Clause Immediate Constituent Composition, Incidence, and Ordering
11.3 Clause Immediate Constituent Ordering
11.3.2 A Descriptive Approach to Positional Syntax
The inconclusiveness of the foregoing leaves us with a question: what approach(es) to visual-izing positional syntax should we use? Our approach to positional syntax is descriptive. For predi-cator root / binyan / voice combinations that occur frequently enough to allow meaningful charac-terization, we generate horizontal bar charts showing various patterns of CIC sequencing. Our goal is to show what is in the texts, not to account for it (at least not at this point in our investigations).
11.3.2.1 Nucleus, Core, and Periphery in Role and Reference Grammar
The basic concepts upon which our description relies have been described by Robert Van Valin, Jr. 17 Van Valin’s role and reference grammar benefits greatly by being the result of his address-ing the question: “What would laddress-inguistic theory look like if it were based on the analysis of lan-guages with diverse structures such as Lakhota, Tagalog and Dyirbal, rather than on the analysis of English?” 18 In response, Van Valin introduces 19
a very different conception of clause structure . . . from that assumed in other approaches. . . . [T]he conception of clause structure that it posits [is] equally applicable to free-word-order, flat-syntax lan-guages such as Dyirbal and Malayalam, to head-marking lanlan-guages like Lakhota and Tzotzil . . . , and to fixed-order configurational, dependent-marking languages like English and Icelandic.
To achieve this sort of coverage, Van Valin relies on his concept of “the layered structure of the clause.” Its full explication consumes much of his book. For our purposes, we need to understand only what he calls “the primary constituent units of the clause.” They are (we quote): 20
• the ‘nucleus’, which contains the predicate (usually a verb), 21
• the ‘core’, which contains the nucleus and the arguments of the predicate,
• a ‘periphery’, which subsumes non-arguments 22 of the predicate, e.g., setting locative and temporal phrases.
“The universal aspects (the nucleus, core, periphery and clause) are all semantically motivated.” 23 Now, here is a crucial point: 24 “Since these hierarchical units are defined semantically and not syn-tactically, . . . the elements in these units may in principle occur in any order, if a given language permits it.” We emphasize that Van Valin’s terminology relates to semantic definitions of nucleus, core, and periphery. Contrary to the spatial imagery that inheres in the names of the hierarchical units (“core,” “nucleus,” “periphery”), they do not imply a particular grouping in surface structure.
In carrying out our investigation of these matters, we will also need to be very diligent in de-tecting ordering constraints that have nothing to do with the semantic-to-syntactic mapping but, rather, result from other constraints. 25 For example, for presentational reasons, objects of address
17. R. D. Van Valin Jr., Exploring the Syntax-Semantics Interface (Cambridge: Cambridge University Press, 2005) 3–8.18. Ibid., 1.
19. Ibid., 3–4.
20. Ibid., 4.
21. “Predicate” here equals “predicator” in our terminology.
22. “Non-arguments” here equals “adjuncts” in our terminology.
23. Ibid., 8.
24. Ibid., 5.
25. Note well: we are describing the ordering of constituents. We are not seeking to account for the ordering, although interesting—if premature in our view—efforts have been made to do so. In this connection, see Christo H. J.
(i.e., speeches) overwhelmingly are clause final. This assists cognition, since for all but the short-est utterances, the processing of “and Moses said [long speech]” is far simpler than would be the processing of “and Moses [long speech] said.”
11.3.2.2 The Positional Syntax of Grammatical Function CICs
The Descriptive Categories Adopted. The number of CIC sequences attested in Biblical He-brew is vast. The 68,089 verbal clauses exhibit 9,623 patterns. This is because we make use of so many CIC categories: three impermanent, three syntactic isolate, six predicator, four operator, four GFs, and 41 SRs—61 categories in all. To examine the positional syntax of GFs, we extract clause cores by removing from the clausal CIC sequences all non-GF / non-predicator types:
1. Each underdetermined “and” is part of a clause only because we have not promoted it to cue phrase status. The other impermanents are so few that their presence or absence can have little effect upon the results of our analysis (see §9.3.1.1).
2. The syntactic isolates, being isolated, are excluded from the clausal sequences (see §9.3.2).
Altogether, these items account for 1.1% of the CIC population.
3. We exclude the operators on the grounds that 98% of them either appear immediately before the verbs that they operate upon or are clause initial.
van der Merwe and Eep Talstra, “Biblical Hebrew word order: The interface of information structure and formal fea-tures,” Zeitschrift für Althebräistik 15 / 16 (2002–3) 68–107.
Predicator Type Rank #1 Rank #2 Rank #3 Rank #4 Rank #5 Rank #6
Anchored VO 32% V 23% VS 16% VSO 7% VIO 4% VSIO 3%
Unanchored V 26% VO 18% VS 12% SV 10% OV 6% SVO 5%
All
4. To reduce the repertoire of CIC sequence patterns, we delete resumed arguments.
5. For the present investigation of the positional syntax of grammatical-function CICs, we delete the SR CICs, leaving only the predicator CICs and GF CICs.
If a verbal clause has either an anchored or an unanchored predicator, and if it may have no more than one subject, direct object, indirect object, and / or complement, then one can show that there are 261 possible core sequence (GF sequence) patterns. In fact, we find 189 core sequence patterns in Biblical Hebrew. Of these patterns, 63 occur once and 23 occur twice.
11.3.2.3 Core Patterns across the Qal Active in Biblical Hebrew
We plot percentages in horizontal bar charts across all Qal active clauses with anchored or un-anchored predicators. We use 19 frequently occurring sequence patterns plus “other.” Note that in all cases lacking a verb (‘V’) in clause-initial position, the relative frequency is higher for the unan-chored verb corpus than it is for the anunan-chored verb corpus (for example, the SVO pattern is far more common when the Qal active verb is unanchored than when it is anchored). See chart, p. 162.
So much for core CIC sequence patterns across the Qal actives of Biblical Hebrew. We are much more interested in the comparative behavior for specific verb root corpora. In chaps. 12–15, we provide side-by-side bar charts of core CIC ordering patterns, verb corpus by verb corpus.
Predicator Type Rank #1 Rank #2 Rank #3 Rank #4 Rank #5 Rank #6
Anchored VO 67% V 33% – – – –
Unanchored VO 27% V 27% OV 17% SVO 8% VS 5% VOS 5%
ץפח
11.3.2.4 Core Patterns for the ץפח Corpus
In upcoming chapters, we show bar charts for clauses with anchored or unanchored predicators.
By way of example, the relative frequencies for the ordering patterns in the ץפח corpus are shown in the chart above, p. 163.
The left-hand bar chart is misleading, because it presents the CIC sequences in the mere three clauses in the ץפח corpus that contain anchored predicators. Because the sample size for the an-chored corpus is so small, we are well advised to avoid making comparisons between the unan-chored and anunan-chored corpora. We do observe, however, that in 32% of the ץפח corpus, the predica-tor is not clause initial (i.e., the sequence is OV, SVO, SV, OVS, or IV).
11.3.2.5 SRs amid GFs
There is one final matter that we investigate in chaps. 12–15: with what frequency do SR CICs appear physically amid the GF CIC constituents? We proceed as follows:
1. Divide the SR CICs into six groups. Table 11.6 shows our chosen grouping.
Table 11.6. The Six SR CIC Groups
Other Participant
2. Across each verb corpus, find all contexts where GF CICs (the predicator and its GF arguments) are not contiguous because of “interpolated” SR CICs.
3. Document these contexts using filled-in versions of table 11.7.
Table 11.7. SR Interpolations To see our perspective, consider this English clause: “Jean said ‘Hello’ to Mike at 6 a.m.”: sbj – vb – obj addr – ind obj – time point. In this clause, the GF CICs are contiguous: sb – vb – address –
ind obj. The time point SR (“at 6 a.m.”) is physically in the clause’s periphery. If we had “Jean at 6
a.m. said ‘Hello’ to Mike”: sbj – time point – vb – obj addr – ind obj, then the GF CICs would not be contiguous, the time point SR being interpolated between the subject and the predicator.
As it happens, the ץפח corpus has no interrupted GF sequences. Consequently, actual data on SR interpolations first appear in §12.6.2.