How did the Hebrew language evolve between the Bible and the Mishnah? Ken Penner’s forthcoming book, The Verbal System of the Dead Sea Scrolls, contributes significantly to the discussion of this issue. He agreed to sit down and answer a few questions, each of which you can see listed here:
List of Questions
- What is the argument of the book?
- What piqued your interest in the verbal system of the Dead Sea Scrolls?
- How has the work developed between dissertation and publication? Any advice for students who hope to publish their dissertation?
- What tools or methodologies did you find most useful in your research?
- How does your analysis of the verbal system shed light on interpretive debates in the broader field of Dead Sea Scrolls scholarship?
- How does the verbal system of the Dead Sea Scrolls compare and contrast with the the verbal system(s) of the Hebrew Bible?
- Do you plan to extend your research and publish on the verbal system of the Hebrew Bible?
- Where does your view of the verbal system of the Hebrew Bible fit within the spectrum of recent debates?
- How accurately do the written sources reflect the spoken Hebrew of the time? Does the verbal system of the Dead Sea Scrolls result from natural linguistic development or intentional archaism? Do the Dead Sea Scrolls attest to a single common verbal system, or are there numerous parallel systems evident in the sources?
Ken, what are you arguing in this book?In this study, I have sought to determine to what extent Hebrew verbs in Qumran texts encode tense, aspect, and mood. General linguists have discussed the question of which “parameter” (usually Tense, Aspect, or Modality, TAM for short), is “prominent,” meaning it provides the main semantic difference between verb forms.
Most standard textbooks claim that Biblical Hebrew primarily encodes aspect and that Mishnaic Hebrew primarily encodes tense. Qimron’s standard reference work for Qumran Hebrew does not attempt to align the verb forms of Qumran Hebrew according to one or the other of these, its two chronological neighbours. There is no established consensus regarding the semantics of the Qumran Hebrew verb conjugations, and the result is inconsistency among scholars interpreting the verb forms in Qumran texts. Most scholars are able to recognize when a verb form is used atypically, but until now, no one has been able to describe just how anomalous a given usage might be.
Such confusion would not be tolerated for long in a field of more physical sciences, given the availability of relevant data. Tests would be devised to solve the question. For example, imagine we are at a time in the past when we are just learning that things expand when heated. We might then hypothesize that water expands at a constant rate as it is heated. We could then test that view by measuring how closely our predictions match the reality we observe. If our predictions don’t match the measurements, we would then adjust the theory that led to those predictions.
In the study of ancient Hebrew, the “measurements” are the Hebrew verb forms that we find in the text. A good model of the Hebrew verbal system needs to be able to predict the verb forms the ancient authors would choose. The more a theory is able to anticipate these choices, the more explanatory power it has, and the better it is. If a theory does not explain the data, we need to revise the theory to explain the data better. We need a model that accounts for the author’s choice of verb form, and the model to be preferred is that which most efficiently explains reality. We need models that are testable, that are falsifiable. It would be unhelpful to say the verb forms do not indicate time but instead are used to structure paragraphs, unless one also stated which structures demand which verb forms. Only then would there be a clear way to test the claim. Granted, linguistics is an imprecise science; and all grammars “leak”. But that need not stop us; we seek not objectivity but repeatability.
I found that Qumran Hebrew does not encode aspect if we mean what general linguists such as Comrie and Binnick meant by “aspect”. Rather, the Qumran Hebrew verb forms align themselves much more closely with tense than with aspect. Future tense may be considered a form of irrealis modality. In Qumran texts, 93% of the yiqtol forms (without waw) are used for 52% of the future statements, and 70% of the qatal forms are used for 62% of the statements regarding the past. Present actions are expressed using yiqtol in 72% of instances, and states in the speaker’s present are expressed using qatal in 64% of instances. In total, 97% of yiqtols are modal, and conversely, 51% of the expressions of modality use yiqtol. It is clear that tense and modality are both strongly correlated to verb form, but because the data is too meagre, it is not perfectly clear which of the two is more prominent in Qumran Hebrew. The balance is tipped slightly in favour of tense, since five of seven past modal expressions use qatal, and future indicative expressions use yiqtol. Qumran Hebrew is not tenseless; quite the opposite: it is tense-prominent.
What piqued your interest in the verbal system of the Dead Sea Scrolls?
I first became fascinated with the meaning of the Hebrew “tenses” in the 1990’s, while learning biblical Hebrew and at the same time writing comprehensive examinations in Greek verbal aspect. None of the textbooks from which I first learned and taught Hebrew explained the conjugations in quite the same way. Some grammars called the difference between qatal and yiqtol “tense” and others “aspect.” When I moved on to the study of the Dead Sea Scrolls, I found that although the majority view regarding biblical Hebrew was that these forms were “perfect” and “imperfect,” and for Mishnaic Hebrew these were thought to be “past” and “future,” for texts written in the intervening period, the time of Qumran Hebrew, no consensus was to be found. Yet our reconstruction of both the history and theology of this community at times depends on the implications of these Hebrew conjugations.
How has the work developed between dissertation and publication? Any advice for students who hope to publish their dissertation?
I revised my dissertation to incorporate recent theoretical developments (notably of Andrason’s pan-chronic theory of the development of the verbal system, and Geiger’s work on the Qumran participle), and another case study (the tense problem Stuckenbruck raised regarding Pesher Habakkuk). I updated my evaluation of the views of Cook and Joosten on the basis of their recently published books.
Note that the longer you put off submitting your proposal to a publisher, the more work you will have to do to revise it! I had my Greek Isaiah commentary also in progress, so when the first editor suggested my Qumran Hebrew research would be better published elsewhere, I decided to concentrate on Greek Isaiah instead. In retrospect, knowing now how much time the commentary has taken, I should have worked on getting the dissertation published first.
What tools or methodologies did you find most useful in your research?
I sought methodological rigour in six significant ways: first, by choosing a synchronic rather than diachronic approach; second, by choosing an empirical rather than theoretical approach; third, by using a statistically significant and representative sample of the relevant linguistic output; fourth, by seeking statistically strong bidirectional correlations between form and function rather than simply listing the possible functions for each form; fifth, by determining the semantic value of each verb form from its context rather than from presuppositions regarding the forms; and finally, by using only verbs having clear tense/aspect/mood reference, rather than including even ambiguous cases. Some helpful tools:
- Computer Database
Today’s technology allows us to create a computer database with an ancient Hebrew corpus to test the various hypotheses. I first wrote a Visual Basic application to extract the formal data from Marty Abegg’s Accordance module “Qumran Sectarian Manuscripts.” I then used the linguists’ models of events and modality described below to tag the semantic values. Finally, I wrote another application in Visual Basic to create tables that showed the correlations between the formal features and the semantic values.
- Event Model
To conduct the experiment, I needed a method that yields repeatable results, something that reduces the dependence on the researcher. I needed clearly defined categories, i.e., what is meant by “aspect,” “mood,” and “tense.” The primary tool that provides some of this objectivity regarding TAM is event model that Hans Reichenbach first proposed.
In this Event Model, a predication has three time frames, and verbs are inflected based on the relationship between these three times. There is the time of the utterance (called the “speech time”), the time about which the utterance is predicated (called the “reference time”), and the time the event or situation takes place (called the “event time”). Reichenbach depicted this “three-place structure of time determination” on a timeline using the initials S, R, and E (for speech, reference, and event). To illustrate how this event model works, if I say now
“At noon yesterday I had already eaten,”
the speech time (S) is today, the reference time (R) is noon yesterday, and the event time (E), is the eating, which must have occurred some time prior to noon yesterday.
Reichenbach also incorporated aspect into his event model by allowing the possibility that the event could occur not just at one point in time, but over a “stretch” of time. The next figure illustrates the following statement.
“Yesterday at noon I was reading a book”
The possible tenses are determined by the relative ordering of these three time points. In the next chart labelled “TAM and the Event Model”, the “less than” and “greater than” symbols indicate temporal sequence (S is less than R means S occurs before R), and the “equals” symbol indicates simultaneity. S “equals” R means the reference point is at the same time as the time of speaking.
Following the “possible worlds” understanding of modality, Galia Hatav argued that since human experience is such that at any point in time it is not known what will happen next, there are multiple alternative futures at that point. For every undetermined possibility, there is a hypothetical world in which that possibility is true. Only one of these possible worlds turns out to be the “real” world. Even for events that happen in the past, at the time of the event, it was not known which of the possible worlds would turn out to be the actual world.
Consider the following example. On Wednesday, John is trying to decide what he will do on the weekend: take a trip, or stay home. Either option is open, a possible world, but only after Saturday will we know which is realized in the actual world. He decides to take a trip. Now new options are open; he must decide where to go: the beach, the mountains, or his parents. In our event model, the possibilities can be visualized as a constantly branching timeline:
According to Hatav, modal propositions are evaluated according to their quantification over branching options and to the relation of their Reference Time with the Speech Time. The statement “John must go to the beach” is true if and only if in every branching option subsequent to the Speech Time John goes to the beach. “John may go to the beach” is true if in at least one branching option John goes to the beach. “John will go to the beach” is true if in every branching option that turns out to be the actual option, John goes to the beach. In this way, future statements may also be considered modal.
How does your analysis of the verbal system shed light on interpretive debates in the broader field of Dead Sea Scrolls scholarship?
One should not easily dismiss the verb forms’ normal time reference. In examining the question of present-realized or future salvation in the Hodayot, I show that because qatals with future reference are uncommon, appealing to the category of “prophetic perfect” is a questionable practice; an interpretation that does not appeal to the “prophetic perfect” is grammatically preferable. Similarly, I maintain regarding the question of the chronology of the “hangings” in pesher Nahum, that habitual actions in the past are routinely expressed using qatal, not yiqtol. Qumran scholars have already recognized that qatals with future reference and yiqtols with past reference were somewhat anomalous; the results of this study now permit us to calculate exactly that probability. The likelihood that any qatal refers to the future is 18%; the likelihood that any yiqtol refers to the past is 1.1%.
How does the verbal system of the Dead Sea Scrolls compare and contrast with the the verbal system(s) of the Hebrew Bible?
- Like Biblical Hebrew, the function of weqatal verbs in Qumran Hebrew is very close to that of yiqtol verbs. In Biblical and Qumran Hebrew, weqatal is used for actions in the future. This is the opposite of what is observed in the Mishnah, where Segal claims weqatal expresses consecutive actions in the past. The correlation between future and weqatal is so strong in Qumran Hebrew and so pervasive across all genres and documents that it could not be simply the result of imitation of biblical style.
- According to standard grammars, in Biblical Hebrew past actions with nonperfective aspect tend to be expressed using yiqtol. In Mishnaic Hebrew, they are expressed using the copula with the participle. Unlike either of these, Qumran Hebrew (apart from MMT) most commonly uses the qatal for such past imperfective or habitual actions (28 times), but occasionally uses the participle (twice with the copula, and twice without) and yiqtol (twice) as well.
- In Biblical Hebrew, the negator often distinguishes a modal use of yiqtol from an indicative use. If לא is used, the verb tends to be indicative future or a gnomic prohibition; if אל is used, the verb is usually a specific negative request. In our corpus, 233 yiqtols are negated with אל; of these 233, 155 are deontic modal. Of the 82 yiqtols negated with אל, all are deontic (volitional).
- The question of the use by the Qumran authors of the wayyiqtol form is complicated by the fact that the (consecutive) wayyiqtol and the (conjunctive) weyiqtol are not normally distinguished orthographically in unpointed texts. In the Masoretic Text, only in ל״ה verbs, hollow verbs (ע״ו and ע״י), and the Hiphil binyan is a distinction between the two forms visible in the consonantal text: ל״ה waw-consecutives have no final ה; hollow ע״ו andע״י waw-consecutives are missing the ו or י, and Hiphil waw-consecutives are missing the characteristic י. However, in Qumran Hebrew it is not clear whether there is any such pattern of orthographic distinction. At least in the case of ל״ה verbs, there is not; the verbs are written in their long or short forms depending solely on grammatical person: all first person wyqtls have a final ה, and almost all others do not. In the case of hollow verbs and Hiphils, the question of whether wayyiqtols can be formally distinguished from weyiqtols in Qumran texts remains unanswered.
Do you plan to extend your research and publish on the verbal system of the Hebrew Bible?
At the 2015 SBL Annual Meeting in Atlanta, I will present a paper applying my method to the first two chapters of 1 Samuel. I wrote the paper for a collection edited by Jackie Naude and Cynthia Miller to be published by Eisenbrauns.
A few important works on the Hebrew verbal system of the Bible have been published recently. I'm thinking of Joosten, Cook, and the articles by Andrason. Where does your view of the verbal system of the Hebrew Bible fit within the spectrum of recent debates?
John Cook and Alexander Andrason both placed more emphasis on examining the diachronic evolution of the Hebrew verb forms, to establish a typologically plausible synchronic model. This evolutionary framework, though diachronic, can help explain the realities of synchronic languages because a grammatical object cannot be reduced to one synchronic function. Although there can be a prototypical meaning of the formation, that is, values expressed most frequently, these will be among the range of possible meanings. Because the meanings of a gram involve multiple semantic spheres, any contrast with another gram cannot be simplified to a single opposition.
Cook argued that Biblical Hebrew is aspect-prominent, for two reasons: languages tend to be more aspectual than tensed; and the fact that Hebrew stative qatals are used for present states points to the perfective aspect.
Andrason considered qatal mainly used for perfect, perfective, and past (Andrason 2013a), with associated senses of “future perfect, simple future, prospective certainty, or inevitability, future imminence and near future, present about-ness as well as ‘almost’ and counterfactual present perfects” (Andrason 2013c).
Matthew Anstey described the Hebrew verbal system as tense-prominent. Wayyiqtol is the standard verb used for narrative. For other statements about the past, qatal is used; for non-past, yiqtol is used, (Anstey 2009, 826).
Ron Hendel claimed qatal and yiqtol are primarily relative tenses, expressing the relationship between R and E; the exact tense is dependent on the situation aspect of the verb. If the verb is dynamic, qatal indicates relative past and yiqtol indicates relative non-past; if stative, qatal is for relative non-future and yiqtol is for relative future. Secondarily, qatal is used for perfective and yiqtol for imperfective expressions (Hendel 1996).
Dennis Pardee described qatal and wayyiqtol as perfective, and yiqtol and weqatal as imperfective. Yiqtol could be used in historical narrative to express iteration and in direct speech to express the future and modalities (Pardee 2012).
Mats Eskhult asserted that the basic oppositions are aspectual: qtl for “state” and yqtl for “motion”; yiqtol and qotel for imperfective “ongoing” “cursive” action, and qatal and wayyiqtol for perfective “constative” action, “viewed as a single whole” (1990, 121).
Elizabeth Robar said the yiqtol is imperfective, and the wayyiqtol is a narrative present, unmarked for aspect or mood; it is relative present, dependent on the preceding clause for its tense (Robar 2013).
Randall Buth maintained the biblical Hebrew verb forms expressed were used for all three parameters: tense, aspect, and mood: qatal and wayyiqtol for definite statements (past or perfective or decisive or contrary to fact) – qatal when discontinuity is intended, and wayyiqtol for thematic continuity; yiqtol and weqatal for indefinite statements (future or imperfective or potential or repetitive) – yiqtol when discontinuity is intended, and weqatal for thematic continuity (Buth 1992, 103).
Jan Joosten took a synchronic approach, taking into account the forms available to express each meaning in tense, taxis, aspect, and modality. He described the qatal as a anterior relative tense (taxis), yiqtol as irrealis mood, wayyiqtol as past tense, and the participle as progressive aspect (Joosten 2012).
Tania Notarius claimed qatal and wayyiqtol are perfective, and yiqtol and weqatal are for future and epistemic and deontic modal expressions. In sequential tenses the Reference time is moved to a point after the event (Notarius 2011, 277). Qotel is progressive aspect (Notarius 2011, 280; Notarius 2008).
Oded Cohen described the semantics of the verb forms explicitly in terms of the Event time “E” and Reference time “R.” Qatal is for an anterior or coincident E:R relationships, and for realis modality that is not habitual or iterative. Wayyiqtol and qotel carry the same modality, but wayyiqtol is for perfective coincident E:R relationships and qotel is for imperfective E:R. Yiqtol and weqatal are both for coincident or posterior E:R relationships, and expressions that are irrealis, future, habitual, iterative, directive or law. (Cohen 2013, 275–277).
Bo Isakssen claimed qatal is perfective and “anterior aspect” (Isaksson 2014; Isaksson 2013). He distinguished between short yiqtol, which is sometimes modal, sometimes narrative perfective, sometimes general present (but is not a preterite), and long yiqtol, which shares the same meaning as weqatal and weyiqtol.
Andersen and Forbes claimed “the tense and aspect of a clause is determined not only by the characteristics of its verb but also by relations between it and other clauses in the discourse of which it is part … there is no point in artificially assigning tenses and aspects to the verbs as such” (2012, 38).
Eep Talstra claimed the yiqtol is durative; it indicates relevance for the reader, and should be translated as a present tense. The weqatal continues a series begun by yiqtol and indicates progression.
My more limited work on biblical Hebrew indicates that the verbs are not selected for their tense or taxis. Even though wayyiqtol is very consistently Absolute Past, Relative Present, and R<S, those semantic values are also commonly expressed by yiqtol, weqatal, and participles. And certainly by Binnick’s definition of “aspect” (Binnick 1991) most of the verbs do not inflect for aspect, although the participle is consistently imperfective. Imperfective statements are commonly expressed using qatal and wayyiqtol. Only if “imperfective aspect” is redefined to include the future tense and irrealis modality as Driver and Garr did can the yiqtol be considered strongly correlated with imperfective aspect (Garr 1998, 29f, 41f, liv, lvi). If one parameter must be chosen to the exclusion of the others, then of the four possibilities (namely, that the verb forms inflect for E relative to S, E relative to R, E overlapping with R, and modality), the hypothesis that accounts most simply for the a biblical author’s choice between a recto (qatal and wayyiqtol) and obliquo (yiqtol and weqatal) form is that modal events (including the future, posterior, and habituality) are expressed using one of the obliquo forms, and all other events are expressed using one of the recto forms.
My best hypothesis (i.e., the one that accounts for the data most completely) is that the choice (whether conscious or not) that biblical authors made regarding the form of the verb followed a decision tree that can be simplified into something like the following. First, the author considered whether the event was modal (irrealis, including future and habitual events) or not (realis). If the event is modal, then performative (qatal), directive (imperative), repetitive (participle if present tense and subject-fronted), and imminent future (participle) are special cases. Otherwise, if something other than the verb needs to be fronted (for example, for subordination, negation, adverbs, topicalization), use yiqtol; but for clause-initial position, use weqatal. On the other hand, if the event is not modal/future, the present tense of dynamic verbs is a special case (yiqtol). Otherwise, if something other than the verb needs fronting, use qatal, and otherwise use wayyiqtol.
Hence, my position is most closely in agreement with those that highlight modality (Buth, Joosten, Hatav, Notarius, Cohen, and Isakssen), and furthest from those who highlight aspect (Cook, Eskhult, Pardee, Robar) or discourse (Andersen & Forbes, and Talstra).
Ken, while discussing this this interview with a few others, Drew Longacre posed a couple additional questions on which we would love to hear your thoughts.
How accurately do the written sources reflect the spoken Hebrew of the time? Does the verbal system of the Dead Sea Scrolls result from natural linguistic development or intentional archaism? Do the Dead Sea Scrolls attest to a single common verbal system, or are there numerous parallel systems evident in the sources?
This is hard to establish. I would suggest two registers: one archaizing literary, and another spoken (MMT), which is naturally developing towards Mishnaic Hebrew. The verbal syntax of most Qumran documents does not agree with Mishnaic Hebrew against Biblical Hebrew. However, there is one exception to this generalization. 4QMMT does show similarity with MH in two respects: (1) its increased use of participles, and (2) its relatively high mood-prominence.
Like the Mishnah, MMT exhibits a vastly increased preference for the participle, especially compared to Biblical Hebrew. The reasons for this increase are uncertain, but there is a clear pattern in which active participles are used in clauses expressing imperfective actions (i.e., when the event time extends beyond the reference time).
The strong mood prominence is the other major distinctive of verb functions in MMT. All yiqtols in MMT are modal, and all qatals are non-modal. This is in agreement with the observation of E.Y. Kutscher that in MH, the yiqtol has become a modal form. The modality-prominence is evident only for the finite verbs; the participles do not fall neatly along modal lines.
These two similarities, namely the frequency of the participle and the modality-prominence of the finite verbs, are not sufficient to classify the language of 4QMMT as “Mishnaic Hebrew.” In a few other respects MMT and the rest of the corpus bear more resemblance to Biblical Hebrew than to the Hebrew of the Mishnah. For example, weqatal in MMT is always used for the future (seven times), never the past. But of all the documents in this corpus, MMT is the one that most closely resembles the Mishnah with respect to the use of the verb forms.
Ken, thank you so much for taking the time to answer a these questions, and thank you for your research. I always find your work to be thorough, clearly presented, and rigorously argued — a model of scholarship we would do well to imitate.
About the Author:Brian W. Davidson
Brian W. Davidson is a PhD student studying Old Testament at the Southern Baptist Theological Seminary. His research interests primarily have to do with Hebrew, Aramaic, and Greek grammar, textual criticism, and the textual history of the Old Testament, with a special focus on the Septuagint.
Website:Brian W. Davidson
Galia Hatav, The Semantics of Aspect and Modality: Evidence from English and Biblical Hebrew (Amsterdam ; Philadelphia: John Benjamins Publishing Company, 1997), 120. ↩
Galia Hatav, The Semantics of Aspect and Modality: Evidence from English and Biblical Hebrew (Amsterdam ; Philadelphia: John Benjamins Publishing Company, 1997), 121. ↩