1 Introduction

In current linguistic theory, the concept of agreement is typically conceptualized as an asymmetrical relationship in which the presence of some feature (notably person, number, or gender) in one lexical unit (“the controller”) triggers the presence of the same feature in another lexical unit (“the target”) within a particular syntactic domain (Corbett 2006; Haig & Forker 2018; Matasović 2018; Melnik 2020). When the relevant domain is that of the clause, agreement is often dubbed verbal agreement, referring to the pattern in which “the verb is – under syntactically or phonologically specifiable conditions – obligatorily modified by a morpheme (affix or clitic) expressing the agreeing category” (Matasović 2018: 20–21). The notion of verbal agreement is typically evoked in relation to the inflectional marking of person-number-gender (henceforth: PNG) in verbal paradigms; thus, person-inflectional markers are termed “agreement markers”. This notion is heavily entrenched in general linguistics, and appears to be taken for granted as an axiom; accordingly, its adequacy is rarely questioned (but see Cysouw 2011). Israeli Hebrew (IH) is no exception to this tendency, and the concept of verbal agreement is often applied to person-inflectional affixes in two paradigms, namely suffix-conjugated verbs and prefix-conjugated verbs.¹ In this paper, I discuss the notion verbal agreement critically with the aim of demonstrating that this notion is unwarranted – both in terms of the observed data and its historiographical origin – using IH as a case study.

Table 1 illustrates the two person-inflected paradigms in IH (adapted from Izre’el 2012: 219).

A survey of the literature on Hebrew reveals that there is no consensus on the theoretical interpretation of the inflectional affixes of suffix- and prefix-conjugated verbs. The prevalent view of these affixes includes two related claims: (1) The inflectional affixes are agreement markers; that is, elements that are dependent on, or controlled by, an external subject; and (2) the inflectional affixes of 1^st and 2^nd person verbs are marked for person, number, and gender, while the inflectional affixes of 3^rd person verbs are marked for number and gender, but not for person (Glinert 1989; Berman 1990; Vainikka & Levy 1999; Ariel 1990; 1998; 2000; Levy & Vainikka 2000; Gutman 2004; Melnik 2007). These claims are illustrated by the following quotes:

• A.  “agreement markers of person […] are confined to 1st and 2nd person; 3rd-person verbs require an independent subject pronoun […]. The 3^rd person forms in (9c) [=gamar ‘he finished’, gamra ‘she finished’, gamru ‘they finished’, yigmor ‘he’ll finish’, tigmor ‘she’ll finish’, yigmeru ‘they’ll finish’; L.S.], although zero-marked for person, are still morphologically distinct from 1^st and 2^nd person-marked forms in past and future” (Berman 1990: 1142)²

• B.  “Though they [=third-person past/future inflections; L.S.] are not morphologically marked for person, their zero marking is a clear identifier of the grammatical person they represent.” (Ariel 1990: 116)

• C.  “Significantly, 3^rd person inflection only marks gender and number. Person is zero-marked.” (Glinert 1989: 53, fn. 2)

• D.  “In the third person [of the future tense; L.S.], as in the past tense, the independent pronouns are obligatory. The verb form of the third person does not include person features. It includes the features of number (singular/plural) and gender (masculine/feminine) but not person” (Coffin & Bolozky 2005: 166)

These quotes are revealing in two respects. First, they hint at a possible motivation for regarding 3^rd person verbs as having a zero person marking, namely the tendency of 3^rd person verbs to require a free subject pronoun, in contrast to 1^st and 2^nd person verbs, with which free subject pronouns are optional (quote A and D). It appears that the apparent requirement of a free subject pronoun has been interpreted as evidence that 3^rd person verbs do not include person marking, only gender and number ones. It is noteworthy that this assumption was further disseminated beyond Hebrew scholarship to general linguistics – WALS studies, for example, categorize IH as a language with “zero in all 3sg forms” (Siewierska 2013). In addition, these quotes make it clear that it is uncontroversial that 3^rd person verbs are morphologically distinct, and that the grammatical persons represented by 3^rd person verbs, i.e., 3sgm, 3sgf, 3pl, are identifiable (quote A and B). What is not clear, however, is how a verb form that is supposedly unmarked for person be unambiguously identified as a specific grammatical person.

A notable exception to this tradition is represented in the works of scholars who follow the theory elaborated by leading Semitic scholar G. Goldenberg (1998, 2013 and elsewhere), such as Bar (2007), Izre’el (2012; 2018; 2020), Zewi (2013), Cohen (2017), and Halevy (2016; 2020). The approach to linguistic description of IH adopted by these scholars is informed by the fact that IH is a Semitic language, with considerably different grammatical features from those of the languages that served as the basis for the general linguistic theory, most frequently Indo-European languages. Scholars operating in this approach regard the suffix- and prefix-conjugated verb as dimorphemic predicational complex consisting of a pronominal subject morpheme and a verbal predicative stem which are linked up into a predicative bond (“nexus”). Crucially, they do not subscribe to the two claims mentioned above, and regard each of the inflectional affixes as pronominal elements, ones that have referential capacities. This view is concisely summarized by Halevy (2020: 544):

“The synthetic inflection of Hebrew means that person, number, and gender affixes representing the subject morpheme are always incorporated in finite verb form, with person marking confined to past and future tense, so that the inflectional affixes of the predicate do not serve merely as agreement markers but rather, they denote the grammatical subject”

In this paper, I will substantiate the latter approach via a critical discussion of the two abovementioned claims, i.e., the interpretation of verbal affixes as agreement markers and the interpretation of 3^rd person verbs as having no person marking. Consequently, the main claim of the paper will be that the verbal inflectional affixes in IH are best regarded as “bound pronouns” (Kibrik 2011). To substantiate this claim, I will first argue for the general inadequacy of the notion of “verbal agreement” based on the following argument: this notion does not conform to the cross-linguistically dominant pattern of pronominal subject expression, having originated in ancient Greek philosophy and logic, and then being transferred to the grammatical traditions of a few highly influential European languages (§2). I will then argue for the inadequacy of the notion verbal agreement in IH based on various types of language-internal evidence (§3).

2 Revisiting “verbal agreement”

The prevailing form-to-form approach to agreement, according to which agreement is a morphosyntactic mechanism of copying features from one linguistic element (controller) to another one (target) has been critiqued in the past by prominent scholars (e.g., Barlow 1999; Mithun 2003; Langacker 2008; Haspelmath 2013; Croft 2001; 2013; Kibrik 2019). These scholars pointed out that a formal approach to agreement is inadequate, as it encounters massive difficulties when confronted with actual data – for example a formal controller may be missing or underspecified, or the controller and the target exhibit a mismatch in agreement features.

Acknowledging the problems associated with the traditional notion of agreement, various types of modifications to that notion have been suggested. Barlow (1999), for example, suggested to reinterpret agreement as a referential morpheme that provides information about the nature of referents, either in combination with some prior referential expression, or serving as the sole representation of the said referent. Kibrik (2019), too, opposes the traditional understanding of agreement as an asymmetric form-to-form relationship, and proposes instead to reinterpret it in a parallel manner. According to Kibrik, agreement should be viewed as a cognition-to-form mapping, according to which agreement features, such as person, number, and gender, are associated with referents in the cognitive representation. As the speaker produces a clause containing the referent, relevant features are mapped onto language-specific sites where these features must be marked in the given language. Apparent agreement between various sites is considered a side effect of mappings from the same cognitive source. Consequently, in this framework, person inflectional affixes are not regarded as agreement markers, but as bound pronouns, being subsumed under the label “reduced referential devices”, along with free pronouns, and zero reference (Kibrik 2011: 73–77; 2013: 228–231).

Reconceptualizing person inflectional affixes as “bound pronouns” (Kibrik 2011; 2019), “incorporated pronouns” (Bresnan & Mchombo 1987), or “pronominal affixes” (Mithun 2003) instead of “agreement markers” highlights the fact that more often than not such affixes function referentially in the languages of the world, serving as bound analogues to the free pronouns of the well-known European languages, notably English, German, and French.³ Significantly, the descriptive traditions of precisely these languages have employed the notion of agreement for the relationship between the (lexical/pronominal) subject NP and the verbal inflectional affixes as early as the 16^th century, but especially so from the 20^th century, starting with Bloomfield (1933) (Cysouw 2011: 155; Kibrik 2011: 216; Haspelmath 2013: 203).

On the face of it, the notion of agreement appears as descriptively suited to the description of the verbal inflection of the influential European languages, such as English, German, French. That is because the typical clause structure in these languages minimally involves a verbal form with an explicit subject NP in the form of a lexical NP or a free subject pronoun. Put differently, verbal forms almost never occur without an explicit subject NP, and by implication, verbal forms are considered as insufficient for the constitution of a grammatical clause. However, this pattern is in no way representative of the cross-linguistic variation in pronominal subject expression, being highly peripheral among the world’s languages (Siewierska 1999: 238; Kibrik 2011: 216; Dryer 2013; Matasović 2018: 115). Thus, in the majority of the world’s languages, pronominal subject expression is typically achieved through bound pronouns, and verbs/predicates can constitute grammatical clauses in themselves (Kibrik 2011; Dryer 2013; Haspelmath 2018). Nevertheless, and quite paradoxically, the “highly exotic” (Kibrik 2011: 220) pattern found in English, German and French has been adopted as the model of canonical agreement by many modern linguistic theories.⁴

As mentioned, this paradox has been evoked by several functionally- and typologically-oriented scholars as evidence against the ‘agreement’ interpretation of verbal inflectional affixes. However, considering that many grammatical notions and concepts were initially conceived in the grammatical tradition of Classical Greek, and subsequently mediated to the descriptive traditions of modern European languages by Classical Latin, there appears to be another reason for reconsidering the ‘agreement’ interpretation of verbal inflectional affixes. It appears that similarly to many other concepts in Western linguistic research (Diver, Davis & Reid 2012; Izre’el 2018; Matthews 2019), the ‘agreement’ interpretation of verbal inflectional affixes is rooted in the grammatical tradition of Classical Greek. Notably, however, this grammatical tradition did not develop out of the study of the Greek language in its own right, but rather was rooted in the philosophical, ontological and logical traditions of ancient Greece, notably those founded by Plato and Aristotle in the 5^th–4^th centuries BC, for whom the study of language was a tool to be used in pursuing higher ends – development and understanding of dialectic, logic, rhetoric, and poetry (Law 2003; Allan 2004). That is why it should not be taken for granted that concepts originating in this tradition are descriptively valid for languages other than Greek, and possibly not even to Greek itself.

Against this backdrop, I would like to argue that the ‘agreement’ interpretation of the verbal inflectional affixes may have its origins in the Greek grammarians’ approach to the structure of the sentence, and more specifically in the assumption that a verbal form alone cannot comprise a complete sentence. Starting from the first study of syntax by Apollonius Dyscolus (2nd century CE), the object of the syntactic inquiry was the ‘complete sentence’, considered as a congruent arrangement of different parts of speech (Lallot 2015: 853). Crucially, the basic syntactic arrangement that accomplishes completion was postulated as the combination of two types of words, a noun and a verb (Lallot 2015: 859; Ildefonse & Lallot 2017: 82; Matthews 2019: 79–80). This view is demonstrated in the following excerpts from the Syntax of Apollonius Dyscolus (1.14–15; Householder 1981: 23–25):

§14 “The ordering [of the parts of speech, L.S] is a reflection of the complete sentence, quite properly placing the noun first, and after it the verb, since any sentence which lacks (either of) these is not complete […] It’s easy to test this with a sentence-structure containing all the parts of speech; if either noun or verb is deleted, then the sense of the sentence is indeterminate, but if any of the others is removed, there’s no defect in the sentence at all.”

§15 “I’m not here claiming that you can’t have a complete sentence with pronominal subjects, such as egō peripatō (I’m walking) or su peripateis (You’re walking)…. For then, too, completeness is achieved, when a pronoun is used in place of a noun, which gives an essentially identical structure.”

We can see that in §14, Apollonius argues that nouns and verbs take the first and second places in the logical hierarchy of parts of speech since they are the only parts of speech that are indispensible for a complete sentence. He then illustrates this through a deletion test that begins with a sentence that includes all parts of speech: ho autos anthrōpos olisthōsas sōmeron katepesen ‘the same man slipping today fell down’. Apollonius shows that deleting the noun or the verb would result in defective sentences, whereas deleting all of the words but the noun and the verb would result in the minimally complete sentence anthrōpos epesen ‘(a) man fell’. Evidently, the noun+verb syntactic configuration is regarded as the primary syntactic structure that achieves completeness. This is also made clear in §15, in which Apollonius clarifies that the minimal sentence can also consist of a pronoun+verb configuration, since it essentially preserves the bipartite structure of the noun+verb configuration by means of replacing the noun with a pronoun.

Paradoxically, however, verbs in Classical Greek are in fact capable of forming a sentence on their own. In contrast to English, German and French – languages whose descriptive traditions are rooted in the grammatical tradition of classical Greek – pronominal subject referents in classical Greek verbal clauses are normally expressed only by the personal ending of the verb, with free subject pronouns normally not present (van Emde Boas et al. 2019: 310). In other words, verbs in Classical Greek require a lexical NP subject only for unactivated subject referents, whereas with activated subject referents a verb alone is sufficient for constituting a sentence. The discrepancy between the inability – as postulated by Appolonius – of verbal forms to form sentences on their own, and their ability based on usage pattern of Greek does not go unnoticed by Householder. With regard to the assumption that the sentence anthrōpos epesen ‘(a) man fell’ is a minimally complete sentence, he comments that “in Greek one could also delete the noun, since epesen ‘he fell’ is a perfectly good sentence, though in isolation […] it would be indeterminate as to identity of ‘he’.” Similarly, with respect to §15, Householder comments that “the egō ‘I’ and su ‘you’ are normally present only if emphatic; peripatō is the ordinary way of saying ‘I’m walking’”. Thus, the verb’s potential to form a sentence on its own depends on the type of subject taken as point of departure. Starting from sentences with pronominal subjects would lead to the conclusion that verbs are syntactically self-sufficient; whereas starting from sentences with lexical subjects was bound to lead to the opposite conclusion. The point of departure in Appolonius’ syntax appears to be the latter, according to which the noun+verb sentence – i.e., a noun in the nominative followed by a verb in the 3^rd person, such as antrōphos epesen ‘(a) man fell’ – is the primary syntactic configuration.

It would appear that the primacy of such a configuration goes back to the philosophical, ontological and logical traditions of ancient Greece, notably those founded by Plato and Aristotle (Izre’el 2018). For these philosophers, such a syntactic configuration was regarded as natural based on the presupposed priority of substance (corresponding to the noun) over attributes, qualities, states, relations, and events (corresponding to the verb) because the latter can only be predicated of something (Allan 2007: 44; Schmidhauser 2010: 501–502; Lallot 2015: 860; Ildefonse & Lallot 2017: 74–77). Crucially, this syntactic configuration represents a particular type of sentence, one that was suited to philosophical and logical inquiries. Nonetheless, the assumption that a verb alone cannot form a complete sentence, and the consequent requirement to have at least two components in a simple sentence, a (pro)noun and a verb, were carried on to be a basic requirement in the Western study of syntax ever since (Seuren 2013; Izre’el 2018). As a consequence of such an approach to sentence structure, and considering that it was the ‘word’ – and not the ‘morpheme’ – that was perceived as the smallest unit that had a meaning (Matthews 2019: 47–48), it is perhaps not surprising that the verbal inflectional affixes were not considered as referential elements that can fulfill the subject role by themselves. Instead, the meaning components of person and number were regarded as properties that are shared by verbs and (pro)nouns, and that are supposed to be congruent in a paricular sentence (Matthews 2019), and therefore the bearers of these properties, verbal inflectional affixes, came to be regarded as agreement markers.

3 Revisiting “verbal agreement” in IH

Having argued for the inadequacy of applying the notion “agreement” to PNG inflectional markers in verbal paradigms in the previous section, I now turn to IH, aiming to reassess the two prevalent claims found in IH scholarship regarding the status of the PNG inflectional affixes of the person-inflected paradigms: (1) the inflectional affixes are agreement markers, i.e., elements that are dependent on, or controlled by, an external subject, themselves being devoid of referential capacity; (2) the inflectional affixes of 1^st and 2^nd person verbs are marked for person, number, and gender, while the inflectional affixes of 3^rd person verbs are marked for number and gender, but not for person (see §1). The discussion of these claims will be done separately for suffix-conjugated verbs (§3.1) and prefix-conjugated verbs (§3.2).

5 Discussion

In the previous section, I proposed a re-assessment of the analysis of verbal agreement inflectional affixes of person-inflected paradigms in IH, according to which the inflectional affixes are agreement markers that are marked either for person, number, and gender (in the 1^st and 2^nd persons), or for number and gender but not for person (in the 3^rd person). Based on previous corpus investigations and on synchronic intra-paradigmatic considerations, I argued that this analysis is unsubstantiated and therefore unwarranted. I suggested instead that all inflected verb forms are predicative complexes consisting of a stem and inflectional affixes, which should be interpreted as referential elements (bound pronouns or person indexes) that are uniformly marked for person. My analysis is summarized in Table 15.

The main arguments in support of my claim are: (1) all verbal forms but two are unambiguous with regard to PNG, and two forms in the prefix-conjugation – Ø~j-STEM and t-STEM – display person syncretism with respect to two PNG options; and (2) all verbal forms but two contain an overt segment that can be assigned the role of marking the relevant PNG combination (“do not posit zero whenever an overt referential device is present that can possibly carry referential function”; Kibrik 2011: 236). The only verbal forms that do not contain any overt segment that can carry the referential function are 3sgm suffix-conjugated verbs (stem-Ø) and 1sg prefix-conjugated verbs (Ø~j-stem) to some extent. Consequently, (morphological) zero should only be assigned to those slots within the respective paradigms (“when no overt referential device is present, posit a free or bound zero in accordance with the language’s dominant pattern”; Kibrik 2011: 236). Crucially, the morphological zero interpretation of the absence of inflection in these two verbal forms is also substantiated by meeting another, stricter criterion for postulating zeroes, according to which a postulated zero sign must have a non-zero sign expressing the same meaning (Haas 1957; Segel 2008). Indeed, 3sgm and 1sg in IH have the following additional overt allomorphs: (1) 3sgm – free (hu), bound (j– in prefix-conjugated verb), or clitic (=(n)o/=av in attributive syntactic positions); and (2) 1sg – free (ani), or clitic (=i/=aj in attributive syntactic positions).

A major objection to my analysis would be that it disregards the degree of gradient transparency of the inflectional affixes, and consequently, that it does not account for the distributional patterns of free subject pronouns with the respective verbal forms that were arguably triggered by the degree of gradient transparency. Notably, Ariel (1990; 1998; 2000; 2007) proposed a classification of person-inflections according to the degree of accessibility they encode. Degree of accessibility, in turn, is (also) a function of the degree of transparency of the respective forms, as summarized in Table 16.¹⁹

As can be observed, the inflectional affixes in Group A are phonologically more salient and informationally more transparent – recapitulating part of the corresponding free pronouns more clearly – than are the corresponding affixes in Group B. The inflectional affixes in Group C bear no trace of the corresponding free pronouns, and thus are the least transparent. Ariel (1990) proposed that this diminishing transparency of person marking from A to C corresponds to an increase in the degree of accessibility coded by the respective affixes; In C there are no accessibility markers whatsoever, while the affixes in A encode a lower degree of accessibility than the B affixes. According to Ariel, this assumed difference in accessibility marking accounts for the distribution of free subject pronouns with A (normally absent) and C (normally present) (see Tables 3 and 7, respectively). The intermediary status of B is reflected in the fact that, in their indicative use (“future tense”), these forms tend to appear with free subject pronouns in informal registers, but not in formal ones. Ariel (2000: 236–237) proposed that this attests to the “vanishing referentiality” of the affixes in B in informal registers, and indicates that these affixes “are hardly felt as referential anymore”, whereas in formal registers, the referentiality of these affixes is retained. According to this account, the affixes in B are register-sensitive, as they function as high-activation referring expressions in formal registers, and as non-referential elements in informal registers (Ariel 2007).

The analysis proposed in the present paper is not in fact incompatible with the view proposed by Ariel (1990; 1998; 2000; 2007) if we allow for a more nuanced and flexible view of the referentiality of the person indexes. Combining Ariel’s tripartite division of the verbal forms based on the degree of gradient transparency of their inflectional affixes with my claim that all verbal forms are marked for person may lead to the conclusion that the various person indexes differ with respect to their degree of referentiality: (1) 1^st and 2^nd person indexes in suffix-conjugation are maximally referential; (2) 1^st and 2^nd person indexes in the prefix-conjugation are moderately referential; and (3) 3^rd person indexes in both conjugations are minimally referential. Specifically, with respect to prefix-conjugated verbs, allowing for a moderate degree of referentiality has the benefit of accounting for the variation in the distribution of free subject pronouns with indicative uses as opposed to modal ones, as well as in formal as opposed to informal registers. Based on this account, the moderate degree of referentiality of 1^st and 2^nd person indexes in the prefix-conjugation is sufficient for the purpose of indicative use in formal registers; therefore, a free pronoun is normally not required (see Tables 8, 9, 10). Conversely, in informal registers, the moderate degree of referentiality is deemed insufficient; therefore, a free subject pronoun is normally required (see Table 7). At the same time, it should be acknowledged that the degree of referentiality of person marking is not the only factor that may exert pressure in favor of requiring free subject pronouns, as such pressure can also be exerted by various combinations of language-internal factors and external-linguistic influences (for example, see Kibrik 2013 for spoken Russian). With regard to IH prefix-conjugation, two additional explanations appear to be relevant.

The first explanation for the emergence of the preference for free subject pronouns with indicative uses pertains to the extant semantic ambiguity between the indicative (“future”) meaning and modal (“imperative”/“cohortative”) meanings (see §3.2) of prefix-conjugated verbs in IH. Specifically, the modal use of 2^nd person prefix-conjugated verbs is fairly well attested in the early decades of the 20^th century, as evidenced by various written sources (Reshef 2015: 288, 328–329). It has recently been suggested that this use does not have its roots in Biblical Hebrew, but is a continuation of a tendency that was already present in the informal Rabbinic Hebrew of the 18^th and 19^th centuries, mediated by factors such as paradigm leveling and a grammaticalization process of subordination, which is speculated to have evolved due to language contact with Yiddish (Ariel 2019). Given that the modal use of 2^nd person prefix-conjugated verbs appears well-established in early Modern Hebrew, it seems plausible to assume that free subject pronouns were felt as necessary to express the indicative (“future”) meaning of prefix-conjugated verbs unequivocally.

Another factor that may have promoted the use of free subject pronouns with indicative prefix-conjugated verbs involves a possible external-linguistic influence. Such an influence is particularly likely in the context of IH – a language that emerged during the first decades of the twentieth century and of which the first speakers were mainly Yiddish- and Russian-speaking immigrants for whom Hebrew was a second language (Izre’el 2003; Spolsky 2014: 255; Doron 2016). It has repeatedly been demonstrated that many structural aspects of IH may be traced to the influence of the contact with the native languages of first-generation Hebrew speakers (Blanc 1965; Doron et al. 2016; 2019; Reshef 2020; Shor et al. 2022). It is also well known that speakers of languages characterized by an extensive use of free pronouns tend to transfer this property of their native languages when speaking a second language in which the “inflection only” pattern prevails. For example, in the context of second language acquisition, it has been shown that patterns of pronominal usage in the target language (the L2) could be transferred from the speaker’s native language (L1). In other words, speakers of languages with an extensive use of free pronouns tend to transfer this property to their target language, essentially overusing free pronouns in their L2. This tendency has been observed in various bilingual populations, including child bilinguals (Hacohen & Schaeffer 2007; Haznedar 2010), adult L2 learners (Liceras Diaz 1999; Li 2014), and heritage speakers (Montrul 2006; Keating et al. 2011). Similarly, pronominal subject expression in creoles and pidgins is affected by the dominant referential pattern in the substrate languages. Creole and pidgin languages in the Atlantic and the Pacific regions tend to have obligatory subject pronouns due to the substrate influence of West African and Austronesian languages in which subject pronouns are normally obligatory. By contrast, creole and pidgin languages in the Indian Ocean and Asia tend to have optional subject pronouns due to the substrate influence of South and Southeast Asian languages in which subject pronouns tend to be optional (Haspelmath and the APiCS Consortium 2013). In line with these findings, the fact that free subject pronouns were the primary means of pronominal subject reference in the native languages of IH speakers of European origin (notably Yiddish and Russian) may have led to an increased use of free subject pronouns with indicatively used prefix-conjugated verbs.²⁰

While the specific details of this process must be confirmed through diachronic corpus-based studies, it would appear that the expression of the future indicative meaning via the construction “free subject pronoun + prefix-conjugated verb” was triggered by a combination of internal factors (the need for the unequivocal expression of the future meaning, coupled with the opaqueness of the person indexes), which was possibly enhanced by external influences, such as the influence of the native languages of the first-generation Hebrew speakers.

6 Conclusion

In this paper, I attempted to re-assess the verbal agreement interpretation of the inflectional affixes of the person-inflected paradigms in IH, according to which the inflectional affixes are agreement markers that are marked either for person, number, and gender (in the 1^st and 2^nd persons), or for number and gender but not for person (in the 3^rd person). Based on previous corpus investigations and synchronic intra-paradigmatic considerations, I argued that this interpretation was unsubstantiated and therefore inadequate, suggesting instead that the inflectional affixes in IH should be treated as referential elements that are uniformly marked for person.

An additional reason for rejecting the notion of verbal agreement was proposed. This notion presupposes the primacy (canonicity/naturalness) of a bipartite clausal format in which the lexical subject NP and the predicate are present and morphologically independent; thus, the inflectional affix is assumed to be positioned in an asymmetrical relationship with the corresponding subject NP with which it is assumed to agree. Consequently, the notion of verbal agreement also presupposes the primacy of morphologically independent person markers over inflectional ones (cf. Cysouw 2003). However, these presuppositions are at odds with cross-linguistic data: With the exception of a few highly influential European languages, such as English, German, and French, free subject pronouns are not normally present in the majority of languages that employ person-inflected predicates; thus, inflectional affixes have no concrete element with which to agree. Crucially, these presuppositions are essentially rooted in the Platonic and Aristotelian logico-philosophical analysis of the proposition into subject (corresponding to the noun) and predicate (corresponding to the verb) which the first Greek grammarians later transferred to the grammatical analysis of sentences (Seuren 2013; Izre’el 2018). The primacy of the bipartite clausal format has a logico-philosophical justification: Substance (corresponding to the noun) is presupposed to be prior to attributes, qualities, states, relations, and events (corresponding to the verb) because the latter can only be predicated of the former. From a usage-based linguistic perspective, however, the primacy of the bipartite clausal format is not justified. The adoption of a logico-philosophical perspective instead of a usage-based linguistic perspective has unfortunately resulted in the theoretical interpretation of verbal inflectional affixes as elements that merely agree with free subjects, rather than as referential elements that can fulfill the subject role by themselves. In this respect, it would be preferable to employ alternative terms and concepts, such as “inflectional person marking” (Cysouw 2003; 2011), “bound pronoun” (Kibrik 2011; 2019), or “person index” (Haspelmath 2013), as these are not biased toward a particular clausal configuration or toward the degree of morphological independence of the person marking form.

Competing interests

The author has no competing interests to declare.