The fourth post in a series on Biblical Hebrew Word Order, introduced here.
In the last two posts I introduced and discussed the criteria of frequency and distribution. In this post I will add the criterion of clause type as yet another important filter for the raw word order data.
The Criterion of Clause Type
The second criterion used to filter raw frequency results concerns ‘clause type’. This criterion is predicated on the observation that languages often exhibit different word order patterns in different clause types; in such cases, not all clause types present the language’s basic word order. Consider English interrogative clauses, such as When did Noah leave? This clause type in English has the inflected Verb, did, before the Subject, in contrast to the declarative counterpart, Noah left yesterday. On this basis, we would exclude interrogative clauses as a source for basic word order in English. Moreover, although interrogatives are typically a minority clause type in English texts and so their exclusion would not normally affect the frequency results, we can imagine a text that consists predominantly of questions, resulting in a highly skewed frequency-based analysis for English word order.
Such observations—that one must consider whether the language in question exhibits word order variation according to clause type—have influenced the typological analysis of basic word order from its beginnings (see Greenberg 1963:80). The result is that basic word order is most often identified as the the word order present in “stylistically neutral, independent, indicative clauses with full nouns phrase (NP) participants, where the subject is definite, agentive and human, the object is a definite semantic patient, and the verb represents an action, not a state or an event” (Siewierska 1988:8; see also Mallinson and Blake 1981:125). Notably, Siewierska also indicates that the basic word order of a language need not be identical to the “dominant linearization pattern” (i.e., statistically prevalent word order) in that language (1988:8). She suggests that this may be the result of the vagaries of human communication, in which diverse structures are utilized, or it may be due to a genre convention (1988:11-12). Genre convention is certainly operative in Hebrew with regard to the restricted distribution of wayyiqtol clauses, which I discussed above. The wayyiqtol form is used as the narrative Verb and, unlike the qatal and yiqtol Verbs, is confined to indicative semantics and a past temporal context.
Another implication for Biblical Hebrew that follows from Siewierska’s basic clause definition concerns the presence of overt Subjects. Hebrew, like many languages (Spanish and Italian, for example) allows the syntactic Subject to be omitted. Such languages are often referred to as null Subject or ‘pro-drop’ languages (see Holmstedt f.c.a for an overview). Null Subject languages often exhibit word order differences between clauses with an overt Noun Phrase as the Subject and clauses without an overt Subject (Siewerska 1988:11); similarly, clauses with overt pronominal Subjects often exhibit word order differences from clauses with lexical Noun Phrase Subjects (Dryer 2007:80). Since Biblical Hebrew allows overt Subject to be omitted (4a) and arguably uses overt Subject pronouns for Topic or Focus marking (4b), any discussion of basic word order must draw primarily on clauses that have overt lexical Noun Phrase Subjects (4c).
(4) Types of Subjects in Biblical Hebrew
a. Null: Gen 9:6 בְּצֶלֶם אֱלֹהִים עָשָׂה __ אֶת־הָאָדָם
b. Pronominal: Gen 27:31 וַיַּעַשׂ גַּם־הוּא מַטְעַמִּים
c. Lexical Noun Phrase: Gen 3:1 כֹּל חַיַּת הַשָּׂדֶה אֲשֶׁר עָשָׂה יְהוָה אֱלֹהִים
That clauses with lexical Noun Phrase Subjects are less common in null Subject languages like Hebrew, Spanish, and Italian makes it more difficult but not impossible to identify basic word order clauses in a text. It simply highlights the necessity of using all the criteria together in the investigation of basic word order.
The clause criterion holds a number of additional implications for the careful study of basic word order in Biblical Hebrew. First, it has long been noted that the dominant VS pattern of narrative becomes less than dominant in direct speech (MacDonald 1975). Consider the following numerical data from Genesis:
(5) Narrative versus Speech in Genesis
a. the wayyiqtol is used 1971x in narrative but only 123x in speech (and only 21x with an overt Subject)
b. of 2507 main (non-subordinate) narrative clauses, only 107 are SV (4.3%) while 896 (including wayyiqtol) are VS (over 50%)
c. of 1748 main (non-subordinate) speech clauses, 134 are SV (7.7%) and 200 are VS (11.4%)
The remarkable non-use of the wayyiqtol in direct speech confirms its primary role as the narrative or story-telling verb. Add the radical shift towards near balance of SV and VS in speech texts and it is clear the one’s position on the discourse type will significantly impact the methodology and conclusions for basic word order.
How do we decide which discourse type is a better candidate for representing basic word order? Narrative is typically taken as the determinative discourse type: “If storyline clauses in narrative discourse in a given language are VSO, then that language should be classified as a VSO language” (Longacre 1995:333); this view holds that dialogue introduces complexities that likely depart from basic word order. However, Payne (1995) suggests that “[m]ost claims about word order have undoubtedly been based on narrative data and, without conscious awareness, the typological cubby-holes to which languages have been assigned are likely biased by formal features correlating with temporal sequentiality” (1995:454). In other words, precisely because clauses in narrative are strung together in some sort of temporal order, narrative (rather than direct speech/dialogue/conversation) may exhibit departures from standard word order (see also Downing 1995:20).
Another distinction that may affect the word order discussion is between main and subordinate clauses. Subordinate clauses often appear to be more conservative, that is, they show less syntactic diversity than main clauses (this is what Ross 1973 named the ‘Penthouse-principle’, i.e., that the rules are different if you live in the penthouse = upstairs/main clause). This observation has interesting implications for both word order typology and diachronic syntax. For identifying basic word order, some, like Bickford (1998:214-16), argue that subordinate clauses take priority in the identification of basic word order. For diachronic syntax, it has been noted that word order changes in, for example, English, German, and Kru, first took place in main clauses and only later (often much later) applied to subordinate clauses (see Matsuda 1998 and Bybee 2002 for discussion and bibliography). Importantly, if it is established that a diachronic word order change has affected main clauses but not subordinate clauses, it reverses the priority of the clauses for word order typology: the new order exhibited in main clauses should be taken as basic.
Biblical Hebrew, I suggest, should be added to the list of languages that exhibit Ross’ Penthouse principle. As the data in (6) illustrate, there is no doubt that subordinate clauses (6b) are overwhelmingly VS in Biblical Hebrew, even when all the criterion are applied, while the number of SV and VS main clauses (6a) are nearly identical.
(6) Main versus Subordinate Clause Word Order in Genesis
224 SV: Gen 2:6 וְאֵד יַעֲלֶה מִן־הָאָרֶץ
216 VS: Gen 27:41 יִקְרְבוּ יְמֵי אֵבֶל אָבִי
15 SV: Gen 31:32כִּי רָחֵל גְּנָבָתַם
126 VS: Gen 4:25 כִּי הֲרָגוֹ קָיִן
There is also comparative evidence in Northwest Semitic that ancient Hebrew was initially a VS language in both main and subordinate clauses. However, by the post-biblical period, Hebrew exhibits a shift in word order character, such that soon after the turn of the era, rabbinic Hebrew appears to be an SV language. Indeed, it has even been argued that Biblical Hebrew itself exhibits an (early BH) VS-to-SV (late BH) shift (Givón 1977). These lines of evidence converge in such as way as to suggest that Biblical Hebrew experienced a shift to SV order in main clauses while the older VS order was preserved in subordinate clauses. If so, two questions proceed from this. First, if the basic word order of main and subordinate clauses differ, which is to be taken as the typological classification for the language? Second, is it possible to identify when this shift occurred? If Genesis exhibits SV basic word order, as I argue below, then the shift must have occurred earlier than previously argued.
Finally, it should not go unnoticed that Siewierska’s definition of the basic clause type includes a semantic notion: indicative clauses rather than non-indicative (i.e., modal, subjunctive) are better candidates for basic word order. In biblical Hebrew, modal clauses—whether with the morphologically modal jussive, the modal of use of the imperfective yiqtol, or the modal use of the perfective qatal—are consistently VS (7a). In contrast, indicative clauses (excluding the wayyiqtol) are not so clearly VS (7b)—indeed, they are SV by more than 2-to-1.
(7) Indicative versus Modal Clause Word Order in Genesis
164x SV: Gen 2:6 אֵד יַעֲלֶה מִן־הָאָרֶץ
77x VS: Gen 27:41 יִקְרְבוּ יְמֵי אֵבֶל אָבִי
14 SV: Gen 16:1 שָׂרַי אֵשֶׁת אַבְרָם לֹא יָלְדָה לוֹ
102 VS: Gen 1:3 יְהִי אוֹר
As this section has illustrated, the clause type criterion covers a lot of linguistic ground. For Biblical Hebrew, applying this criterion leads one to set aside clauses without lexical Noun Phrase Subjects, work with an awareness that narrative and direct speech exhibit different patterns (primarily due to the conventionalized use of the wayyiqtol in the narrative genre), distinguish between the word order of main and subordinate clauses (and choose which has priority for determining basic order), and look to the order exhibited by indicative clauses rather than non-indicative clauses. Although filtering the date through this criterion requires significant effort and results in a smaller database of ‘basic word order’ clauses than some prefer:
Determining the word order of a language on the basis of such a small sample seems somewhat precarious. More importantly, basic word order in this approach bears little resemblance to the way that the language is most frequently used. (Moshavi 2010:15)
I used this quote in my first post, and pointed out the deep problems with Moshavi’s appeal to language use. Now it is time to point out the fundamentally unscientific attitude reflected in her statement. Regardless whether we “like” the size of the database that results from applying various tests to a raw corpus, we must deal with the facts. Wishing for more, or ignoring the tests that should be applied is hardly responsible scholarship. And since it is clear that Hebrew does pattern differently along each of these divides (Subject type, genre, clause level, and semantic type), it is a filtering process that is absolutely required.
We’re getting there! One more criterion and then a concluding post.
Ahhh … there’s nothing quite so invigorating as arguing that a centuries-old view is just plain wrong.
(No ominous music or devilish chuckles today, just a sigh of ‘almost there’ satisfaction. Even with the best of databases, good scholarship requires checking and re-counting all the examples and my eyeballs hurt from a month of it!)
• Bickfoard, J. Albert. 1998. Tools for Analyzing the World’s Languages: Morphology and Syntax. Dallas: Summer Institute of Linguistics.
• Bybee, Joan. 2002. Main Clauses are Innovative, Subordinate Clauses are Conservative: Consequences for the Nature of Constructions. Pp. 1-18 in Complex Sentences in Grammar and Discourse: Essays in Honor of Sandra A. Thompson, ed. Joan Bybee and Michael Noonan. Amsterdam: John Benjamins.
• Downing, Pamela. 1995. Word Order in Discourse: By Way of Introduction. Pp. 1-27 in Word Order in Discourse, ed. Pamela Downing and Michael Noonan. Amsterdam: John Benjamins.
• Dryer, Matthew S. 2007. Word Order. Pp. 61-131 in Language Typology and Syntactic Description, Volume I: Clause Structure, ed. Timothy Shopen. Cambridge: Cambridge University Press.
• Givón, Talmy. 1977. The Drift from VSO to SVO in Biblical Hebrew: the Pragmatics of Tense-Aspect. Pp. 184-254 in Mechanisms of Syntactic Change, ed. Charles N. Li. Austin: University of Texas Press.
• Greenberg, Joseph H. 1963. Some Universals of Grammar with Particular Reference to the Order of Meaningful Elements. Pp. 73-113 in Universals of Language, ed. Joseph H. Greenberg. Cambridge, MA: MIT Press.
• Longacre, Robert E. 1995. Left Shifts in Strongly VSO Languages. Pp. 331-54 in Word Order in Discourse, ed. Pamela Downing and Michael Noonan. Amsterdam: John Benjamins.
• MacDonald, John. 1975. Some Distinctive Characteristics of Israelite Spoken Hebrew. Bibliotheca Orientalis 23(3/4):162-75.
• Mallinson, Graham, and Barry J. Blake. 1981. Language Typology: Cross-linguistic Studies in Syntax. North-Holland Linguistic Series. Amsterdam: North-Holland Publishing Company.
• Matsuda, Kenjirô. 1998. On the Conservativism of Embedded Clauses. Pp. 255-68 in Historical Linguistics 1997, ed. Monika S. Schmid, Jennifer R. Austin, and Dieter Stein. Current Issues in Linguistic Theory 164. Amsterdam: John Benjamins.
• Payne, Doris L. 1995. Verb Initial Languages and Information Order. Pp. 449-85 in Word Order in Discourse, ed. Pamela Downing and Michael Noonan. Amsterdam: John Benjamins.
• Ross, John Robert. 1973. The Penthouse Principle and the Order of Constituents. Pp. 397-422 in You Take the High Node and I’ll Take the Low Node, ed. C. T. Corum, T. C. Smith-Stark, and A. Weiser. Chicago: Chicago Linguistic Society.
• Siewierska, Anna. 1988. Word Order Rules. Croom Helm Linguistics Series. London: Croom Helm.