Valency: the intersection of syntax and semantics

Valency seems to be an up-and-coming area in Biblical Hebrew linguistics. I was more or less thrown into the issue through my involvement with the Accordance syntax project (see here and here): as someone who was already obsessed by the verbal system, it made sense to task me with overseeing the valency analysis for the project. This task has mainly entailed addressing the necessity of deciding between verbal complements and adjuncts in the database tagging, which in turn has led to a developing valency dictionary that will contribute to Hebrew lexicography by supplementing the current lexica with a specific focus on verbal valency.
But as I said, it seems that this is a burgeoning field. I had the pleasure of being invited to a workshop on the Bible and computing in Leiden this past February and discovered a number of other scholars who were working on valency and the Biblical Hebrew verb. Out of that meeting emerged a session at last month’s SBL annual meeting (somewhat off the beaten path of sessions, but kindly hosted by the International Syriac Language Project program unit) in which I, Janet Dyk, Nicholai Winther-Nielsen, and A. Dean Forbes each delivered a paper addressing the issue of valency in Biblical Hebrew. In turn, out of these meetings came an invitation that Janet present on valency at next year’s SBL meetings in the Bible Translation unit. In the meantime, the four papers from this year’s fruitful meeting will be appearing in the upcoming volume of Gorgias’ series Perspectives on Linguistics and Ancient Languages (see here).

1. Introduction

Along with the familiar triad of tense, aspect, and mood, valency is a defining property of verbs. Although in Hebrew all these properties involve interaction among verbal lexemes, inflection, and syntax, valency is particularly associated with the system of binyanim in contrast to the association of tense, aspect, and mood foremost with the verbal conjugations. Traditionally, valency has been treated under the rubric of either voice or transitivity. However, a valency approach to Biblical Hebrew has two distinct advantages over these traditional categories: firstly, valency analysis is not hampered by the traditional categories of classical grammar; secondly, valency focuses on the nexus between verbs (i.e., lexeme and binyanim) and argument structure (syntax). Because of this particular focus, valency studies can potentially contribute to Biblical Hebrew lexicography and our understanding of the binynaim, as well as to the decipherment of Biblical Hebrew syntax.
In this paper I introduce the concept of valency and contrast it with voice and transitivity (§2). After this introduction, I briefly summarize approaches to valency in Hebrew grammars (§3), explore some of the issues involved in analyzing valency patterns in Biblical Hebrew, including addressing some objections to such a study (§4), and finally, I illustrate with specific examples how my approach to valency advances our understanding of the Biblical Hebrew lexicon and and syntax (§5).

2. Understanding valency 

The term valency derives from the field of chemistry; in linguistic usage the term refers to the number of syntactic elements a verb requires or permits combining with; in short, valency refers to a verb’s syntactic “combining capacity” (Crystal 2008: 507). Although theoretically limitless, the typical range of verbal valency is zero to three constituents. These four patterns—avalent, monovalent, bivalent, and trivalent—are illustrated in (1) by both English and Biblical Hebrew examples. The constituents that define each verb’s valency pattern are underlined and marked by a subscript in each example.

(1) a. Avalent: It is raining.

תַּשְׁלֵג בְּצַלְמוֹן

‘(It) was snowing on Zalmon.’ (Ps 68:15)

b. Monovalent: She slept.

‏לָכֵן שָׂמַח לִבִּי

‘Therefore my heart rejoices.’ (Ps 16:9)

c. Bivalent: He kicked the ball.

‏יְהוָה תְּפִלָּתִי יִקָּח

‘Yhwh will accept my prayer.’ (Ps 6:10)

d. Trivalent: They gave him a drink.

‏וַתַּשְׁקוּ אֶת־הַנְּזִרִים יָיִן

‘You gave wine to the Nazarites.’  (or ‘You made the Nazarites drink wine.’) (Amos 2:12)

These examples are self-explanatory. However, let me note that the avalent pattern illustrated in (1a) is relatively rare, because a well-formed clause typically requires both a subject and a predicate. What defines the examples in (1a) as avalent is that even where they employ a “dummy” subject pronoun, as in the English example and gloss, that pronoun fulfills no thematic or theta role. (This understanding of the “dummy pronoun” importantly distinguishes true subjectless constructions from those with “indefinite” subject referents, such as the impersonal constructions in Biblical Hebrew; cf. Waltke and O’Connor 1990: §4.4.2 and §22.7a). Therefore, the null-subject strategy in Biblical Hebrew should not be interpreted as a valency-reducing feature of the language (cf. Andersen and Forbes 2012: 167); in all cases except the rare avalent pattern illustrated in (1a), clauses that lack an overt subject in Biblical Hebrew are best analyzed as having a null-subject constituent that serves the appropriate thematic role in the clause.

Transitivity is the analysis of the relationship of a verb to its dependent constituents, and as such clearly intersects valency.  However, transitivity is a more narrow concept than valency in two crucial ways. First, transitivity analyzes only “internal arguments”; that is, the verb-phrase–internal constituents, in contrast to valency’s scope of analysis that includes both internal and external arguments (i.e., the subject; see Crystal 2008: 34). Second, transitivity treats only the verb-dependent constituents that are found in traditional grammar, that is, direct and indirect objects; it does not take into account other constituents governed by the verb that might be included in a valency analysis. As such, the transitivity approach of traditional grammar leads to awkward discussions about so-called accusative noun phrases that function as something other than direct object and other “objects” of the verb as mediated by prepositions (e.g., Waltke and O’Connor 1990: §10–11). Given transitivity’s exclusion of the subject and some prepositional constituents in its analysis, it only partially correlates with valency, as illustrated in (2).

(2) a. Avalent verbs are intransitive.

b. Monovalent verbs are intransitive, but intransitive verbs may be any valency.

c. Bivalent verbs may be intransitive or transitive.

d. Transitive verbs are at least bivalent; they cannot be monovalent.

e. Trivalent verbs are often ditransitive, but they may be transitive or, rarely, intransitive.

f. Ditransitive verbs are always trivalent; they cannot be monovalent or bivalent.

More importantly, a valency analysis better clarifies the relatedness between argument structures such as those in (3) than the traditional grammar analysis in terms of transitivity allows: valency theory identifies both the noun phrase in (3a) and the prepositional phrase in (3b) as complements of the verb אחז in each example.

(3) a. Bivalent אחז with noun phrase complement:

יֹּאחֲזוּהוּ פְלִשְׁתִּים

The Philistines seized him.’ (Judg 16:21)

b. Bivalent אחז with ב prepositional phrase complement:

‏וָאֹחֵז בְּפִילַגְשִׁי

‘(I) seized my concubine.’ (Judg 20:6)

Voice analyzes the relationship between the syntactic subject and object and the thematic roles of agent and patient as determined by the verb. For example, the transitive verb with active voice in (4a) takes a subject as agent and the object as patient, whereas the corresponding passive verb in (4b) expresses the same underlying sense while switching the patient role to subject and encoding the agent role with a prepositional phrase.

(4) a. The opera singer(Subject & Agent) sang an aria(Object & Patient).

b. An aria(Subject & Patient) was sung by the opera singer(Agent).

Voice is therefore, like transitivity, both a more narrow concept than valency and derives from traditional grammar, in which the Latinate orientation focuses on morphological distinctions of voice. In Biblical Hebrew voice distinctions are expressed in large part by binyanim, and in her study of the binyanim Maya Arad (2005) has observed several correlations among transitivity, voice distinctions, and the binyanim: according to Arad, both Nifal and Hitpael verbs are intransitive, and the Nifal may also frequently be passive; the Pual and Hofalbinyanim are limited to verb-derived verbs, as opposed to root-derived verbs, in that they always encode the passive counterpart of the Piel and Hifil verb of the same root, respectively. However, because valency is broader than either transitivity or voice, these correlations do not help us escape having to determine the valency patterns of these passive and intransitive verbs, despite the fact that they will tend to have lower valency than verbs in the Qal, Piel, and Hifil binyanim.

3. Approaches to valency

Valency gets only the slightest mention in recent Hebrew grammars, whose approach generally still betrays a traditional-grammar orientation to valency phenomenon. For example, Waltke and O’Connor (1990: §10.2a) note that “[g]rammarians sometimes distinguish between adjuncts and complements, the former signifying an optional constituent of a sentence, the latter an obligatory constituent.” However, they proceed to translate these notions into the traditional-grammar categories of “direct-object accusative” and “adverbial accusative.” Van der Merwe, Naudé, and Kroeze (1999: §33) escape the tradition-grammar approach somewhat more successfully than Waltke and O’Connor and embrace the terminology of complement and adjunct in a more thorough-going approach. In addition, they revise the inherited and simplistic understanding of these categories embraced in Waltke and O’Connor by focusing on the semantic factors rather than the syntactic ones. They define complements as constituents that “cannot be omitted without changing the meaning of the clause or without making the clause ungrammatical,” whereas adjuncts “add information to the core of the clause and may be omitted without changing the basic meaning of the clause.” Further, they state in an accompanying footnote that “[t]he complement of a verb may be omitted, but then only when it can be inferred from the context of the sentence.” Unfortunately, measuring meaning change and grammaticality on a closed corpus for an ancient language is no simple task.

Most recently, Andersen and Forbes (2012: 165–168) in an “aside” on valency in their Biblical Hebrew Grammar Visualized reject a valency approach as problematic on three fronts. First, adopting Crystal’s (2008: 508) standard definition of valency as analyzing the number of valents with which a verb combines to create a well-formed sentence, Andersen and Forbes object that the notion of well-formedness is too vague to be analytically useful for Biblical Hebrew. To illustrate, they provide a statistical analysis of the five verbs that most frequently occur with subjects and those that most frequently appear with a direct object to illustrate how inconsistently the valency pattern of these verbs are. Second, they draw attention to the inherent danger of analyzing English translations of the Hebrew data rather than the Hebrew valency patterns themselves insofar as the semantics and accompanying valency patterns do not match between languages. Third, they note that valency analysis has limited applicability because of the dearth of data; specifically the high incidence of low-frequency verb forms does not allow us to draw valid generalizations from the data.

However, the latter two objections are no serious grounds for abandoning a valency analysis of Biblical Hebrew inasmuch as they apply equally to any linguistic study of Biblical Hebrew. For instance, I have drawn attention to precisely the danger of translation confusion with the target language in my study of tense, aspect, and modality in Biblical Hebrew (Cook 2012: 56). And given the closed and uneven data set that constitutes Biblical Hebrew, any linguistic generalizations about the language must be seen as tentative to one degree or another. By contrast, their objection regarding well-formedness is more serious, especially given the lack of native speakers of Biblical Hebrew: methodologically we must assume that all of the Biblical Hebrew data is well-formed until a case is made to the contrary. However, even in valency studies of spoken languages, well-formedness fails as the central criterion for distinguishing complements and adjuncts, and some studies retreat to the use of statistics in making such judgements (e.g., Villavicencio (2002) sets a threshold of 80% occurrence for identifying a type of constituent as a complement rather than an adjunct). I submit that by reassessing the complement-adjunct distinction as I propose below, this difficulty which Andersen and Forbes point out can be obviated to the degree that it is no longer a serious hinderance to a fruitful valency analysis of Biblical Hebrew.

4. Issues in valency analysis

This brief survey suggests two cautions in developing an analytically useful valency approach to Biblical Hebrew: first, we must develop a more sophisticated understanding of complements and adjuncts than the obviously simplistic identity of these two arguments as obligatory and optional, respectively; on the other hand, we need more rigorous guidelines than a simple vague notion of “well-formedness,” as Andersen and Forbes point out. What is needed is an approach that recognizes the instinctually correct idea that complements are more integral to the predication than adjuncts and analyzes this distinction in a nuanced way that involves both syntactic and semantic factors, given that valency involves the intersection of these two domains. In this way valency study can contribute to our understanding of Biblical Hebrew syntax and lexical semantics, and contribute to the philological task of deciphering the Hebrew texts of the Bible.

Consider the English examples in (5): despite the variation of valency and transitivity of the verb give, all three expressions are equally “well-formed,” grammatically speaking.

(5) a. I give and I give, but do I ever receive any thanks? (monovalent/intransitive)

b. When I heard of her passing, I gave flowers in her memory. (bivalent/transitive)

c. I gave flowers to my wife on Valentine’s day. (trivalent/ditransitive)

The simplistic and binary distinction between complement and adjunct is insufficient for analyzing these various argument structures. The two “graded” divisions in (6) have been suggested as alternatives to the traditional binary distinction of complements and adjuncts.

(6) a. Primary complements — Secondary complements — Adjuncts (DeArmond and Hedberg  1998; 2003).

b. (Obligatory) complements — Optional complements — Contextually optional complements — Adjuncts (Herbst 1999).

The intermediate category of secondary complements in (6a) is based on the contrastive results of linguistic tests applied to benefactive, instrumental, and some types of locative prepositional phrases: while do-so and pseudo-cleft tests identify these constituents as complements, the preposition-stranding test identifies them as adjuncts. Unfortunately, the application of linguistic tests of these sort to Biblical Hebrew is difficult given the absence of native speakers and a closed corpus of data.

Herbst’s three-way complement distinction in (6b) is a more promising basis for valency analysis of Biblical Hebrew verbs, not only because it does not rely on linguistic tests, but because the nature of these distinctions is more obvious and measurable from the data. Let me explain Herbst’s graded categories as they apply to Biblical Hebrew, based on the ongoing use of this model in the development of the Accordance Bible software syntax module (see post). Because our concern is with distinguishing complements and adjuncts and due to the infrequency of avalent or subjectless constructions, I will simply ignore the subject-role complements in my analyses.

First, a verb may have a syntactically obligatory complements; that is, the absence of these constituents makes the expression ungrammatical. However, “obligatory” is in parenthesis in reference to this category in (6b), because different complement patterns may be associated with a single verb. Often a distinction in meaning can be discerned among the different patterns, such as illustrated in (78) below.

(7) סמך (qal)

a. Bivalent with NP complement ‘support someone/something’:

הֱקִיצוֹתִי כִּי יְהוָה יִסְמְכֵנִי

‘I awake, because Yhwh supports me.’ (Ps 3:6)

b. Trivalent with NP and על-PP complements ‘lay something on someone’.

וְסָמַךְ יָדוֹ עַל רֹאשׁ הָעֹלָה

‘(He) should lay his hand upon the head of the burnt offering.’ (Lev 1:4)

Examples (7a–b) illustrate two distinct meanings for the qal verb סמך, which are associated with the two distinct valency patterns: a bivalent one and a trivalent one. The trivalent pattern appears to be a technical meaning, appearing only in sacrificial contexts with the exception of one occurrence in Amos 5:19.

A different sort of semantics-based variation is illustrated in by the examples in (8): a trivalent pattern with a noun phrase and prepositional complements is associated with the meaning to ‘give’ (or ‘place’), as illustrated by (8a), whereas a trivalent pattern with an noun phrase and complementary infinitive is associated with the meaning ‘allow’, as illustrated by (8b).

(8נתן (qal)

a. Trivalent with NP and PP complements: ‘give something to someone’

‏‏הָאֵל הַנּוֹתֵן נְקָמוֹת לִי

‘The God who gives vengeance to me.’ (Ps 18:48)

b. Trivalent with NP and INF complements: ‘allow someone something’

‏‏לֹא־יִתְּנֵנִי הָשֵׁב רוּחִי

‘(He) will not allow me to catch my breath.’ (Job 9:18)

Finally, variation among obligatory valency patterns might admit other explanations. For example, the monovalent intransitive pattern for the hifil of נגע ‘to arrive’, illustrated in (9), occurs only in the books of Songs, Qoheleth, and Esther, which may be dialectally diachronically significant.

(9נגע (hifil)

הַנִּצָּנִים נִרְאוּ בָאָרֶץ עֵת הַזָּמִיר הִגִּיעַ

‘The blossoms have appeared in the land, the time of pruning has arrived.’ (Song 2:12)

Secondly, a complement is “optional,” according to Herbst (1999), if it is implied by the structure of the predicate itself. Consider the English examples in (10): the former examples in each pair imply a “generalized” complement based on the semantics of the verb itself: one normally reads something with words; one normally cooks food. If the meaning departs from these general senses, a complement is required to cancel the implied complement, as in the second pair in each example.

(10) a. She is reading. (implied complement: something with words)

cf. She could always read his face.

b. He is cooking. (implied complement: food)

cf. He is cooking up trouble.

A Biblical Hebrew example that falls into this category is the qal verb שׁיר ‘to sing’: in its monovalent pattern in (11a), the verb implies a generalized complement of  ‘song’ or the like. However, the verb may also exhibit a bivalent pattern, as in (11b), in which what is sung or sung about is specified by a noun phrase complement. An important piece of evidence supporting the claim of an implied complement is the occasional presence of a cognate complement with such verbs in the bivalent pattern, as in example (11c), in which שירה ‘song’ is the cognate complement of שׁיר. In these cases the cognate complement reinforces the generalized complement implied by the verb itself.

(11שׁיר (qal)

a. Monovalent with implied complement

‏אָשִׁירָה וַאֲזַמְּרָה לַיהוָה

‘(I) will sing and make melody to Yhwh.’ (Ps 27:6)

b. Bivalent with overt complement: e.g., ‘sing something’

וַאֲנִי אָשִׁיר עֻזֶּךָ

‘But I will sing of your strength.’ (Ps 59:17)

c.Bivalent with cognate complement

אָז יָשִׁיר־מֹשֶׁה וּבְנֵי יִשְׂרָאֵל אֶת־הַשִּׁירָה הַזֹּאת לַיהוָה

‘Then Moses and the children of Israel sang this song to Yhwh.’ (Exod 15:1)

Finally, “contextually optional” complements refer to constituents that are recoverable or identifiable from the discourse context, in contrast to being implied by the verbal semantics alone, as in the previous case. One indication of this category of valency variation is the infrequency with which a complement might be absent. For example, only 3 of 59 occurrences of the hifil of שׁקה ‘to give a drink’ lack a complement referring to the recipient of a drink. In each instance a good case can be made that the complement is elliptical—that is, null but identifiable from the context. The null constituent and its antecedent are in parentheses in the examples in (12). Note also that what is offered to drink is unspecified, being an optional complement, as in the case of שׁיר, just discussed.

(12) a. Deut 11:10

‏‏אֲשֶׁר תִּזְרַע אֶת־זַרְעֲךָ וְהִשְׁקִיתָ בְרַגְלְךָ

‘. . . where (you) sowed your seed and watered (it = your seed) with your foot’

b. Ps 78:15

‏יְבַקַּע צֻרִים בַּמִּדְבָּר וַיַּשְׁקְ כִּתְהֹמוֹת רַבָּה

‘(He) split rocks in the wilderness and gave (them = them v. 14) drink as the great depths’

c. ‏‏Esther 1:7

וְהַשְׁקוֹת בִּכְלֵי זָהָב וְכֵלִים

‘. . . giving (them = all the people v. 5) drinks in gold vessels’

Determining contextually optional complements is complicated by the previous category of optional complements, because a verb might exhibit both types of valency variation with the result that in the case of a contextually optional example the text is not syntactically “fragmentary” as we expect for elliptical structures. Consider the examples of the qal verb אכל ‘to eat’ in (13): as with English eat, אכל may imply a generalized complement of food as in (13a); but in Genesis 3:6, cited in (13b), the verb has an contextually optional (that is, elliptical) complement whose antecedent is מִפִּרְיוֹ.

(13) a.Ruth 3:7

‏וַיֹּאכַל בֹּעַז וַיֵּשְׁתְּ וַיִּיטַב לִבּוֹ

‘Boaz ate and drank and his heart became merry.’

b. Gen 3:6

‏‏וַתִּקַּח מִפִּרְיוֹ וַתֹּאכַל

‘She took some of the fruit and she ate (it = some of the fruit).’

This type variation among an optional and contextually optional complement may appear in a single passage, as in (14).

(14) 1 Kings 19:5–8‏

‏וַיִּשְׁכַּב וַיִּישַׁן תַּחַת רֹתֶם אֶחָד וְהִנֵּה־זֶה מַלְאָךְ נֹגֵעַ בּוֹ וַיֹּאמֶר לוֹ קוּם אֱכוֹל׃ ‎6‏ וַיַּבֵּט וְהִנֵּה מְרַאֲשֹׁתָיו עֻגַת רְצָפִים וְצַפַּחַת מָיִם וַיֹּאכַל וַיֵּשְׁתְּ וַיָּשָׁב וַיִּשְׁכָּב׃ ‎7‏ וַיָּשָׁב מַלְאַךְ יְהוָה שֵׁנִית וַיִּגַּע־בּוֹ וַיֹּאמֶר קוּם אֱכֹל כִּי רַב מִמְּךָ הַדָּרֶךְ׃ ‎8‏ וַיָּקָם וַיֹּאכַל וַיִּשְׁתֶּה וַיֵּלֶךְ בְּכֹחַ הָאֲכִילָה הַהִיא אַרְבָּעִים יוֹם וְאַרְבָּעִים לַיְלָה עַד הַר הָאֱלֹהִים חֹרֵב׃

‘He lay down and fell asleep under a broom bush. Suddenly an angel touched him and said to him, “Arise and eat.”  6 He looked about; and there, beside his head, was a cake baked on hot stones and a jar of water! He ate (it = the cake) and drank (it = the water), and lay down again.  7 The angel of the LORD came a second time and touched him and said, “Arise and eat, or the journey will be too much for you.”  8 He arose and ate (it = the cake) and drank (it = the water); and with the strength from that meal he walked forty days and forty nights as far as the mountain of God at Horeb.’ (NJPS)

In verse 5 the angel awakes Elijah and tells him קוּם אֱכוֹל. Here the imperative אֱכוֹל appears to be monovalent, with an optional complement implied by the predicate itself: ‘Eat (something).’ In the following verse (v. 6), however, Elijah looks near his head and finds a stone-baked cake (עֻגַת רְצָפִים) and a jar of water (וְצַפַּחַת מָיִם), and the text reports וַיֹּאכַל וַיֵּשְׁתְּ. Both these verbs should be treated as bivalent with contextually optional (i.e., elliptical) complements. Thus, we can intelligibly render them: ‘He ate it and drank it’. This command-narrative pattern is repeated in the following two verses (vv. 7–8), where the angel tells Elijah to eat and drink again. Although the bread and water are now known entities in the discourse, the expression in verse 7 is parallel with that of verse 5, suggesting that as in the previous case the repeated command here is likewise monovalent with a generalized optional complement. The fact that the angel does not specifically tell him to eat and to drink lends some weight to this monovalent interpretation. Similarly, for the repeated report in verse 8 that Elijah וַיֹּאכַל וַיִּשְׁתֶּה ‘ate and drank’ we should understand the two verbs as bivalent, their null complements referring to the cake and water that the reader will infer the angel resupplied or were left over from Elijah’s previous meal.

5. Some illustrations

Having explained valency and advocated a specific approach to valency analysis in Biblical Hebrew, it remains to illustrate the value of carrying out such an analysis. The contribution of valency analysis to our understanding of Biblical Hebrew goes in two directions. In the one direction, valency studies can contribute to lexicography by providing a syntactic basis for distinguishing different nuances of meaning as they are demonstrated to align with specific valency patterns. In the other direction, valency analysis can inform philology by providing data to arbitrating between alternative analyses of some clauses in  the text. Let me illustrate each of these with an example.

First, analyzing lexical meaning in terms of valency patterns may undergird distinctions among homonymy in the lexicon in ways that simple semantic analysis cannot. For example, HALOT lists together under the single root עלל I the poel meanings ‘treat severely’ and ‘glean’, which are illustrated by the examples in (15).

(15עלל I (poel)

a. ‘Treat severely’

‏עֵינִי עוֹלְלָה לְנַפְשִׁי

My eye treats me severely.’ (i.e., ‘afflicts me’) (Lam 3:51)

b. ‘Glean’

וְכַרְמְךָ לֹא תְעוֹלֵל

‘And your vineyard (you) do not glean (completely).’ (Lev 19:10)

While one might be able to do some acrobatics to see how these are etymologically semantically related, HALOT’s entry is only marginally helpful in pointing out that these meanings are distinguished by valency pattern: ‘treat severely’ has a ל prepositional phrase complement whereas the meaning to ‘glean’ has an noun phrase complement. The former occurs only in Lamentations, where the passive poal also occurs with the sense of ‘be treated severely’. Based on this semantic alignment with the different valency patterns, it may be best to see these as two separate verbs, as indeed BDB treats them: though it ultimately relates the verbs to the same root as HALOT, BDB identifies the meaning ‘glean’ as a denominative form from the feminine noun עֹלֵלוֹת ‘a gleaning’.

An example in which attention to valency patterns aids philological analysis of the text is provided by the passage in (16).

(16) דרך (qal)

גַּת דָּרַךְ אֲדֹנָי לִבְתוּלַת בַּת־יְהוּדָה

NRSV: The Lord has trodden as in a wine press the virgin daughter Judah.
NJPS: As in a press the Lord has trodden Fair Maiden Judah.
‘The Lord has trodden the wine press for my the virgin daughter Judah.’ (Lam 1:15c)

Both the NRSV and NJPS treat לִבְתוּלַת as the complement of the verb דרך, and גַּת as some sort of adverbial accusative. But דרך does not elsewhere take as its complement a ל prepositional phrase, though it does appear 5 times with על and 10 times with ב prepositional complements both with a locative idea ‘upon’ or ‘on’. The majority of the time, however, it takes an noun phrase complement. Based on this valency pattern, it is best to identify גַּת as the complement and the prepositional phrase לִבְתוּלַת בַּת־יְהוּדָה as an adjunct, as indicated by the third translation option in (16). And indeed, this is how Keil (1989: 8.372–73) takes the text, explaining: “These [i.e., the young men mentioned in 1:15b] celebrate a feast like that of the vintage, at which Jahveh treads the wine-press for the daughter of Judah, because her young men are cut off like clusters of grapes (Jer. vi. 9), and thrown into the wine-press (Joel iv. 13).”

6. Summary and conclusion

The study of verbal valency in Biblical Hebrew is in its infancy. It is crucially focused on the intersection of syntax and semantics, with the result that it can inform our understanding of both syntax and lexical semantics of Biblical Hebrew. The approach I have proposed and illustrated above, and which is being worked out in the Accordance Bible software syntax module, is one that successfully overcomes possible objections to valency analysis of Hebrew and provides a usable approach to the analysis of Biblical Hebrew argument structure.

Works cited

Andersen, Francis I. and A. Dean. Forbes. Biblical Hebrew grammar visualized. Winona Lake, Ind.: Eisenbrauns, 2012.

Arad, Maya. Roots and Patterns: Hebrew Morpho-Syntax. Studies in Natural Language and Linguistic Theory. Dordrecht: Springer, 2005.

Cook, John A. Time and the Biblical Hebrew Verb: the Expression of Tense, Aspect, and Modality in Biblical Hebrew. Linguistic Studies in Ancient West Semitic. Winona Lake, IN: Eisenbrauns, 2012.

Crystal, David. A Dictionary of Linguistics and Phonetics. 6th ed. Oxford: Blackwell, 2008.

Dearmond, Richard C. and Nancy Hedberg. On Complements and Adjuncts. Proceedings of the 1998 Annual Conference of the Canadian Linguistics Association, 1998.

———. More Issues on the Argument Structure of Primary Complements. Proceedings of the 2003 Annual Conference of the Canadian Linguistics Association : 50–61, 2003.

Herbst, Thomas. English Valency Structures: A First Sketch. Erfurt Electronic Studies in English 6, 1999.

Keil, Carl Friedrich and Franz Delitzsch. Commentary on the Old Testament. Translated by James Martin. Peabody, MA: Hendrickson, 1989.

Van Der Merwe, Christo H. J., Jackie A. Naudé, and Jan H. Kroeze. A Biblical Hebrew Reference Grammar. Sheffield: Sheffield Academic, 1999.

Villavicencio, Aline. Learning to Distinguish PP Arguments from Adjuncts. Proceedings of the 6th Conference on Natural Language Learning 20: 1–7, 2002.

Waltke, Bruce K. and M. O’Connor. An Introduction to Biblical Hebrew Syntax. Winona Lake, IN: Eisenbrauns, 2002.

11 Responses to “Valency: the intersection of syntax and semantics”

  1. matthaiti Says:

    Thanks for this, Dr. Cook. Very helpful as I’m currently working on my PhD with David Firth in the UK on syntactic analysis of the Egyptian Hallel. You’ve added to my bibliography yet again!
    Blessings and Merry Christmas to your family,

    • johncookvw Says:

      I’m glad it was helpful, Matt. I’ll be interested in the conclusions from your study of the Egyptian Hallel psalms. The book of Psalms was what I started with for the Accordance tagging, and, though it is released, I’m sure there are plenty of corrections still to be made to that syntactic tagging!

  2. Jonathan Kiel Says:

    Thank you, Dr. Cook!

    Over the past year, I have had a few conversations with my advisor (Peter Gentry) regarding the usefulness of a Hebrew lexicon that would incorporate verbal valency. I am thankful for your contribution and am now eager to track down the work that is already complete. (I was previously unaware this discussion was already underway!) Any suggestions on where to start prior to the next installment of ‘Perspectives on Linguistics…’?

    Thanks again,

    • johncookvw Says:

      Thanks for your reply, Jonathan. A useful valency lexicon is sorely needed—as I’ve become increasingly aware of as I’ve analyzed these verbs and examined what the entries in the standard lexica have (or lack) in this regard. To be honest, most of the valency studies in linguistics seem either scattered, often unpublished papers, or come from the Continent, where they have shown much more interest in the topic. This paper is in essence a very “baby step” forward through the application of Herbst’s ideas to BH. Herbst is the lead author of the Valency Lexicon of English (deGruyter 2004), and the forward of that probably lays out his theory better than anywhere else.

      • Jonathan Kiel Says:

        Thanks, Dr. Cook!
        I’ll check out Herbst. Looking forward to an ongoing discussion on the topic…

  3. Mike Aubrey Says:

    Very nice summary, John. My only concern would be that a lexicon that followed this kind of analysis could potentially get rather bloated in listing all the various patterns for a given verb.

    • johncookvw Says:

      Thanks Mike. Though your concern is valid, though the way we are currently progressing in the “working” lexicon for the Accordance project, it does not appear that such a lexicon would appear quite “fit” next to HALOT and downright anorexic next to the multi-volume DCH. Since it is only verbs and the basic entry can be quite condensed (we include root + binyan, basic gloss, range of patterns, sigla denoting patterns, such as S V NP, PP etc., and frequency), it really comes down to a matter of (1) whether to list verses (under 25 occurrences we tend to list the verses; above that the references only) and (2) how much “comment” about the patterns is helpful versus cluttering. That latter will ultimately prove the biggest challenge.

      • Mike Aubrey Says:

        I would suggest looking for schematic constructions in valency patterns across multiple verbs. If you haven’t read Adele Goldberg’s Constructions: A Construction Grammar Approach to Argument Structure, I would definitely recommend it.

  4. johncookvw Says:

    Thanks Mike, I’ll have to check into that. I’ve generally shied away from more cognitive approaches; they just never seem very satisfying to me. Probably a bias I should overcome!

    • Mike Aubrey Says:

      Yeah, I know. I think we talked a bit about that in Leiden (personally, I’m a big fan of cognitive linguistics). But even independent of that, she does a nice job articulating the challenges of standard approaches in way that I would expect you might be able to find something useful. This book is, in a sense, her least cognitive work. Prototype semantics, metaphor, and other important ideas from cognitive linguistics don’t play a particularly large role in this particular book.

  5. Keep ‘em coming back with the December Biblical Studies Carnival | Words on the Word Says:

    […] started in December, with a first entry on information structure in Jonah 1. John Cook discusses valency and verb theory in Biblical […]

Comments are closed.

%d bloggers like this: