A Discourse Analysis of First Corinthians	Ralph Bruce Terry
	Home | Chapters: 1 | 2 | 3 | 4 | 5 | 6

Clause Word Order

Greek is a synthetic language rather than analytic; that is, grammatical slots, such as subject and object, are shown by inflectional case endings, rather than by word order. This means that word order in Greek is free to serve other functions, usually on a discourse level. Roberts has noted, "the classical writers make the freest use, for rhetorical effects (such as emphasis, euphony, variety, etc.), of the departure from normal order which, in an inflected language, is usually possible without ambiguity" (1912, 178). But this raises the question of what was normal word order in a clause for Koiné; Greek.

Although it was possible to form clauses in all six permutations of subject, object, and verb (SVO, SOV, VSO, VOS, OVS, OSV) in Koiné Greek, in actual fact certain constructions were preferable to others. In general, those orders in which the subject preceded the object were used more frequently than those in which the object preceded the subject. Roberts refers to the subject-object order as "natural" and the object-subject order as "unnatural" (1912, 178). Evidence from this study will bear that out. This is the same order that Friberg found in his study of word order in the gospel of Luke (1982, 204-206; cf. Callow 1983a, 13).

But when the question arises as to which of the three possible subject-object word orders is normal, there is little consensus among Greek grammarians. Blass has suggested that predicate-subject-object is the normal Greek word order for New Testament narrative, but Debrunner qualifies this by noting that the verb in initial position is "common only with verbs of saying" (Blass, Debrunner, and Funk 1961, 248 [§472]). Debrunner further suggests that the good number of clauses with verb initial order in Mark are due to Semitic influence in that book (Blass, Debrunner, and Funk 1961, 248 [§472]). Robertson (1934, 417) refers to Blass with approval, broadening the restriction from narrative to prose, but then qualifying his statement to say that spontaneity is the only unalterable Greek rule.

Moulton and Howard (1920, 416-417) refer to Wellhausen and Norden as authorities who note that the verb first in a clause is a sure sign of Semitism, but they cite Lagrange and Torrey as experts who oppose this. They also give the results of Kieckers' research that suggests that the normal order for the verb in classical literature was primarily in the middle position, followed by the final position, and last of all in initial position, while the New Testament gospels vary from this in that initial position is used more than final (Moulton and Howard 1920, 417-418). Turner (1963, 347) suggests that the normal order of ancient Greek was subject-object-verb, and that the verb initial order is an idiosyncrasy of Biblical Greek through Semitic influence. In this, he refers to Rife (1933, 250), who, restricting himself to main declarative clauses with noun subjects and objects, found that classical Greek favored the SOV word order followed closely by the SVO word order. By contrast, he shows that the Greek translation of the first five books of the Old Testament slavishly follows the Hebrew word order, restricting itself to VSO most of the time (Rife 1933, 250). Turning to the New Testament, Rife's results are shown in Table 29, supplemented by data from the ANACLAUS databases for James and I Corinthians.

According to Table 29, the primary word order for both narrative and hortatory books is SVO, followed by VSO in the narrative books and SOV in the hortatory books. For object emphasis orders, OVS is preferred to both VOS and OSV for both narrative and hortatory books. Rife's point is to show that the gospels are not translations of an Aramaic original, but Matthew's gospel approaches the 65.7% SVO order of Biblical Aramaic from the book of Daniel (1933, 251).

The analysis of Greek word order, however, cannot be limited to statistical tabulations of data. As Turner notes, "requirements of emphasis will everywhere upset rules of word-order" (1963, 348). And Robertson says, "emphasis consists in removing a word from its usual position to an unusual one" (1934, 417).

PRIMARY WORD ORDER FOR MAIN DECLARATIVE CLAUSES

IN SELECTED NEW TESTAMENT BOOKS

                                                               
| Book         |  VSO  |  SVO  |  SOV  |  VOS  |  OSV  |  OVS  |
| Matthew      | 12.5% | 62.5% | 12.5% |   -   |   -   | 12.5% |
|              | (n=1) | (n=5) | (n=1) | (n=0) | (n=0) | (n=1) |
| Mark         | 16.7% | 50.0% | 22.2% |   -   |  5.6% |  5.6% |
|              | (n=3) | (n=9) | (n=4) | (n=0) | (n=1) | (n=1) |
| Luke         | 23.1% | 48.7% | 20.5% |  5.1% |   -   |  2.6% |
|              | (n=9) | (n=19)| (n=8) | (n=2) | (n=0) | (n=1) |
| John         |   -   | 40.0% | 30.0% | 10.0% | 20.0% |   -   |
|              | (n=0) | (n=4) | (n=3) | (n=1) | (n=2) | (n=0) |
| Acts         | 13.3% | 68.9% |  2.2% |  2.2% |  2.2% | 11.1% |
|              | (n=6) | (n=31)| (n=1) | (n=1) | (n=1) | (n=5) |
| Narrative    | 15.8% | 56.7% | 14.2% |  3.3% |  3.3% |  6.7% |
|    Books     | (n=19)| (n=68)| (n=17)| (n=4) | (n=4) | (n=8) |
| Romans       |   -   | 40.0% | 50.0% |   -   |   -   | 10.0% |
|              | (n=0) | (n=4) | (n=5) | (n=0) | (n=0) | (n=1) |
| I Corinthians|   -   | 52.9% | 23.5% |  5.9% |  5.9% | 11.8% |
|              | (n=0) | (n=9) | (n=4) | (n=1) | (n=1) | (n=2) |
| James        | 11.1% | 44.4% | 44.4% |   -   |   -   |   -   |
|              | (n=1) | (n=4) | (n=4) | (n=0) | (n=0) | (n=0) |
| Hortatory    |  2.8% | 47.2% | 36.1% |  2.8% |  2.8% |  8.3% |
|   Books      | (n=1) | (n=17)| (n=13)| (n=1) | (n=1) | (n=3) |

Data are restricted to main declarative clauses with noun subjects and objects.
Data for the first six books are taken from Rife 1933, 250. Data from Matthew, John,
and Romans are restricted to a sampling of the first ten occurrences only.

In the following analysis, it will be shown that the usual word order positions differ from clause to clause according to the syntactic constructions of each clause. The two primary clause orders apart from discourse considerations are SVO and SOV.

In recent years, some progress has been made in identifying factors on a discourse level that cause changes in word order in Greek. Friberg (1982, 9-10) has classified these factors into four categories: syntactic, semantic, pragmatic, and stylistic. An example of his syntactic category is the relative clause, which regularly begins with a relative pronoun. Among the semantic factors he lists are emphasis and topicalization. His pragmatic factors include afterthought, interruption, and the hearer's or reader's ability to process the information signal. Finally, he lists variation as a stylistic factor.

To this list, Radney adds the concept of motif, which he defines as "a usually-recurring item (i.e., character, prop, or concept) in a text that contributes in some prominent way to the theme of a larger grammatical unit" (1982, 110; reprinted 1988, 60). Callow (1983b, 15) sees this as comparable to Friberg's category of general emphasis. Other factors which Radney identifies as influencing word order include relative and interrogative clauses, pronouns, correlative constructions, negation, and the introduction of new subjects (1988, 13, 19, 27, 40, and 48).

A factor which neither Friberg nor Radney considers but which Rife (1933) has shown to be very real is the influence of the original order of a translated text. The Hebrew VSO order has markedly affected word order in the translation of the Old Testament into Greek. This in turn produced a word order for "Biblical Greek" which seems to have influenced writers trying to sound scriptural (cf. Moulton and Howard 1920, 478). This influence of another language on a synthetic language such as Greek is also seen when the writer is a speaker of Greek as a second language. The word order of a primary language can have an ordering effect on the order of Greek. This means that native speakers of Hebrew would prefer a VSO order (Turner 1963, 348), native speakers of Aramaic a SVO order (Rife 1933, 251), and native speakers of Latin a SOV order (Wheelock 1960, 15) in their use of Greek. The fact that Coptic has a VSO word order (Plumley 1977, 143) helps explain why Rife found that the Greek papyri, mostly from Egypt, had a 40% VSO word order (1933, 250), which is an unusually high percentage with the exception of translation Greek from a Hebrew original.

Now although Friberg and Radney have done New Testament discourse studies a service by suggesting discourse level factors that help to account for Greek word order, they have both suggested that the underlying unmarked word order for Greek is VSO (Friberg 1982, 207; Radney 1988, 2; cf. Callow 1983a, 20-21; Callow 1983b, 4, 18). This is less than desirable for several reasons.

First, word order typology is traditionally based upon dominant order rather than underlying order. Greenberg's first universal principle of word order applies to the "dominant order" of "declarative sentences with nominal subject and object" (1963, 61), which is very similar to the restrictions with which Rife (1933) worked. The reason for this restriction is that certain syntactic categories show a different word order. Greenberg's universals 11 and 12 relate to different orders that questions may have (1963, 65). Comrie (1981, 83) has pointed out that clauses with pronouns may also have a different word order than those with nouns. He also notes that there has been some discussion as to the basic word order of German; the dominant order for main clauses is SVO, while for subordinate clauses it is SOV (Comrie 1981, 83). Comrie goes on to point out that "surface structure typologists tend to opt for main clause order as less marked, while transformational-generative grammarians tend to opt for subordinate clause word order as more basic" (1981, 83). It is possible to maintain that SOV is the underlying order that is transformed to SVO through a rule that the verb or its auxiliary must come second in a main clause. But even here, the underlying order is dominant in subordinate clauses and the debate is about which dominant order is basic. The word order VSO is not dominant for any syntactic category in Koiné Greek, as the next point shows.

Second, the statistical data seems to be against taking Greek as a verb initial language. This will be demonstrated below for I Corinthians from the ANACLAUS database, but even Friberg's and Radney's research illustrate it. Of 55 clauses in Luke that have noun subjects, simple noun objects, and indicative verbs, only 15 (27.3%) show VSO word order (Friberg 1982, 30, 33-34). In fact, taking all clauses in Luke that show both a single overt subject and a verb, only 481 of 1179 (40.8%) clauses show the verb initial (Friberg 1982, 191). Now in defining the unmarked case for subject, object, and indirect object orders, Friberg appeals to the greater frequency as defining the unmarked case (1982, 204, 207). It is only in defining verb-subject order that Friberg abandons his principle that greater frequency indicates the unmarked case (1982, 192).

Radney too must deal with the fact that not many examples of VSO occur in the book of Hebrews. He gives only two examples (Hebrews 4:2 and 4:4; cf. Radney 1988, 11) and comments on "the outlandish rarity of VSO clauses in the data" (Radney 1988, 12). One of his examples is a quotation from Genesis (which is highly VSO in order). To be sure, the Hebrew original and Septuagint translation do not have an overt subject (they are VO order). But the preceding clause in Genesis 2:2 is VSO with subject spreading into the clause quoted, and Radney overlooks the possibility of a Biblical Greek word order as mentioned above being used to make the quote sound scriptural.

Third, two major factors given by Friberg for the frequency of the subject preceding the verb in Luke are problematic. The first of these factors is topicalization (Friberg 1982, 197-199). Callow (1983a, 18) notes that Friberg accounts for 31% of fronted subjects using this device. The problem with this is in separating the topic from the subject. To be sure, there is such a thing as object topicalization (cf. Foley and van Valin 1985, 355-358). But the unmarked situation is that the subject is the topic of a sentence; even Friberg admits this for his unmarked VSO order (1982, 197). But if this is the case, in what sense can the subject of a SVO clause be said to be topicalized in a way different than the subject of a VSO clause? Even a VSO clause must have a topic. The problem is so severe that Radney (1988, 13) disavows using the term topic to explain subject fronting. He rightly notes, "there is the question of why marked topicalization is necessary at all, since the grammatical subject of a clause should automatically be interpreted as the topic" (Radney 1988, 13).

The other problematic factor that Friberg identifies for subject fronting is emphasis. To be sure, there is such a thing as emphasis. When an object is fronted, it is emphasized because this is a less usual position. But if the subject before the verb is the more usual case, in what sense can a subject be emphasized by moving it from the less common to the more common position? If Robertson is correct in saying, "emphasis consists in removing a word from its usual position to an unusual one" (1934, 417), the shift of a subject from a position following the verb to one preceding it would serve to deemphasize that subject.

Fourth, the identification of Greek as a dominantly VSO language tends to be an abstract interpretation, since that order is not reflected as the principal word order in any Greek text studied by scholars except for those that are translations of a Hebrew original (cf. the chart in Rife 1933, 250). There is the problem that any abstract solution can be interpreted in alternative ways. In addition, if topicalization is allowed as a legitimate reason for fronting a subject, English could be abstractly interpreted as a VSO language in which the subject is fronted whenever it is the topic. Since the subject in English is almost always the topic, this could explain the fact that English shows SVO order in its surface structure. An abstract interpretation such as this is unacceptable for English. By analogy, it should not be acceptable for Greek either.

Finally, Friberg chooses the VSO word order as primary because he states that "attempts to explain the subject following the verb are in vain" (1982, 192). Actually, there are several reasons why the subject may follow the verb. It has already been noted that Debrunner points out that many of the cases of the subject following the verb occur with verbs of speaking while others are due to Semitic influence (Blass, Debrunner, and Funk 1961, 248 [§472]). In fact, the two VSO independent clauses in I Corinthians that have both a noun subject and accusative noun object (in 6:16 and 15:45) are quotes from Genesis 2:24 and 2:7, respectively. And both of them have oblique objects in Greek; that is, their objects are objects of the preposition ει�"ς 'into' which is functioning as a literal translation of the Hebrew direct object marker et. This is hardly standard Greek form.

There is, in fact, evidence that Luke's Greek in both the gospel of Luke and the book of Acts imitates the biblical style of the Greek Old Testament, the Septuagint (LXX). In his commentary on Luke, Leon Morris notes, "From 3:1 on the Gospel is written in a type of Hellenistic Greek which is strongly reminiscent of the Septuagint" (1974, 26). And in his Greek commentary on Acts, F. F. Bruce states, "Luke can use this 'Biblical Greek' as readily as any other writer. . . . it is found most abundantly in those parts of his work where he has a Palestinian or Jewish setting" (1951, 28). Turner also notes "the Jewish Greek of some parts of the Gospel and the early chapters of Acts," but he is not sure whether this is due to Luke's skill for writing in "a deliberate LXX style" or whether this was "perhaps Luke's natural speech" (1976, 56). This Jewish or Biblical Greek style may account for the relatively high percentage of verb initial clauses which Friberg found in Luke and which Levinsohn (1987, 3) also found in Acts. To be sure, there is a possibility that this higher percentage means that narrative and non-narrative texttypes have different underlying or even dominant word orders, but until stylistic factors are accounted for, it would be premature to conclude this. Just as translation Greek in the Old Testament has produced a narrative style that is more characteristic of Hebrew than of Greek, so imitation of that style may have produced a modified word order.

Turning to I Corinthians, there are seventeen clauses that have VSO word order. All of them are marked by grammatical features that are usually considered secondary. Four of the VSO clauses use a subjunctive verb (3:4; 12:15, 16; 15:28); two use an imperative verb (10:15 and 11:28); and one has a participle (9:20). In line with the tendency noted above, six of the seventeen are used with verbs of speaking (3:4; 7:10; 12:15, 16; 15:12, 35). In addition, five are used with forms of the equative verbs εἰμί 'be' and γίνομαι 'become' (6:16; 9:20; 12:19; 15:28, 45). Five are used with cognitive verbs (1:21; 7:16 bis; 10:15; 11:28) and one with a depictive verb (1:20). No VSO clauses use the more common action or stative verbs. Two VSO clauses are used in quotes; seven are used in dependent clauses; four are used in questions; and four are used with negatives. At least nine of them have a clause or a quotation as their object. Perhaps more significantly, thirteen of the seventeen VSO clauses occur as the first clause in a colon. This distribution of VSO clauses across many secondary features is hardly what one would expect to find if VSO were in fact the underlying unmarked order.

Further, when all possible cases of the four types of verb-subject order (VS, VSO, VOS, and OVS) are considered, the following items are statistically significant as opposed to the four types of subject-verb order (SV, SVO, SOV, OSV) in I Corinthians. VS order is found in 32% of the dependent clauses that precede the independent clause, as opposed to less than 20% of independent clauses. VS order is used in 67% of adverbial clauses in I Corinthians, as opposed to only 25% of main clauses. The subjunctive mood is used 41% of the time in VS order, while the indicative mood is used in VS order only 24% of the time. The imperfect tense is found 43% of the time in VS order, while the present is found in VS order less than 18% of the time. The middle and passive voices are used with VS order 30% of the time, while the active is used only 19% of the time with VS order. Verbs that relate to motion are used 43% of the time in VS order, while those that relate to creative activity are used 56% of the time in VS order. By way of contrast, action verbs are used in VS order only 14% of the time and stative verbs are used with VS order less than 13% of the time.

Clauses with compound subjects appear in VS order 56% of the time as opposed to 23% for clauses with singular subjects. Clauses that have mixed first person and third person subjects are in VS order 67% of the time, while clauses with just third person subjects are in VS order just 22% of the time. Clauses with pronoun objects occur in VS order 38% of the time and clauses with no object use VS order 26% of the time. These figures contrast with clauses with noun objects, which occur in VS order only 6% of the time. When new information is introduced in a subject slot, the clause is in VS order 53% of the time. This is in contrast with clauses with the same subject as previous clauses, which have VS order only 15% of the time. And previous clause order seems to have some effect on VS ordering. The VS order is found 57% of the time following VSO order and 41% of the time following VS order (that is, in a clause with no object). This tendency has been noted by both Dover (1960, 53, 56) and Levinsohn (1987, 3). On the other hand, VS order clauses occur following SVO order clauses 19% of the time, after SOV clauses 21% of the time, after SV clauses 16% of the time, after VO clauses 15% of the time, after OV clauses 20% of the time, and after V clauses (those with neither overt subject nor object) 22% of the time.

To be sure, the mere quoting of percentages does not prove that VS order is not unmarked. But these items with high VS order are mostly secondary features. The chi-square test shows that for all of them the difference between these frequent occurrences and the less frequent ones of more primary features has less than a 5% probability of being due to random distribution. For some of them, the percentage of use of VS is so high that these features become factors in explaining why the VS order is used in these cases.

It is beyond the scope of this dissertation to formulate rules as to why a particular order is or is not used in certain situations. This analysis is limited to the influence on word order of the syntactic and semantic variables in the ANACLAUS database for I Corinthians. However, the material presented below will suggest that there are good reasons for supposing that a verb-subject order is secondary and thus marked. There is no reason why attempts to find justification for using VS order in Greek should be in vain. The data above suggest several possible reasons. Friberg's argument that there can be no other explanation is not valid.

Now if Koiné Greek was not in fact a VSO language, the question remains as to what its unmarked order was. The following data for I Corinthians will suggest that for epistolary material the underlying order (if such exists for a synthetic language) was either SVO or SOV, the two dominant orders. It is difficult to decide between the two since there are preferred orders for different situations. It would be as easy to write rules to move from SVO to SOV as from SOV to SVO since each order predominates in mutually exclusive syntactic constructions. It would seem that for each of several such constructions, there was an unmarked order.

Table 30 lists the distribution of clauses throughout I Corinthians into preceding dependent clauses, independent clauses, and following dependent clauses for each of fourteen different types of word order. The category labeled other in the chart represents clauses with missing components; it contains units that English teachers have traditionally labeled as "sentence fragments." It is perhaps stretching the definition of the term clause to include these units here, but they function as clauses with implied constituents.

From Table 30 it is possible to see that the largest category of Greek clauses are those that contain neither an overt subject nor object but only a verb. The next three orders in terms of frequency are VO, OV, and SV with nearly equal numbers of clauses. The fifth most frequent order is SOV; it is the most frequently used order with all three main constituents present. The sixth most frequent order is SVO. These six categories of clauses, each of which has over a hundred examples in I Corinthians, account for more than 78% of the clauses in I Corinthians. The other eight categories in descending order of frequency are "other" (the miscellaneous category), VS, SO, OVS, OS, OSV, VSO, and VOS. Of the fourteen categories, VSO is next to last in frequency.

When all fourteen categories are used to try to determine significant differences in word order, twenty-four of twenty-six variables show statistically significant differences of some sort. In order to bring the analysis under control, it is useful to restrict the study to those clauses that contain all three major constituents: subject, object, and verb.

WORD ORDER DISTRIBUTION IN I CORINTHIANS

                                                           
|  Word Order  |  Preceding  |  Independent  |  Following  |
|  Type        |  Dependent  |  Clauses      |  Dependent  |
|  V           |     39      |      164      |      61     |
|  VO          |     25      |      131      |      37     | 
|  SVO         |     15      |       77      |      16     | 
|  VSO         |      5      |       10      |       2     | 
|  VOS         |      1      |       10      |       2     | 
|  OV          |     20      |      136      |      36     |
|  SOV         |     16      |      105      |      12     |
|  OSV         |      0      |       13      |       5     |
|  OVS         |      2      |       21      |       4     | 
|  SV          |     24      |      135      |      27     | 
|  VS          |     18      |       39      |       8     |
|  SO          |      4      |       32      |       8     | 
|  OS          |      1      |       26      |       0     | 
|  other       |     13      |       71      |       7     | 
|  Totals      |    183      |      970      |     225     |

There is a good theoretical reason to do this as well. Table 31 shows that there is a marked difference in the percentage of verb initial clauses that occur with objects and without objects, that is, with transitive and intransitive verbs. This holds true for data from both Luke (primarily a narrative text) and I Corinthians (primarily a non-narrative text). For both books, the difference is statistically significant. The fact that clauses without objects are at least 11% more likely to have initial verbs raises serious questions about the validity of determining verb-subject-object order by looking at all clauses with verbs and subjects. In passing, it is worth noting that the difference in verb ordering between Luke and I Corinthians is statistically highly significant. Whether this difference is caused by a different dominant word order between narrative and non-narrative text or by Luke's use of a style in imitation of the Greek Old Testament is beyond the scope of study of this dissertation.

THE EFFECT OF OBJECTS ON WORD ORDER DISTRIBUTION

IN LUKE AND I CORINTHIANS

                                                      
|                 |      Luke       | I Corinthians  |
|                 |                 |                |
|                 |  VS    |  SV    |   VS   |  SV   |
|                 |        |        |        |       |
|With Objects     | 31.5%  | 68.5%  |  19.9% | 80.1% |
|                 |(n=133) |(n=289) | (n=36) |(n=145)|
|                 |        |        |        |       |
|Without Objects  | 42.7%  | 57.3%  |  33.3% | 66.7% |
|                 |(n=212) |(n=284) | (n=45) |(n=90) |
|                 |        |        |        |       | 
|Totals           | 37.6%  | 62.4%  |  25.6% | 74.4% |
|                 |(n=345) |(n=573) | (n=81) |(n=235)|

The data for Luke are taken from Friberg 1982, 30, 191. For comparison
purposes with Friberg, the data are limited to third person indicative
verbs.

When the data are limited to clauses with all three major constituents, six variables show differences between their features for different word orders that are highly significant: clause type, clause relationship, subject type, object type, object case, and texttype. In the following analysis, SV is a combination of SVO, SOV, and OSV; VS of VSO, VOS, and OVS; VO of SVO, VSO, and VOS; OV of SOV, OSV, and OVS; SO of SVO, SOV, and VSO; and OS of OVS, OSV, and VOS. These combinations are also used in the following two tables.

Clause type distinguishes four types of clauses: independent, dependent, quotations, and relative clauses. Among these, independent clauses show a 64.8% usage of OV order; dependent clauses show a 51.7% usage of OV order; quotations show a 68.8% usage of VO order; and relative clauses show a 66.7% usage of VO order. These are different enough to be highly significant on an VO-OV comparison.

There are seven different kinds of clauses found in the variable labeled clause relationship: main, subordinate, conditional, embedded, adjectival, adverbial, and absolute. The latter two can be omitted from the study of three element clauses since there is only one such example of an adverbial clause and only two such examples of absolute clauses. Among the remaining five clause types, main, subordinate, and embedded favor an OV order with percentages of 64.1%, 60.7%, and 50.7%, respectively. Conversely, the OV order of conditional and adjectival clauses drop in percentage to 46.4% and 22.2%, respectively; that is, these clause types have more VO clauses than OV clauses. The difference is statistically significant. In addition, none of the conditional clauses has the object fronted before the subject. Only 6.8% of the embedded clauses (i.e., clauses serving in subject or object slots) show this kind of object fronting. The other three kinds of clauses have over 20% of their objects fronted before their subjects. The difference in distribution of object fronting is highly significant. This may indicate that conditional clauses do not allow object fronting.

Table 32 presents the distribution of the six full word orders over the four types of text found in I Corinthians. In each case the two most common orders are SOV and SVO. For narrative, hortatory and persuasive texttypes the most common order is SOV; for expository it is SVO. The percentage of SVO clauses varies at most 12.5%, which is hardly significant. The occurrence of SOV, however, varies somewhat more—from a high of 53.1% in hortatory to a low of 17.6% in expository. The combination statistics at the bottom of the table show that expository is the only texttype in which more objects are found after the verb than before it. It is also marked by much more object fronting than the other texttypes. This object fronting makes the distributions of subject-object ordering highly significant. The other combinations are not significant.

It is difficult to tell whether the varied distribution by texttype is due to the texttype itself or to other factors based on the types of clauses found within each texttype. To the extent that word ordering may be determined by verb semantics and this factor influenced by topic of discussion, the distribution may vary greatly from discourse to discourse. It is only after several texts of Koiné Greek have been analyzed that the significance of the distribution shown in Table 32 will be apparent.

TEXTTYPE AND WORD ORDER IN I CORINTHIANS

                                                                       
|       |                          Texttype                            |
| Word  |                                                              |
| Order |  Narrative   |  Expository   |  Hortatory    |  Persuasive   |  
| SVO   |  37.5% (n=3) |  43.1% (n=22) |  30.6% (n=45) |  34.5% (n=38) |  
| VSO   |              |               |   4.1% (n=6)  |  10.0% (n=11) |             
| VOS   |              |  13.7% (n=7)  |   2.7% (n=4)  |   1.8% (n=2)  |  
| SOV   |  50.0% (n=4) |  17.6% (n=9)  |  53.1% (n=78) |  38.2% (n=42) |  
| OSV   |              |  11.8% (n=6)  |   2.7% (n=4)  |   7.3% (n=8)  |  
| OVS   |  12.5% (n=1) |  13.7% (n=7)  |   6.8% (n=10) |   8.2% (n=9)  |  
| Total |           8  |           51  |          147  |          110  |
| VO    |  37.5% (n=3) |  56.9% (n=29) |  37.4% (n=55) |  46.4% (n=51) | 
| OV    |  62.5% (n=5) |  43.1% (n=22) |  62.6% (n=92) |  53.6% (n=59) | 
| SV    |  87.5% (n=7) |  72.5% (n=37) |  86.4% (n=127)|  80.0% (n=88) |
| VS    |  12.5% (n=1) |  27.5% (n=14) |  13.6% (n=20) |  20.0% (n=22) | 
| SO    |  87.5% (n=7) |  60.8% (n=31) |  87.8% (n=129)|  82.7% (n=91) |
| OS    |  12.5% (n=1) |  39.2% (n=20) |  12.2% (n=18) |  17.3% (n=19) |

Table 33 combines the important factors from the variables subject type, object type, and object case, with the addition of object article, which has significance for noun objects in the accusative case. Some features listed in Table 18 for these variables have been omitted from this chart since there are too few instances of their use to draw conclusions as to the significance of the data. A complete listing of all clauses in I Corinthians with the three main constituents is found in Appendix D.

In addition, there are several minor patterns that can be inferred from the data which are not presented in Table 33. Participial clauses that have an article functioning as a subject in lieu of a pronoun regularly take an SVO order: of 26 examples in I Corinthians, 20 (76.9%) have an SVO order and 6 (23.1%) have an SOV order.

All eleven relative clauses that have a relative pronoun as the subject follow the SVO order. Of the six relative clauses that have relative pronouns as the object, three have OVS order, two have OSV order, and one (15:36) has SOV order; the latter order seems to be influenced by a vocative (αφρων 'stupid') followed by a second person singular pronoun συ 'you' as subject. In all other cases the relative pronoun comes first in the clause, whether it is subject or object.

There are also five examples of demonstrative adjectives modifying an articular noun subject. In every case but one, the order is SVO and the demonstrative follows the noun; the one exception (11:25) with SOV order is a quote from the words of Jesus in which the demonstrative precedes the noun.

SUBJECT AND OBJECT INFLUENCES ON WORD ORDER IN I CORINTHIANS

                                                                     
|  Object      |                      Subject Type                   |
|  Type        |                                                     |
|              |     Pronouns     |     Clauses     |     Nouns      |
|  Clauses     |  sVO 13  sOV  1  |  SVO  1  SOV -- | SVO  3  SOV    |
|              |  VsO  4  OsV --  |  VSO --  OSV -- | VSO  3  OSV    |
|              |  VOs --  OVs --  |  VOS --  OVS -- | VOS --  OVS  1 |
|              |  VO 94%  OV   6% |  VO 100% OV   0%| VO  86% OV  14%|
|              |  SV 78%  VS  22% |  SV 100% VS   0%| SV  43% VS  57%|
|              |  SO 100% OS   0% |  SO 100% OS   0%| SO  86% OS  14%|
|  Articular   |  sVO  7  sOV  7  |  SVO  1  SOV -- | SVO 13  SOV  6 |
|  Accusative  |  VsO --  OsV  3  |  VSO --  OSV -- | VSO  2  OSV  1 |
|  Nouns       |  VOs --  OVs --  |  VOS --  OVS -- | VOS --  OVS  2 |
|              |  VO  41% OV  59% |  VO 100% OV   0%| VO  63% OV  37%|
|              |  SV 100% VS   0% |  SV 100% VS   0%| SV  83% VS  17%|
|              |  SO  82% OS  18% |  SO 100% OS   0%| SO  88% OS  12%|
|  Anarthrous  |  sVO  6  sOV 11  |  SVO --  SOV  2 | SVO  8  SOV 13 |
|  Accusative  |  VsO --  OsV  3  |  VSO --  OSV -- | VSO --  OSV -- |
|  Nouns       |  VOs --  OVs  1  |  VOS --  OVS -- | VOS  1  OVS -- | 
|              |  VO  29% OV  71% |  VO   0% OV 100%| VO  41% OV  59%|
|              |  SV  95% VS   5% |  SV 100% VS   0%| SV  95% VS   5%|
|              |  SO  81% OS  19% |  SO 100% OS   0%| SO  95% OS   5%|
|  Predicate   |  sVN  5  sNV  3  |  SVN  1  SNV  4 | SVN  5  SNV 12 |
|  Nominatives |  VsN  1  NsV  2  |  VSN --  NSV -- | VSN --  NSV -- | 
|              |  VNs --  NVs --  |  VNS --  NVS -- | VNS --  NVS -- | 
|              |  VO  55% OV  45% |  VO  20% OV  80%| VO  29% OV  71%|
|              |  SV  91% VS   9% |  SV 100% VS   0%| SV 100% VS   0%|
|              |  SO  82% OS  18% |  SO 100% OS   0%| SO 100% OS   0%|
|  Predicate   |  sVA  4  sAV  8  |  SVA --  SAV  6 | SVA  4  SAV 11 |
|  Adjectives  |  VsA --  AsV  3  |  VSA --  ASV -- | VSA --  ASV  1 |
|              |  VAs --  AVs --  |  VAS --  AVS  2 | VAS --  AVS  2 |
|              |  VO  27% OV  73% |  VO   0% OV 100%| VO  22% OV  78%|
|              |  SV 100% VS   0% |  SV  75% VS  25%| SV  89% VS  11%|
|              |  SO  80% OS  20% |  SO  75% OS  25%| SO  83% OS  17%|
|  Pronouns    |  sVo  4  soV  9  |  SVo --  SoV  1 | SVo  1  SoV  8 |
|              |  Vso --  osV  2  |  VSo --  oSV -- | VSo  2  oSV  2 |
|              |  Vos --  oVs  2  |  VoS --  oVS  1 | VoS  7  oVS 10 |
|              |  VO  24% OV  76% |  VO   0% OV 100%| VO  33% OV  67%|
|              |  SV  88% VS  12% |  SV  50% VS  50%| SV  37% VS  63%|
|              |  SO  76% OS  24% |  SO  50% OS  50%| SO  37% OS  63%|

Two other tendencies need to be noted. One is the tendency for pronouns to fill slots next to the verb, whether they be subjects, objects, or indirect objects (cf. Radney 1988, 21). The other is for compound objects to come at the last of a clause. These are not hard and fast rules, only tendencies that occur more often than not.

Table 33 shows the distribution of various word orders within clauses for three types of subjects and six types of objects. For each combination, the six types of clause orders are combined to show the percentages of ordering between two of the three elements: subject (S), object (O), and verb (V). Predicate nominatives (N) and predicate adjectives (A) are presented as types of objects. Where any of these constituent elements are pronouns, they are represented in the table and in the description below using lower case letters.

The clearest pattern that emerges from Table 33 is that clauses with predicate adjectives are usually found in SOV order. The equative verb, which in this type of Greek construction is sometimes omitted, does not add a lot of semantic content to the clause and thus seem to prefer the last slot. A similar situation occurs with predicate nominative constructions; most of the clauses are in SOV order. Here, however, the predicate nominative with a pronoun subject seems to be an exception to the rule. There are five examples of an sVN (a type of SVO) order. Closer examination reveals that there are other reasons why these particular clauses follow an SVO order. Three of these counter-examples (1:30; 3:11; and 4:17) are relative clauses with the pronoun subject being a relative pronoun; the order for this type of construction is regularly SVO. In addition, two of these cases (1:30 and 12:27) have compound objects, and as noted above, there is a tendency for compound objects to come last in a clause. The clause in 1:30 contains both a relative pronoun as subject and a compound object. This leaves only one counter-example (11:24) to SOV order and it is a quotation of a reported saying of Jesus. It seems therefore that SOV is the standard order for predicate adjectives and nominatives.

In contrast with predicate adjectives and nominatives, accusative objects show a mixed pattern. Those without a definite article (i.e., anarthrous) modifying the object show a regular SOV word order. There are eight counter-examples to this when the subject is a noun, but four of these eight have a demonstrative adjective modifying the subject, and as previously noted, this factor seems to favor SVO order. When the accusative noun object is modified by an article, two patterns emerge. For clauses with nouns or embedded clauses as subjects, the preferred order is SVO. The three cases with noun subjects and objects fronted to the beginning of the clauses (i.e., the OSV and OVS clauses) are all found in expository texttype. Table 32 shows that expository texttype has over 25% object fronting.

The situation with pronoun subjects and articular accusative noun objects is not so clear. There are seven examples of both sVO and sOV order, and the larger percentage (59%) for OV clauses is caused by the three OsV clauses. These are definite cases of object fronting and should not be considered in deciding between SVO and SOV since object fronting in either could produce an OSV clause. For the sake of simplicity and consistency, it would be nice to formulate an underlying SVO order so that all articular accusative noun objects would have the same order. On closer examination, however, the SOV order would seem to be primary. All of the pronoun subject clauses which occur with OV order for articular accusative noun objects are found in main and subordinate clauses; those with VO order occur in conditional, embedded, and adjectival clauses. Further, three of the sVO clauses are relative clauses with a relative pronoun as subject; such clauses always have SVO order. Thus the primary order in this case seems to be SOV. To the extent that noun and clause subjects can be spoken of as being heavier than pronoun subjects, and articular accusative nouns can be spoken of as being heavier than anarthrous accusative nouns in an object slot, it seems that the lighter constructions favor SOV order and the heavier ones favor SVO order.

An examination of the data with clause objects seems to confirm this tendency. Clause objects would be considered heavy constructions and an SVO order would be expected. This is exactly what is found. The major counter-examples to this view are found among those clauses with noun subjects and clause objects. In 1:28 a clause occurs in OVS word order with a compound object, a combination of a substantival articular accusative adjective and an articular participial clause. This would seem to be a heavy object in the sense spoken of above. The clause in question, however, is a clear case of object topicalization, as an examination of (65) shows.

(65)	τὰ ἀγενη̑	του̑ κόσμου καὶ τὰ ἐξουθενημένα	ἐξελέξατο	ὁ θεός
	the ignoble	of the world and the scorned	chose	the God

The other three counter-examples using noun subjects and clause objects all have VSO word order. But of these, two (12:15 and 12:16) are subjunctive mood in conditional clauses and one (7:10) has a compound subject. As noted previously, 56% of compound subjects in I Corinthians follow the verb. Further, all three are verbs of speech, and Debrunner has noted the tendency for subjects to follow verbs of speech (Blass, Debrunner, and Funk 1961, 248 [§472]). Since there are good explanations for these counter-examples, it seems reasonable to conclude that the primary word order for clauses with embedded clauses as objects is SVO.

Finally, Table 33 shows the order for clauses with pronoun objects. Where both subject and object are pronouns, the primary word order is SOV. One of the four examples of SVO word order (1:8) is a relative clause with a relative pronoun as subject. Since the word order in such clauses is always SVO, this clause should not be considered. In the same way, three of the object fronted clauses (2:8; 3:17; and 10:30) are relative clauses with relative pronouns as objects. As noted above, object fronting is always standard in such cases and these cannot be considered valid counter-examples.

The two examples of clauses with embedded clause subjects and pronoun objects are not enough from which to draw significant conclusions. But since by analogy SOV seems to be the predominant order for all clauses with pronoun objects, and since object fronting seems to result in a secondary order, it seems logical to conclude that further work in other Koiné Greek texts will indicate that SOV is the primary order in this case also.

The most difficult of the combinations is that of clauses with noun subjects and pronoun objects. This is the only combination that results in all six possible word orders. There are several factors that must be considered as contributing to this wide distribution of forms. Four of the eight clauses with SOV forms (7:19 bis and 10:19 bis) are actually predicate nominative clauses with pronoun objects in the nominative case, and it has already been established that SOV is the primary word order for predicate nominative clauses. Also, three of the OVS clauses (2:7; 2:9; and 6:18) are relative clauses with a relative pronoun as object, and this is a standard order for such a case. A fourth (12:28) begins with the relative pronoun οὕς 'whom' used as a demonstrative pronoun meaning 'these' (Arndt and Gingrich 1957, 589); in such a construction, the relative pronoun regularly comes first. In addition, three of the OVS clauses (3:5 bis and 9:18) have the interrogative pronouns τί 'what' and τίς 'what' in the object slot; such questions regularly have fronted objects. Four of the VOS clauses (7:15; 16:19 bis; and 16:20) contain verbs of speech, which tend to take initial position in a clause. When these factors are considered and the clauses listed above are eliminated from consideration, the primary word order seems to be the SOV order that would be expected by analogy with pronoun and clause subjects. At least the SOV order has the most examples unexplained by syntactic considerations.

Furthermore, there are good semantic factors to explain the object fronting in the three remaining clauses with OVS order. The OVS clause in 4:1, shown in (66), is a good example of object topicalization as the topic of Christian workers is resumed after being mentioned in 3:22.

(66)	Οὕτως	ἡμα̑ς λογιζέσθω	ἄνθρωπος	ὡς ὑπηρέτας Χριστου̑
	Thus	us let consider	men	as attendants of Christ
	καὶ	οἰκονόμους	μυστηρίων	θεου̑.
	and	stewards	of [the] mysteries	of God

The OVS clause in 7:17, shown in (67), is a good example of object emphasis when the distributive pronoun is used.

(67)	ἕκαστον	ὡς κέκληκεν ὁ θεός
	each	as has called the God

Last, the OVS clause in 12:11, shown in (68), is a good example of summation, as the previous items listed are brought together.

(68)	πάντα	δὲ ταυ̑τα ἐνεργει̑ τὸ ἓν καὶ τὸ αὐτὸ πνευ̑μα
	all	but these works the one and the same spirit

In all these cases, the subject is God or the Spirit of God. Since God can be assumed to be the ultimate actor, the mention of God can be placed in the final slot in the sentence without doing damage to the sense. In fact, it is quite common when God is the actor with a passive verb simply to omit any mention of Him (Nida and Taber 1974, 114). He can be inferred from the context.

In summary, there is not a single dominant word order for all clauses of Koiné Greek. Rather, the primary and preferred word order seems to be dependent upon the syntactic construction of the clause. SOV is the primary word order for predicate adjective and predicate nominative clauses, as well as for clauses with anarthrous accusative direct objects and pronoun direct objects. It also seems to be the preferred order when the clause contains an articular direct object and a subject pronoun.

SVO is the primary word order for clauses with embedded clauses as objects and for clauses with articular accusative direct objects and without pronoun subjects. In some sense, these clauses have heavy objects that tend to come at the end of a clause.

At this point, it is impossible to tell whether SOV or SVO should be considered the primary underlying word order for Koiné Greek. Since Friberg has shown in relating his data to universals that "Greek is largely consistent with our expectations of the behavior of a VO language" (1982, 334), it might be better to favor SVO. But the search for an underlying word order is beyond the scope of this dissertation.

The point here is that each of these orders in the constructions listed above should be considered unmarked. The way to produce emphasis is to move a constituent from its normal word order. Since these orders are the most common, they should be considered normal, and thus orders that differ from these are emphatic and also marked, in some sense. But it is only after syntactic considerations have been given that discourse roles such as those presented in (66) through (68) can have any real significance in marking a text.

Finally, no attempt has been made in this study to search for word order differences between foregrounded and backgrounded text. Some languages mark the main line with a different word order; Angwak is an SVO language of Africa that marks the storyline in narrative text with SOV word order (Longacre 1990b, 87). Longacre has suggested that something similar happens in Greek, perhaps even marking narrative storyline with the VSO order that several investigators claim to be primary (1993, personal communication). This suggestion certainly merits further investigation.

Back to top | Next: Chapter 5.4 Quotations