonology but also syntax (as well as, very rarely, morphology; Andrason & Koo 2020:26-28). The main reason for non-canonicity is the presence of the empty linker or dummy juncture w and, as mentioned above, the use of disjunctive accents. In some instances, the two verbs are not contiguous - however, if waw is disregarded, only one constituent, invariably the subject, can separate the two verbs (ibid. 26). Once, non-canonicity results from the presence of distinct grammatical (albeit not logical) subjects on V1 and V2 and, as a result, two different inflectional markers (ibid.). It should be noted that the fully or nearly canonical examples involve the motion minor verbs 'zl 'go' and 'th 'come', the postural minor verb qwm 'rise', and the activity minor verb ns' 'take'.²³ In contrast, the stative verbs bns 'to be angry' and rbh 'to grow, become bigger', as well as the activity verb slh 'send' only exhibit non-canonical uses (ibid. 26-27). The most canonical uses are affirmative and inflected in the imperative, as well as, albeit less commonly, in the q^etal form (ibid. 26). All canonical examples also attest to the contiguous position of the V1 and V2 (ibid.).

To conclude, SVCs occupy two stages in Biblical Aramaic: non-canonical and (nearly) con-canonical. Although non-canonicity prevails and its contribution to the map is greater than that of canonicity, canonicity and near canonicity - where the only non-canonical features pertain to phonology - are also well attested. All non-canonical instances emerge due to the insufficient extent of cohesiveness, thus attesting to pre-canonicity. This means that no cases of excessive cohesiveness, and post-canonicity, are attested. This relative extent of canonicity is correlated with the quite considerable grammaticalization degree of verbal serialization in the language.

 

4. Discussion

The comparison of the profiles exhibited by SVCs in Canaano-Akkadian, Ugaritic, Biblical Hebrew, and Biblical Aramaic reveals a patent increase in the grammaticalization and canonicity levels of verbal serialization. That is, the four languages, which follow one another chronologically, can be arranged into a (more or less linear) sequence - the manifestation of a shared evolutionary path. This path leads from a low to an extensive degree of the grammaticalization of verbal serialization and from predominant pre-canonicity to (gradually) more prominent (con-)canonicity and, subsequently, to (still scarcely attested) post-canonicity.

With regard to the grammaticalization of serializing patterns, its low extent is characteristic of Canaano-Akkadian. Ugaritic attests to a considerable degree of grammaticalization of verbal serialization. In Biblical Hebrew and Biblical Aramaic, this grammaticalization is even more extensive. This steady grammatical development is evident through a number of phenomena: the expansion of the set of minor verbs (1x in Canaano-Akkadian > 8x in Ugaritic > 14x in Biblical Hebrew > 12x in Biblical Aramaic); the growing number of SVC examples (from a handful in Canaano-Akkadian to a few hundred in Biblical Hebrew²⁴ via nearly two dozen in Ugaritic (the twenty cases attested in Biblical Aramaic are most likely due to the very limited corpus of this language)); the use of semantically diverse minor verbs (from the only motion verb in Canaano-Akkadian to several types, including stative roots, attested in Biblical Hebrew and, especially, Biblical Aramaic); the use of more diverse major verbs (from motion and posture verbs in Canaano-Akkadian to all types of verbs in Biblical Hebrew and Biblical Aramaic); compatibility with negative polarity (absent in Canaano-Akkadian and Ugaritic but present in Biblical Hebrew and Biblical Aramaic); and the expansion to all types of diathesis (from exclusively active uses in Canaano-Akkadian and Ugaritic to active and passive uses in Biblical Aramaic). Despite being considerably advanced in the two biblical languages of the NWS family, the grammaticalization process is not completed as the use of non-human (whether animate or inanimate) subject referents is attested in neither Biblical Hebrew (at least with the minor verb qwm) nor Biblical Aramaic.

Similarly, with regard to the canonicity of verbal serialization, Canaano-Akkadian only attests to non-canonical uses (always of a pre-canonical type). The canonicity of SVCs increases in Ugaritic and Biblical Hebrew. In Biblical Aramaic, it seems lower than in Ugaritic and Biblical Hebrew, although significantly higher than in Canaano-Akkadian.²⁵ This gradual proliferation of more canonical examples is principally visible through the following phenomena: the increasing number of (exclusively) mono-event interpretations (from no undisputed cases of mono-eventhood in Canaano-Akkadian to several such cases in Ugaritic, Biblical Aramaic, and especially Biblical Hebrew); and the increasing number of linker-free uses (from the predominance of the linker in Canaano-Akkadian to its sparce presence in Ugaritic and, as far as the overt type is concerned, in Biblical Hebrew; in Biblical Aramaic, the linker is less common than in Canaano-Akkadian although more frequent than in Ugaritic and Biblical Hebrew). With regard to contiguity, this feature seems to be relatively stable across all the languages and thus throughout the entire development.

Overall, the most grammaticalized and the most canonical uses tend to emerge if SVCs: are built around minor verbs derived from motion (e.g., hlk 'go' in Canaano-Akkadian, Ugaritic, and Biblical Hebrew, as well as 'zl 'go' and 'th 'come' in Biblical Aramaic) and postural (e.g., qwm 'rise' in Biblical Hebrew and Biblical Aramaic) roots; are inflected in the (2^nd singular masculine) imperative (Canaano-Akkadian, Ugaritic, Biblical Hebrew, Biblical Aramaic) and, although less so, the (3^rd singular masculine) preterite (qatala in Ugaritic, wayyiqtol in Biblical Hebrew, and q^etal in Biblical Aramaic); and appear in affirmative active contexts (all four languages). SVCs that draw on the motion verbs - especially those with hlk 'go' - and exhibit the above-mentioned properties are also the only ones attested in all the languages analyzed, including the most ancient ones, i.e., Canaano-Akkadian and Ugaritic.

The above results demonstrate that NWS serializing patterns have been evolving by increasing their grammaticalization and canonicity levels. The attested sources of this evolution are non-canonical SVCs in which the two verbs are linked by an empty linker (L) u/w - captured by the schema V1^minor-L-V2^major. As the linker u/w used in such sequences is homophonous with the lexemes widely used as conjunctive coordinators in the respective languages, this type of a non-canonical serializing construction could alternatively be viewed as an instantiation of the category of pseudo-coordination (see Johannessen 1998; L0drup 2002; Yuasa & Sadock 2002; De Vos 2005; Kvist Darnell 2008; Giusti, Di Caro & Ross 2022). Given the radiation properties explained above, such pseudo-coordinated structures initially drew on the motion verb hlk 'go' as V1, as well as, although (most likely) at a slightly later stage, the postural verb qwm 'rise', that were inflected in the (2^nd masculine singular) imperative and, again (most likely) somewhat later, the (3^rd masculine singular) preterite. The advancement along the evolutionary path has consisted in the expansion of this pattern to a greater variety of contexts (e.g., more types of V1 and V2, other subject referents, TAM constructions, and diathesis types, as well as negative polarity values) and the increase in the cohesiveness of the pseudo-coordinated construction and especially the generalization of the canonical serializing pattern without the linker: V₁^minor _-V₂^major

The evolutionary path revealed above, i.e., from a less cohesive profile (V1^minor-L-V2^major) to a more cohesive profile (V1^minor-V2^major), suggests that the stage preceding lesser cohesiveness was most likely that of conjunctive coordination (i.e., V1^full-verb and V2^full-verb). This may be proposed because of four phenomena: first, from a typological perspective, pseudocoordination regularly derives from genuine conjunctively coordinated clauses (i.e., clauses connected by means of a conjunctive coordinator 'and'); second, as mentioned above, the linker u/w used in the four NWS languages is homophonous (or nearly homophonous) with the lexeme that in one of its functions acts as a conjunctive coordinator; third, the earliest non-canonical SVCs attested in Canaano-Akkadian allow for alternative bi-event readings that are all of a conjunctively coordinated type; and fourth, conjunctively coordinated clauses with u/w are widely attested in the ancient NWS languages, constituting a highly productive template, certainly prone to be reused for other grammatical purposes. In other words, given the attested radiation across the oldest NWS languages that cover a (more advanced) portion of the evolutionary path - from non-canonical serialization (pseudo-coordination) to canonical serialization - one can "travel" the remaining fragment of the path and reconstruct the input construction as conjunctive coordination with the predecessor of the coordinator u/w 'and', a Proto-NWS *wa. Given the properties of the earliest pseudo-coordinated structures and the profiles radiating chronologically in the subsequent languages, this conjunctive coordinated input most likely involved motion/posture verbs (especially hlk 'go' as well as, most probably, slightly later qwm 'rise') inflected in the imperative (2^nd person) and preterite (3^rd person) and used in affirmative and active contexts.

Consequently, the above discussion suggests that, in NWS languages, verbal serialization has emerged via a clause-fusion mechanism. Conjunctive coordination - i.e., two clauses linked by the coordinator u/w 'and' (V1^full-verb and V2^full-verb) - gradually increased its cohesiveness by transmuting into a non-canonical SVC with an empty linker w/u homophonous with the coordinator u/w 'and', a type of pseudo-coordination (V1^minor-L-V2^major), and ultimately acquiring a canonical serializing status, thus giving rise to an SVC in which the linker has been eliminated (V1^minor-V2^major).

In addition to answering the research question, the results of the present study may enhance our typological knowledge of the diachrony of SVCs. First, similar to what characterizes NWS languages, in Polish, both the canonicity of SVCs and their grammaticalization are the most advanced if the serializing constructions are inflected in the (2^nd masculine) imperative and, albeit slightly less so, the (3^rd singular) perfective past, all used in affirmative active contexts (Andrason 2018a; 2018b). This congruence between NWS and Slavonic languages may suggest some type of universal hierarchy governing the formation and stabilization of serialized patterns whereby certain tenses and moods (e.g., imperative) are more propitious to yield SVCs than others. Second, the use of motion verbs such as hlk 'go' and postural verbs such as qwm 'rise' as the first and/or the most advanced serializing strategies in NWS languages corroborates the serialization hierarchy proposed by Aikhenvald (2006; 2018), whereby these types of minor verbs tend to generate SVCs before other types of verbs and thus acquire canonical profiles more rapidly.

 

5. Conclusion

The present article demonstrates that NWS verbal serialization has most likely emerged from clause fusion. The analysis of the synchronic profiles of SVCs in the oldest attested NWS languages, i.e., Canaano-Akkadian, Ugaritic, Biblical Hebrew, and Biblical Aramaic, reveals an evolutionary path from less cohesive, non-canonical serializing patterns (of a pseudo-coordinated character) to more cohesive, canonical serializing patterns (more consistent with a prototype of SVC). Given the necessary source of pseudo-coordination in conjunctively coordinated clauses, the homophony of the linker used in non-canonical (pseudo-coordinated) SVCs with conjunctive coordinators, the alternative bi-event and (conjunctively) coordinated reading of the oldest serializing examples, and the common attestation of genuine conjunctively coordinated structures in all analyzed languages, the ultimate origin of verbal serialization, or its 'Big Bang', is reconstructed as conjunctive coordination. This original structure arguably consists of two clauses that are connected by the predecessor of the lexemes surfacing as u/w in the analyzed languages, i.e., the Proto-NWS *wa. The most likely context in which this type of serialization has emerged involves of the motion verb hlk 'go' and postural verb qwm 'rise' inflected in the (2^nd masculine singular) imperative and the (3^rd masculine singular) preterite affirmative active.

 

References

Aikhenvald, A. 2006. Serial verb constructions in typological perspective. In A. Aikhenvald and R. M. W. Dixon (eds.) Serial Verb Constructions: A Cross-linguistic Typology. Oxford: Oxford University Press. pp. 1-68. https://doi.org/10.1093/oso/9780198791263.003.0010        [ Links ]

Aikhenvald, A. 2011. Multi-verb constructions: Setting the scene. In A. Aikhenvald and P. Muysken (eds.) Multi-verb Constructions: A View from the Americas. Leiden: Brill. pp. 1-26. https://doi.org/10.1163/ej.9789004194526.i-313.8 '        [ Links ]

Aikhenvald, A. 2018. Serial Verbs. Oxford: Oxford University Press.         [ Links ]

Aikhenvald, A. forthcoming. Serial verb constructions in Papuan languages.

Andrason, A. 2016. From vectors to waves and streams: An alternative approach to semantic maps. Stellenbosch Papers in Linguistics 45: 1-29. https://doi.org/10.5774/45-0-211        [ Links ]

Andrason, A. 2018a. The WZIAC gram in Polish. A serial verb construction, or not? STUF -Language Typology and Universals 71(4): 577-629. https://doi.org/10.1515/stuf-2018-0022        [ Links ]

Andrason, A. 2018b. From coordination to verbal serialization - The pójsc (serial verbal) construction in Polish. Research in Language 16(1): 19-46. https://doi.org/10.2478/rela-2018-0001        [ Links ]

Andrason, A. 2019a. A pseudo-coordinated Serial Verb Construction "wziqc i V2" in Polish. Slovo a Slovesnost 80: 163-191.         [ Links ]

Andrason, A. 2019b. A serial verb construction with the verb alàku 'go' in Canaano-Akkadian. Antiguo Oriente 17: 11-38.         [ Links ]

Andrason, A. 2019c. Syntactic gradience and fuzziness - The QWM gram (serial verb construction) in Biblical Hebrew. In G. Kotzé, C. Locatell and J. Messarra (eds.) The Ancient Text and Modern Reader. Leiden: Brill. pp 100-126. https://doi.org/10.1163/9789004402911006        [ Links ]

Andrason, A. and B. Koo. 2020. Verbal serialization in Biblical Aramaic - A dynamic network approach. Altorientalische Forschungen 47(1): 3-33. https://doi.org/10.1515/aofo-2020-0001        [ Links ]

Andrason, A. and J. P. Vita. 2014. From glosses to the linguistic nature of Canaano-Akkadian. Folia Orientalia 51: 155-175.         [ Links ]

Andrason, A. and J. P. Vita. 2018. Amorite: A Northwest Semitic language? Journal of Semitic Studies 63(1): 19-58. https://doi.org/10.1093/jss/fgx035        [ Links ]

Andrason, A. and J. P. Vita. 2020. Serial verb construction in Ugaritic. Aula Orientalis 38(1): 5-33.         [ Links ]

Bisang, W. 2009. Serial verb constructions. Language and Linguistics Compass 3(3): 792-814. https://doi.org/10.1111/j.1749-818x.2009.00128.x        [ Links ]

Bond, O. 2019. Canonical typology. In J. Audring and F. Masini (eds.) The Oxford Handbook of Morphological Theory. Oxford: Oxford University Press. pp. 409-431. https://doi.org/10.1093/oxfordhb/9780199668984.013.26        [ Links ]

Brown, D. and M. Chumakina. 2013. What there might be and what there is: An introduction to canonical typology. In D. Brown, M. Chumakina and G. Corbett (eds.) Canonical Morphology and Syntax. Oxford: Oxford University Press. pp. 1-19. https://doi.org/10.1093/acprof:oso/9780199604326.003.0001        [ Links ]

Buccellati, G. 1997. Akkadian and Amorite phonology. In A. S. Kaye (ed.) Phonologies of Asia and Africa (Including the Caucasus). Vol. 1. Winona Lake: Eisenbrauns. pp. 3-38.         [ Links ]

CAD G = Gelb, I., T. Jacobsen, B. Landsberger and A. L. Oppenheim (eds.). 1956. The Assyrian Dictionary of the Oriental Institute of University of Chicago. Vol 5. Chicago/Glückstadt: The Oriental Institute and J.J. Augustin Verlagsbuchhandlung.         [ Links ]

CAD T = Biggs, R., J. Brinkman, M. Civil, W. Faber, I. Gelb, A. L. Oppenheim, E. Reiner, M. Roth and M. Stolper (eds.). 2006. The Assyrian Dictionary of the Oriental Institute of University of Chicago. Vol 18. Chicago: The Oriental Institute.         [ Links ]

Chrzanowski, J. 2011. Verbal Hendiadys Revisited: Grammaticalization and Auxiliation in Biblical Hebrew Verbs. PhD dissertation, The Catholic University of America.         [ Links ]

Croft, W. 2003. Typology and Universals. Cambridge: Cambridge University Press. https://doi.org/10.1017/cbo9780511840579        [ Links ]

Crowley, T. 2002. Serial Verb in Oceanic: A Descriptive Typology. Oxford: Oxford University Press.         [ Links ]

Dellaire, H. 2014. The Syntax of Volitives in Biblical Hebrew and Amarna Canaanite Prose. Winona Lake: Eisenbrauns.         [ Links ]

De Vos, M. 2005. The Syntax of Pseudo-coordination in English and Afrikaans. Utrecht: LOT.         [ Links ]

Dixon, R. M. W. 2006. Serial verb constructions: Conspectus and coda. In A. Aikhenvald and R. M. W. Dixon (eds.) Serial Verb Constructions: A Cross-linguistic Typology. Oxford: Oxford University Press. pp. 338-350.         [ Links ]

Dobbs-Allsopp, F. W. 1995. Ingressive qwm in Biblical Hebrew. Zeitschrift für Althebraistik 8: 31-55.         [ Links ]

Durand, J-M. 2012. Réflexions sur un fantôme linguistique. In C. Mittermayer and S. Ecklin (eds) Altorientalische Studien zu Ehren von Pascal Attinger. Fribourg: Academic Press. pp. 165-191.         [ Links ]

Edzard, L. 2011. Biblical Hebrew. In S. Weninger (ed.) The Semitic Languages. An International Handbook. Berlin: Mouton De Gruyter. pp. 480-514. https://doi.org/10.1515/9783110251586.480        [ Links ]

Eskhult, M. 1998. The verb sbb as a marker of inception in Biblical Hebrew. Orientalia Suecana 47: 21-26.         [ Links ]

Evans, V. and M. Green. 2006. Cognitive Linguistics: An Introduction. Edinburgh: Edinburgh University Press.         [ Links ]

Fales, F. M. 2011. Old Aramaic. In S. Weninger (ed.) The Semitic Languages. An International Handbook. Berlin: De Gruyter. pp. 555-573. https://doi.org/10.1515/9783110251586.555        [ Links ]

Folmer, M. L. 2012. Old and Imperial Aramaic. In H. Gzella (ed.) Languages from the World of the Bible. Berlin: De Gruyter. pp. 128-159. https://doi.org/10.1515/9781934078631.128        [ Links ]

Gesenius 1910 = Gesenius, W., E. Kautzsch and A.E. Cowley. 1910. Gesenius' Hebrew Grammar. Oxford: Clarendon.         [ Links ]

Georgakopoulos, T and S. Polis. 2018. The semantic map model: State of the art and future avenues for linguistic research. Language and Linguistic Compass 12(2): 1-33. https://doi.org/10.1111/lnc3.12270        [ Links ]

Gisborne, N. and A. Patten. 2011. Construction grammar and grammaticalization. In D. Heine and H. Narrog (eds.) The Oxford Handbook of Grammaticalization. Oxford: Oxford University Press. pp. 92-104. https://doi.org/10.1093/oxfordhb/9780199586783.013.0008        [ Links ]

Giusti, G., V. Di Caro and D. Ross. 2022. Pseudo-coordination and multiple agreement constructions. A overview. In G. Giusti, V. Di Caro and D. Ross (eds.) Pseudo-Coordination and Multiple Agreement Constructions. Amsterdam: John Benjamins. pp. 1-32.         [ Links ]

Gzella, H. 2004. Tempus, Aspekt und Modalitat im Reichsaramaischen. Wiesbaden: Harrassowitz Verlag.         [ Links ]

Gzella, H. 2011a. Northwest Semitic in general. In S. Weninger (ed.) The Semitic Languages. An International Handbook. Berlin: Mouton De Gruyter. pp. 425-451. https://doi.org/10.1515/9783110251586.425        [ Links ]

Gzella, H. 2011b. Imperial Aramaic. In S. Weninger (ed.) The Semitic Languages. An International Handbook. Berlin: Mouton De Gruyter. pp. 574-586. https://doi.org/10.1515/9783110251586.574        [ Links ]

Hamawand, Z. 2016. Semantics. A Cognitive Account of Linguistic Meaning. Sheffield: Equinox.         [ Links ]

Haspelmath, M. 2003. The geometry of grammatical meaning: Semantic maps and cross-linguistic comparison. In M. Tomasello (ed.) The New Psychology of Language. Mahwah: Lawrence Erlbaum. pp. 211-242. https://doi.org/10.4324/9781410606921-11        [ Links ]

Hasselbach, R. and J. Huehnergard. 2007. Northwest Semitic languages. In K. Versteegh et al. (ed.) Encyclopedia of Arabic Language and Linguistics. Leiden: Brill. pp. 408-442. https://doi.org/10.1163/1570-6699ealleallcomvol30234        [ Links ]

Heine B. 1997. Cognitive Foundations of Grammar. Oxford: Oxford University Press.         [ Links ]

Heine B., U. Claudi and F. Hünnemeyer. 1991. Grammaticalization. A Conceptual Framework. Chicago: The University of Chicago Press.         [ Links ]

Hopper, P. and E. C. Traugott. 2003. Grammaticalization. Cambridge: Cambridge University Press. https://doi.org/10.1017/cbo9781139165525        [ Links ]

Hornkohl , A. 2014. Ancient Hebrew Periodization and the Language of the Book of Jeremiah. Leiden: Brill. https://doi.org/10.1163/9789004269651        [ Links ]

Huehnergard, J. 1992. Languages. In D. N. Freedman (ed.) The Anchor Bible Dictionary. Vol. 4. New York: Doubleday. pp. 155-170. https://doi.org/10.5040/9780300261905-175        [ Links ]

Huehnergard, J. 2008. Languages of the Ancient Near East. In K. D. Sakenfeld (ed.) The New Interpreter's Dictionary of the Bible. Nashville: Abingdon. pp. 576-580        [ Links ]

Huffmon, H. B. 1965. Amorite Personal Names in the Mari Texts. A Structural and Lexical Study. Baltimore: John Hopkins Press. https://doi.org/10.2307/3264256        [ Links ]

Izre'el, S. 2005. Canaano-Akkadian. München: Lincom Europa.         [ Links ]

Izre'el, S. 2012. Canaano-Akkadian: Linguistics and sociolinguistics. In R. Hasselbach and N. Pat-El (eds.) Language and Nature: Papers Presented to John Huehnergard on the Occasion of his 60th Birthday. Chicago: The Oriental Institute of the University of Chicago. pp. 171218.         [ Links ]

Janda, L. 2015. Cognitive Linguistics in the Year 2015. Cognitive Semantics 1: 131-154. https://doi.org/10.1163/23526416-00101005        [ Links ]

Johannessen, J. B. 1998. Coordination. Oxford: Oxford University Press.         [ Links ]

Kaniewski, D., E. Van Campo, K. Van Lerberghe, T. Boiy, K. Vansteenhuyse, et al. 2011. The Sea Peoples, from cuneiform tablets to carbon dating. PLoS ONE 6(6): e20232. https://doi.org/10.1371/journal.pone.0020232        [ Links ]

Kaufman, S. 2005. Aramaic. In R. Hetzron (ed.) The Semitic Languages. London: Routledge. pp. 114-130.         [ Links ]

Knapp, B. A. and S. W. Manning. 2016. Crisis in context: The end of the Late Bronze Age in the Eastern Mediterranean. American Journal of Archaeology 120: 99149. https://doi.org/10.3764/aja.120.1.0099        [ Links ]

Knudsen, E. E. 1991. Amorite grammar. A comparative statement. In A. S. Kaye (ed.) Semitic Studies in Honor of Wolf Leslau. Vol. 1. Wiesbaden: Harrassowitz Verlag. pp. 866-885.         [ Links ]

Knudsen, E. E. 2004. Amorite vocabulary. In J. G. Dercksen (ed.) Assyria and Beyond. Studies Presented to Mogens Trolle Larsen. Leiden: Nederlands Instituut voor het Nabije Osten. pp. 319-331.         [ Links ]

KTU³ 2013 = Dietrich, M., O. Loretz and J. Sanmartín. 2013, Die keilalphabetischen Texte aus Ugarit, Ras Ibn Hani und anderen Orten. 3^rd edition. Münster: Ugarit-Verlag.         [ Links ]

Kvist Darnell, U. 2008. Pseudosamordningar i Svenska: sarskilt sádana medverben sitta, ligga och stá. PhD dissertation, Stockholms Universitet.         [ Links ]

Lambdin, T. O. 1971. Introduction to Biblical Hebrew. New York: Charles Scribner's Sons.         [ Links ]

Lipinski, E. 2001. Semitic Languages. Outline of a Comparative Grammar. Leuven: Peeters.         [ Links ]

L0drup, H. 2002. The syntactic structures of Norwegian pseudocoordinations. Studia Linguistica 56(2): 121-143. https://doi.org/10.1111/1467-9582.00090        [ Links ]

Moran, W. L. 1992. The Amarna Letters. Baltimore: The Johns Hopkins University Press.         [ Links ]

Muysken, P. and T. Veenstra. 1994. Serial verbs. In J. Arends, P. Muysken and N. Smith (eds.) Pidgins and Creoles. An introduction. Amsterdam: John Benjamins. pp. 289-301. https://doi.org/10.1075/cll.15.30muy        [ Links ]

Pardee, D. 2011. Ugaritic. In S. Weninger (ed.), The Semitic Languages. An International Handbook. Berlin: Mouton De Gruyter. pp. 460-471.         [ Links ]

Rainey, A. 1996a. Canaanite in the Amarna Tablets. Vol. 2: Morphosyntactic Analysis of the Verbal System. Leiden: Brill.         [ Links ]

Rainey, A. 1996b. Canaanite in the Amarna Tablets. Vol. 3: Morphosyntactic Analysis of the Particles and Adverbs. Leiden: Brill.         [ Links ]

Rainey, A. 2015. The El-Amarna Correspondence. A New Edition of the Cuneiform Letters from the Site of El-Amarna based on Collations of all Extant Tablets. Vol. 1. Leiden: Brill.         [ Links ]

Rendsburg, G. A. 1997. Ancient Hebrew phonology. In A. S. Kaye (ed.) Phonologies of Asia and Africa. Winona Lake: Eisenbrauns. pp. 65-84.         [ Links ]

Round, E. and G. Corbett. 2020. Comparability and measurement in typological science: The bright future for linguistics. Linguistic Typology 24(3): 489-525. https://doi.org/10.1515/lingty-2020-2060        [ Links ]

Röllig, W. 2011. Phoenician and Punic. In S. Weninger (ed.) The Semitic Languages. An International Handbook. Berlin: De Gruyter. pp. 472-479.         [ Links ]

Rubin, A. 2010. A Brief Introduction to the Semitic Languages. Piscataway: Gorgias Press. https://doi.org/10.31826/9781463224936        [ Links ]

Sanmartín J. 2014. 1st "Altsyrisch" eine Sprache? Und wenn ja, wie viele?. Ugarit-Forschungen 45: 487-507.         [ Links ]

Steiner, R. 1997. Ancient Hebrew. In R. Hetzron (ed.), The Semitic Languages. London: Routledge. pp. 145-173.         [ Links ]

Streck, M. P. 2011. Amorite. In S. Weninger (ed.) The Semitic Languages. An International Handbook. Berlin: De Gruyter. pp. 452-459. https://doi.org/10.1515/9783110251586.452        [ Links ]

Streck, M. P. 2013. Remarks on two recent studies on Amorite. Ugarit-Forschungen 44: 309327.         [ Links ]

Traugott, E. C. 2011. Grammaticalization and mechanisms of change. In D. Heine and H. Narrog (eds.) The Oxford Handbook of Grammaticalization. Oxford: Oxford University Press. pp. 19-30. https://doi.org/10.1093/oxfordhb/9780199586783.013.0002        [ Links ]

Tropper, J. 2000. Das amurritische Onomastikon der altbabylonischen Zeit. Ugarit-Forschungen 32: 733-744.         [ Links ]

Tropper, J. 2012. Ugaritische Grammatik. 2^nd edition. Münster: Ugarit-Verlag.         [ Links ]

Tropper, J. and J. P. Vita 2010. Das Kanaano-Akkadische der Amarnazeit. Munster: Ugarit-Verlag.         [ Links ]

Tropper, J. and J. P. Vita 2020. Lehrbuch der ugaritischen Sprache. Münster: Zaphon.         [ Links ]

Waltisberg, M. 2011. The case functions in Amorite - A re-evaluation. Journal of Semitic Studies 56: 19-35. https://doi.org/10.1093/jss/fgq057        [ Links ]

Wilson-Wright, A. 2019. The Canaanite languages. In J. Huehnergard and N. Pat-El (eds.) The Semitic Languages. London: Routledge. pp. 509-532. https://doi.org/10.4324/9780429025563-20        [ Links ]

Yuasa, E. and Sadock, J. 2002. Pseudo-subordination: a mismatch between syntax and semantics. Journal of Linguistics 38(1): 87-111. https://doi.org/10.1017/s0022226701001256        [ Links ]

Zadok, R. 1993. On Amorite material from Mesopotamia. In M. E. Cohen, D. C. Snell and D. B. Weisberg (eds) The Tablet and the Scroll. Near Eastern Studies in Honor of William W. Hallo. Bethesda: CDL Press. pp. 315-333.         [ Links ]

 

 

1 Often scholars propose that Amorite was the oldest (attested) member of the NWS branch (see Huffmon 1965, Knudsen 1991; 2004, Zadok 1993, Tropper 2000; Hasselbach & Huehnergard 2007, Huehnergard 2008; Streck 2011; 2013; Gzella 2011a; Waltisberg 2011). Others view this classification as problematic (Huehnergard 1992; Buccellati 1997; Durand 2012; Sanmartín 2014). In my study dedicated to the genetic classification of Amorite, I conclude: "[T]he evidence available currently does not enable us to equivocally determine the position of Amorite within the Semitic family. The data can link Amorite to the Central, Northwest and East Semitic branches. Additionally, various pieces of evidence are missing or their interpretation is uncertain. As a result, the definite answer to the question of the genetic filiation of Amorite presently seems to remain beyond the reach of Semitic linguistics" (Andrason & Vita 2018:55). It should also be noted that there are other old NWS languages: Old Aramaic (from the 9 until the 7 c. BCE; Fales 2011:555) and Phoenician (from the 10 c. BCE until the 1 c. CE; Röllig 2011:473). These two languages are not part of the present study.
2 Henceforth in this section, I will only use the singular 'a construction' to render the text less arduous and verbose. However, in all cases where I mention 'a construction', I also imply 'constructions' in plural.
3 I will use the term 'evolutionary path' rather than 'grammaticalization path' because the concept of grammaticalization will be reserved to a particular type of the development, namely, an increase in the productivity, entrenchment, and generalization of a construction (see further below in this section).
4 Although the concepts of a prototype and canon are sometimes treated as essentially distinct (see Brown & Chumakina 2013), I view them as highly similar and compatible. The former merely emphasizes the cognitive plausibility of an ideal, whereas the latter stresses its rationality or logical construal.
5 Asymmetrical SVCs, in which the 'minor' verb modifies the event expressed by the 'major' verb constitute one of the two main types of SVCs - the other one being symmetrical SVCs (see Aikhenvald 2006; 2018). Given that only asymmetrical SVCs are relevant for my study, the subsequent discussion is limited to this class of SVCs.
6 For a review of the (more or less) common violations of all the prototypical features mentioned in this paragraph, consult Aikhenvald (2006; 2018; forthcoming).
7 Such cohesive structures may be of two types. First, an SVC may be reduced to a synthetic verbal gram with minor verbs acting as a clitic or affix. Second, the minor verb may be converted into a particle. In both cases, the biverbal structure becomes mono-verbal, mono-clausal and mono-predicative: the minor verb loses its verbal properties, and the major verb develops into the main verb.
8 It should be noted that my study involves corpora-lects or languages that emerge from determined sets of texts, in some cases, spanning several centuries (see especially Biblical Hebrew and Biblical Aramaic in sections 3.3 and 3.4). Therefore, although I treat them as single language systems, I am fully aware of their internal complexity and diversity. This, however, does not pose any problem given the dynamic approach in terms of the vectored maps that I use.
9 As mentioned in footnote 1, a number of authors consider Amorite to be the oldest attested NWS language. Even if this were unproblematic (which is not the case), Canaano-Akkadian would still be the first NWS language that offers corpora with real texts (see, however, the peculiar nature of these texts as discussed further below in this paragraph). Thus, the Canaano-Akkadian corpora contrast with the Amorite corpus, which almost entirely consists of proper names (Andrason & Vita 2018:19; the remaining Amorite material includes loanwords found in Akkadian and Sumerian and other non-Akkadian elements present in the Akkadian variety of Mari (ibid. 20)).
10 To my knowledge, no other SVCs have been reported in scholarship as possibly serializing in Canaano-Akkadian. It should, however, be noted that the scarcity of Canaano-Akkadian SVCs may, at least partially, also be due to the limited size of the available corpus. In other words, the actual status of verbal serialization in the language - especially, the Canaanite variety underlying the language of the El-Amarna letters - is unknown.
11 The clause-combining marker u may also be used in other connective senses, thus heading result, cause, purpose, temporal, adversative, and relative clauses (see Rainey 1996b:97-107).
12 That is, as two independent events.
13 Canaano-Akkadian also attests to genuine examples of bi-verbal coordination with the clause-combining maker (conjunctive coordinator) u 'and'. In all such sequences, the construction does not instantiate an SVC but rather a bi-clausal, bi-predicative, and bi-event structure, in many cases of a conjunctive type (see Rainey 1996b; Izre'el 2005; Tropper & Vita 2010).
14 This example with ngr can also be analyzed as a symmetrical SVC (Andrason & Vita 2020:26).
15 These different semantic domains may be mixed and cooccur in a single usage (Andrason & Vita 2020:27).
16 The canonical class of SVCs seems to be the most productive in Ugaritic by attesting to "the greatest semantic and morphological variety of verbs" (Andrason & Vita 2020:27).
17 This non-canonical type of SVC is also less diversified morphologically and semantically.
18 Nevertheless, this example can also be interpreted as an SVC (Andrason & Vita 2020:21).
19 The term ' serial verb construction' has only been used recently in BH scholarship (see Dobbs-Allsopp 1995:37 for certain linker-free sequences and, especially, Andrason 2019c). Traditionally, serializing patterns have rather been referred to by terms such as 'hendiadys' (Lambdin 1971:238; see also Dobbs-Allsopp 1995:37) or 'auxiliarization' (Chrzanowski 2011).
20 The prosodic properties discussed here following Andrason (2019c) are derived from Masoretic cantillation accents. I interpret conjunctive and disjunctive accents as signs of "phonological unity" and "phonological separation", respectively (Andrason & Koo 2020:11). Although Masoretic accents are not genuine orthographic or intonation-related symbols, they "constitute the best available tool to hypothesize the phonological unity or disjunction of the SVCs" (ibid.) for the biblical languages, i.e., Biblical Hebrew and Biblical Aramaic (see section 3.4 below).
21 Minor fragments found in the books of Jeremiah (10.11) and Genesis (31.47).
22 This opinion is, however, not universal. For instance, Kaufman (2005:115-116) considers the Aramaic of Ezra as related to Imperial Aramaic, while that of Daniel is classified as a Middle Aramaic variety.
23 These verbs also exhibit non-canonical profiles.
24 There are more than 250 cases of SVCs for the four minor verbs for which such frequencies exist (see section 3.3 above).
25 This somehow higher position of Ugaritic may be related to the presence of accentuation signs in Biblical Hebrew and Biblical Aramaic, which render several examples less canonical or nearly canonical. The lower canonicity level in Biblical Aramaic in comparison to Biblical Hebrew may also be related to the absence of w-driven tenses (i.e., wayyiqtol and weqatal) in Biblical Aramaic, which in Biblical Hebrew are analyzed as lacking an overt linker w (cf. Andrason & Vita 2020; Andrason & Koo 2020).