by Vern S. Poythress
[Published in Novum Testamentum 26/4 (1984):312-340. Used with permission.]
This article presents rules describing the use of de, oun, kai, and asyndeton in the Gospel of John. The rules are precise enough so that they can be used in a sequel article as the basis for a test of unity of authorship.
Scholars have recognized for a long time that oun and asyndeton are unusually frequent in the Gospel of John, and that John’s usage of them is at times somewhat idiosyncratic.1 But now it has become feasible to give a more thorough explanation of the phenomena. Linguists have, in the past decades, developed analytical tools for studying discourse phenomena, and in particular the interaction between particles and large-scale discourse features. Robert E. Longacre, especially, has pioneered in the analysis of discourse features in “exotic” languages.2 He and his colleagues have demonstrated that in many cases the best explanation for the use and nonuse of puzzling particles and morphemes is to be found in concerns for discourse and paragraph cohesion. Particles can be used in some languages to indicate identity of participants, shift of
participants, paragraph beginning or ending, main events vs. background in narrative, climax, and still other features.
Can the same type of approach help us in understanding some of the peculiarities of the Gospel of John? I think it can. To begin with, oun and asyndeton are best treated in the context of at least two other Greek conjunctions3 used frequently in John: de and kai. When oun and asyndeton are used at the beginning of a sentence, they are used instead of de, kai, or a small number of other conjunctions.4 Their particular meaning or significance is found in their opposition to or contrast with other grammatical elements that might replace them. The idea that meaning resides largely in the paradigmatic contrast of one element over against others has already become familiar in connection with the theory of semantic fields.5 But contrast is an even more important factor in small closed grammatical classes like the class of conjunctions. The true range of use of each particle or conjunction can be effectively assessed only in comparison to its neighbors.
1. The sentence
Not all occurrences of kai and de, however, are relevant. Only those that link two sentences rather than two words, phrases, or clauses are functioning like oun and asyndeton when these occur between two sentences. But what is a sentence in Greek? Different people have used the word “sentence” in different ways. I am using the word in a way close to Longacre’s and Pike’s usage.6 This takes some explanation.
In Greek as in other languages there is a hierarchy of “surface structure” grammatical units. There are in Greek morphemes, words, phrases, clauses, sentences, and discourses. Whole discourses can be broken down without remainder into a string of grammatical units of any one size.7 In particular, the whole Gospel of John consists of a string of sentences (a fragment sentence used as a whole would, of course, count as a sentence for this purpose). Any one sentence is composed of one or more clauses.
A clause in turn can be roughly defined as either (a) a verbless equative or attributive clause with an implicit verb einai, or (b) a grammatical unit with a single central verb, together with other material grouped around it by various types of modification, including especially case relations, or (c) an instance like (b) but with coordinate verbs.
Clauses can be embedded in one another. For example, John 14:4a
hopou ego hypago
and John 14:4b
oidate ten hodon
are both simple clauses. Together they form a single complex clause
hopou ego hypago oidate ten hodon,
in which 14:4a modifies ten hodon of 14:4b. Similarly, in the different construction in 14:5,
is a simple clause functioning as the object of the verb oidamen in the clause
ouk oidamen pou hypageis.
Thus the simple clause is “embedded in” the complex one. From my point of view nearly all infinitives and all participles except periphrastic participles function as the nuclei of clauses. Thus in 14:2 hetoimasai topon hyminis a clause qualifying the purpose of the clause poreuomai. In 14:9 heōrakōs eme is a clause functioning as the
nucleus of the noun phrase ho heōrakōs eme. In 14:21 two coordinate clauses, echōn tas entolas mou and tērōn autas, together form a compound clause functioning in the same way. Similarly, following the introductory kaiof 14:13, the whole verse
ho ti an aitēsēte en tō onomati mou, touto poiēsō, hina doxasthē ho patēr en tō huiō
forms a single complex clause with several simpler clauses making it up. In 14:25 the adverbial participle menōn is a nucleus for the dependent participial clause par’ hymin menōn.
Now we can come back to sentences. A sentence is simply a “maximal” clause, that is a clause not embedded in or modifying a still larger clause, together with the intersentence conjunction at its beginning. The main intersentence conjunctions are asyndeton, de, kai, oun, gar, ara, men, te, alla, dio, tote, hōste when followed by an indicative or imperative, and compounds of these particles (e.g., ara oun, mentoi, menounge, etc.).8 Dia touto should, I think, also be included on the list instead of being treated as a special case of asyndeton.
The definition of sentence above threatens to become circular inasmuch as sentence is defined in terms of intersentence conjunctions. But sentences are identifiable in a partially noncircular fashion by several further features. (1) They have a certain grammatical “closure” not characteristic of all clauses.9 They are “maximal.” (2) an entire discourse can be analyzed into a string of sentences with no remainder. (3) Except when the main clause is an equative or attributive clause with an implicit verb einai, sentences have as a rule one and only one “main” verb or coordinate verbs linked by kai. The main verb cannot be an infinitive or participle. It cannot be a verb governed by a subordinating conjunction like hina, ei, ean, hopou, hote, hou, etc. Nor can the main verb belong to a relative clause. Hence, for example, John 14:10 consists of exactly three sentences, with main verbspisteueis, lalō, and poiei. Hoti normally functions as a subordinating conjunction within a sentence, but there appear to be a few instances where it functions as an intersentence conjunction: John 1:16, 1:17, 8:43, 12:49, 17:8a.
2. Ambiguities with kai
One remaining difficulty with defining the sentence in koine Greek concerns the fact that often kai and occasionally de and alla connect together two clauses within a single sentence (e.g., kai in John 14:17). Kai can also be used to coordinate two words or phrases, and can be used with the sense “even” or “also.” But kai and de also connect larger units. Kai and de occur even at the beginning of paragraphs (John 9:1, 7:37, 11:1, 11:55, 12:20, etc.). Hence it cannot be denied that they function at more than one level.
The chief problem arises with kai. In some instances it is clearly a link between clauses. Thus the kai in John 14:10 links clauses and not sentences. In
egō en tō patri kai ho patēr en emoi estin,
there is only one explicit verb estin, and both parts of the coordination are subordinated to a preceding hoti. Similarly, the first kai of 15:7 links two clauses subordinated to ean. The kai of 14:31 links a hina clause and akathōs clause both modifying houtōs poiō. But, on the other hand, at times kai also clearly links sentences or things larger than sentences (9:1). Hence at times there is potential ambiguity. In 14:8,
deixon hēmin ton patera, kai arkei hēmin,
does kai link two clauses into a single sentence, or link two sentences? Either analysis seems to be consistent with the formal grammatical properties of kai. In this case I prefer to analyze kai as a link between sentences. After all, one verb is imperative, the other indicative. One part is a request, the other a statement. The two separate illocutionary10 forces of the two parts are better seen as associated with two sentences than with one. John 15:1,
egō eimi hē ampelos hē alēthinē, kai ho patēr mou ho geōrgos estin,
is a much more difficult case of ambiguity, because the two parts are much more closely coordinated to one another. In all such cases of real ambiguity I prefer to treat kai as an intersentence conjunction. The frequency of its appearance in long strings in narrative
(particularly in Mark) seems to favor this approach. Such long strings of coordinated material do not occur in any instance that is unambiguously at the clause level (in distinction from the sentence level).
We can also quickly eliminate from consideration the uses of kai in the sense “even, also.” Kai occurs in three types of grammatical constructions:
X kai Y
kai X kai Y.
In the last two of the three, X and Y must ordinarily be of the same grammatical type, so that they can be properly coordinated. In the first construction only, kai has the sense “even, also.” In the last, it has the sense “both-and.” The second type of construction only is the one that we are interested in at the sentence level. But sometimes kai occurs at the beginning of a sentence in a context in which it may easily have the sense “even, also.” In many such cases it is probably playing a dual role: it means “even” or “also,” but also functions as the link between sentences. No other link is normally required.
Something should also be said at this point about the occurrences of kai in the crasis kagō. Almost all such occurrences seem to be ambiguous between (a) kai as a coordination between sentences and (b) kai with the sense “also,” in “I also.” I will regularly construe these cases in sense (b), since the presence of the emphatic nominative egō seems to point in this direction.11
Before going on, we should take note of the text-critical problems. Intersentence conjunctions seem, on the average, to be even more subject to textual corruption than are other parts of speech. What do we do about this? We cannot naively depend on the weight of external evidence in all disputed cases. A text type showing high value in some other respects may not necessarily be of high value in its witness to intersentence conjunctions, since the forces leading to corruption are somewhat different for different types of textual variation.
Hence, the safest procedure is to try first of all to account for the use of conjunctions at all those places where the textual evidence is unanimous or nearly unanimous. After building up our views on the basis of the clear evidence, we will then proceed to apply the conclusions to the disputed texts as well. The rules we derive from unambiguous cases then become standards for the intrinsic probability in the disputed cases. The rules thereby help us to make decisions about the textual variants. But the argument is not circular because it ultimately rests on the cases where textual testimony is undivided.
In the meantime, we need an unambiguous way of citing instances of occurrence of the conjunctions. For this purpose, I take as a starting point the Greek text of the Nestle-Aland Novum Testamentum Graece, twenty-sixth edition. If there is more than one occurrence of a given conjunction in a verse, the second and third occurrences of that same conjunction will be referred to by an extra parenthetical numeral accompanying the verse number. Thus the three kai‘s in 16:22 will be designated 16:22, 16:22(2), and 16:22(3), in that order. Alternatively, the first kai can also be denoted 16:22(1).
3. Expository discourse and narrative discourse
Now we are ready to restate the major question with respect to the Gospel of John. The whole text of John can be analyzed without remainder into a string of sentences. An additional complexity arises because whole discourses can be embedded in a sentence as the object of a verb of saying (cf., e.g., 6:43, 8:31).12 But this will not bother us. Now between any two successive sentences there will be a conjunction: usually kai, de, oun, or asyndeton. (Asyndeton, the absence of an explicit conjunction, will be treated here as an “entity” in its own right, in a manner more or less analogous to the “zero” morpheme of linguistic theory.) In John gar and alla will also occur occasionally, but little else. Now, what regularities, if any, govern the occurrence of these intersentence conjunctions? In any given spot, under what conditions does one occur rather than any other?
The conditions governing the occurrence of one conjunction rather than another are semantic rather than grammatical. Any one
of the intersentence conjunctions is grammatically permissible between any two sentences. One conjunction occurs rather than another to indicate to the reader meaning-relationships or more narrowly semantic and communicative-content relationships in unfolding of the discourse and its sentences.13
But the uses of the conjunctions are slightly different in the case of two types of discourse: expository discourse and narrative discourse. (Following Longacre,14 I am using “discourse” as a general term for all types of units several paragraphs long within a monologue. The so-called “discourse” material in the Gospel of John, in the terminology of New Testament scholars, would be only a particular subspecies within this family.) Expository discourse, in the terminology of Longacre,15 is discourse primarily integrated in logical, argumentative, and topical fashion. In John, Jesus’ speeches and the Evangelist’s commentary (linking up with the speeches and scarcely distinguishable from them) are expository discourse. But the backbone of the Gospel as a whole is the third- person, predominantly past-tense account of incidents in the life of Jesus. This is narrative discourse.16The expository discourses are almost all speeches embedded in a linguistic sense in the over-arching narrative.
4. Gar and alla
Let us begin with expository discourse in John. Gar and alla are used in the same way here that they are in most of the rest of the New Testament. Roughly speaking, gar is used in order to indicate that the sentence or sentences that it introduces is a reason or
ground or some kind of vague logical support for what precedes.17 In my terminology for prepositional relations, gar indicates a relationship of implication-grounds or result-reason (in Beekman-Callow’s terminology, conclusion-grounds or result-reason).18
Next, alla is used between sentences to separate two strongly contrasting statements.19 For one thing, alla is nearly always used in separating two sentences that say (roughly) the same thing, first negatively and then positively.20 John 15:16 is a good example. The propositional relation here is negation-of-negative/positive.21
Alla is also the conjunction to use for “intensification”: to introduce a statement that greatly heightens or exceeds the preceding in John, only 16:2).24 It also connects two strongly contrasting statements, often contrasting almost to the point of contradiction (e.g., John 14:31, 15:21, 15:25, 16:6, 16:7, 16:12, etc.). The usual propositional relation expressed in this case is positive-negative Beekman-Callow’s “contrast”).25 As Abbott points out, with this
5. De in expository discourse
This last use of alla is difficult to separate clearly from de. De is also used to separate contrasting statements. But in the case of de the contrast is “weaker.” How do we characterize this “weakness”? In general, alla is used to eliminate or almost eliminate the previous statement, whereas de “balances two opposed ideas.”27 Moreover, when alla separates two sentences, the contrast is between the whole statements. When de does so, more often than not the contrast is focused on an individual phrase of word within the statements. Thus at 14:26 de points to a contrast focused on Jesus vs. the Paraclete. Contrast this use of de with the use of alla at 14:31. In 14:31 it is not a matter of contrast focused on two individual elements, but strong global contrast between Jesus’ freedom from Satan on the one hand (14:30) and his self-surrender to those on Satan’s side on the other (14:31). Nevertheless, the range of usage of de certainly does border on that of alla. There are intermediate cases where it may depend on very subtle preferences on the part of the writer. Does he or does he not prefer to mark the contrast more intensively using alla? Does he or does he not wish to view the contrast primarily as global (alla) or primarily as focused on individual elements (de)?
The over-all rule, then, is that de is used between successive sentences whenever the writer wishes to indicate that two elements in the two sentences are in contrast. If, however, two statements are globally in contrast, or if the contrast is “strong,” alla rather than de will be used. Moreover, the above circumstances are the only circumstances where de is used in expository discourse in John. At this point de is much more restricted in its use in John than in non-Johannine expository discourses in the New Testament. Outside of John, de is used in circumstances of contrast, but it is also used in many instances where the element of contrast is quite weak or almost invisible. Thus from Acts we have de connecting sentences at
(e.g.) 3:14, 3:23, 3:24. In Paul, Rom 2:3, 2:5, 3:7, etc. In Hebrews, 1:6, 2:6, 3:17, 3’18, etc. So also James 1:4, 1:5, 1:6, 1:9, etc. As an oversimplification, we may say that in the bulk of the New Testament de should be translated sometimes as “but” and sometimes as “and.” In the expository discourses of John, however, it can consistently be translated “but.”
6. Exceptional uses of de in expository discourse
There are three significant exceptions to the above generalizations about de. (1) First, contrary to the above rule, asyndeton is sometimes used instead of de between two sentences with a contrast, even when it is a contrast focused in one element of each sentence. The asyndeton is so used in one circumstance only: namely, when both sentences are quite short and nicely balance one another rhetorically. An example of this is found in John 15:5,
egō eimi hē ampelos, hymeis ta klēmata.
Also note 4:22,
hymeis proskyneite ho ouk oidate· hēmeis proskynoumen ho oidamen….
(Cf also 6:63(2), 8:15(2).)
(2) Second, de is used when otherwise we would expect a kai when kai already appears near the beginning of the sentence with the meaning “even, also” (John 15:27, 18:18(2)). This is presumably a stylistic alteration in order to avoid having two colliding kai‘s.
(3) Third, de sometimes occurs together with forms of the demonstrative pronoun houtos, if this houtos occurs at the beginning of a sentence and refers to something after it (3:19, 6:39, 17:3). (Reference downwards in the text makes the pronoun “kataphoric,” as opposed to the more common anaphoric pronoun.) But asyndeton and kai also occur with kataphoric houtos (15:12, 15:17, 1:19). It is probably best at this point in research to say that there is no clearly perceptible rule with respect to kataphoric pronouns. This is understandable. In terms of the cohesive properties of the text, kataphoric pronouns inevitably represent somewhat exceptional cases. They dofunction to increase textual cohesion by tying together with what follows. However, anaphoric pronouns are the more frequent cohesive device, and their effects on sentence joining are very different (cf. § 8(1)).
With these qualifications, then, we can account for virtually every occurrence of de in the expository discourse of John. But there are still a few questionable occurrences of de. These are almost all ambiguous cases where demight or might not be contrastive. That is it might mean “but” or it might represent merely a bland continuation (compare English “and”). Because the great bulk of cases of occurrence of de in John are demonstratively contrastive, it is better to interpret potentially ambiguous cases as indeed contrastive and translate them with English “but.”
7. Oun in expository discourse
In expository discourse, the use of oun in John is very like its use outside John. Oun functions to indicate a logical inference.28 In the terminology of prepositional relations, oun typically indicates a relationship of reason-result or grounds-implication.29 Its function is similar to that of ara. But John does not use ara. So in effect oun covers whatever ground oun and ara together cover in other NT literature.
8. Kai in expository discourse
The use of kai in John is more difficult to describe. In instances where kai is used to connect sentences, there are several different overlapping circumstances where it occurs. (1) Kaiis used to connect two successive sentences expressing closely related ideas. The ideas are topically related to one another, and the prepositional relation is usually that of coordination.30 Kai is particularly preferred in connecting short clauses of this kind: the clauses often have in common not only a topic but some individual words (e.g., 14:21(2), 14:28, 15:1, 2, 16). But for cases of looser coordination than this, asyndeton is used (e.g., 14:7, 10). This includes some cases where there is some degree of tension between the two sentences, and one might insert “and yet” between them (e.g., 8:49, 57).31 Moreover, asyndeton is preferred before forms of the
(2) Kai is used when the second sentence is a kind of step-wise addition to the first, either intensifying the idea in the first, adding information about one aspect only of the first, or making a meta-linguistic comment or other parenthetical comment about the first. Thus we have as examples of intensification 14:3, 12, 13, 17:26; adding information about one aspect 14:4, 30; making a meta-linguistic statement, 14:24, 29, 17:13; another type of parenthetical statement, 16:3, 17:10.
Finally, (3) kai is used in sequences of short quasinarrative sentences within an over-all framework of expository discourse. Thus the kai‘s at 14:16(1), 14:16(2), 14:21(2), 21(3), 14:23(2), 23(3), 23(4), 15:6(1), 6(2), 6(3), 6(4), 6(5), should be viewed as constituting parts of small-scale narratives within the exposition of chapters 14-17. Even the kai at 14:8 constitutes a minimal narrative. A rapid rehearsal of successive connected events is regularly expressed using connecting kai‘s. That is, within the overall framework of expository discourse, kai links any series of sentences which refer to a series of chronologically successive events connected by at least vague causal relations.
9. Asyndeton in expository discourse
Finally, what about the use of asyndeton? Asyndeton, in my view, is the “unmarked” conjunction in expository discourse in the Gospel of John. That is, it is the construction used whenever there is no special reason for usingde, oun, kai, or some other conjunction. Or, to put it another way, it is the “default” option. One uses asyndeton in default of any positive ground for using some other conjunction. In this respect it can be compared to the aorist aspect in Greek. The aorist is the unmarked or “zero” aspect in Greek. It is
Some vague connection can still be seen between the asyndeton in John and the asyndeton used elsewhere. But asyndeton is far more frequent in John. This is partly because his mode of expression and logical organization produce propositional relations suitably expressed by asyndeton. For instance, the propositional relation general-specific (Beekman-Callow generic-specific)34 lends itself to realization using asyndeton. But the range of usage of asyndeton has also been expanded. At the same time the range in which de is used is noticeably contracted. Thus we may compare John with the rest of the New Testament in terms of the amount of “semantic space” allocated to the conjunctions oun, kai, de, and asyndeton. Diagram 1 gives a simplified result.
10. De in narrative discourse
Now we can go on to analyze the use of de, oun, kai, and asyndeton in narrative. To some extent we can refer back to the results for expository discourse. But narrative offers some complications not observed in expository discourse.
For the sake of clarity, it is best to confine ourselves first to the use of conjunctions within narrative episodes. The use of conjunctions between narrative episodes will be discussed afterwards (§ 14).
Gar and alla show no peculiarities. What about de? The patterns for de are the same as those observed for expository discourse in John, together with three main further uses. (1) First, de is regularly used at the beginning of a sentence in connection with the idiom ho de and hoi de functioning as a subject (e.g., John 2:8, 18:7, 20:25). It always indicates a switch in subject. (Constructions with a noun, pronoun, participle, etc., in the nominative case following ho de (e.g., ho de theos) do not count as instances of this construction.) If it were not for the presence of this idiom, the sentences in question would presumably have begun with asyndeton. Unfortunately, I am unable to give a rule for defining accurately when ho de or hoi de
|John||Other NT Expository Discourse|
as a complete subject is used instead of a proper name or noun phrase (the latter two are statistically far more common).
(2) The second exceptional use of de is its use to introduce a sentence or sentences that do not continue the main line of events, the narrative “backbone.” De introduces parenthetical information, background information, or explanatory information, almost always information given in the imperfect tense.35 John 18:10, “the slave’s name was Malchus,” is an excellent example of parenthetical information introduced by de. John 19:41, giving information about the tomb in which Jesus was laid, is hardly parenthetical in the same way. But it too is introduced by de and given in the imperfect, because it is not a continuation of the main events of the narrative. Similarly 20:24 is an essential bit of information, necessary to the understanding of the events to come, but not itself an event in the main line of narrative development.36
(3a) Under this use of de we should also classify instances where touto de, houtos de, or the like introduces an explanation of the significance of some statement or event (7:39, 11:51, 12:33; cf. 12:6). Possibly the uses ofkataphoric demonstratives with de are similar to this (3:19, 6:39, 17:3; cf. § 6(3)).
(4) De normally occurs in the construction hōs de when a temporal hōs clause begins a new sentence (e.g., 2:9, 6:12, 7:10, 8:7). But this rule can be overridden by the rules § 11(1) and § 11(2) below (see 4:40, 11:6, 18:6, 20:11, 21:9). This use of de should probably be seen as related to the use of de at the beginning of new major narrative episodes (cf. § 14 below).37
(5) There is still one noticeable exception to the above rules about using de. Contrary to rule (3) above, asyndeton rather than de introduces the parenthetical notes about the hour (1:39, 4:6, 19:14).
11. Oun in narrative discourse
Now what about the use of oun in narrative?
(1) Of course, oun occurs where the ordinary sense “therefore” is required (e.g. 4:40, 13:6, 20:2). but this is relatively seldom. Oun has three main uses in narrative.
(2) Oun regularly occurs at the point in a narrative when the narrator returns to the main line of events after a digression, a parenthesis, or the supplying of background information (cf., e.g., 18:6, 11, 16, 19, 19:21). This is what Blass-Debrunner have in mind when they describe a certain use of oun as “resumptive.”38 But they do not mention that this use is particularly characteristic of John.39 Abbott, along with some commentators, has noted that “a parenthesis is frequently followed by a resumptive oun.”40 Phyllis M. Healey states things in more linguistic terms:
I have observed that oun marks resumption of the narrative in other cases than parentheses marked by de. It seems that any non-narrative material embedded within a Narrative Paragraph or a Dialogue Paragraph within a Narrative
I need qualify Phyllis Healey’s remarks in only one respect. Oun is not always used in resuming the main line of events. Rather, oun is the unmarked conjunction (the default option) to use in the resumption of the main line of events. That is, oun is used except when for some reason it makes sense to resume using kai (e.g., 9:16, 18:15(2), 20:4) or asyndeton (e.g., 1:41, 4:7), because of the rules with respect to these conjunctions.
(3) Third, oun is the unmarked way of continuing the narrative whenever there is a shift to a new agent in the action described in the sentence immediately following oun. For example, in 18:24 the action shifts from Jesus as agent (18:23) to Annas as agent (18:24). Hence an oun occurs at the beginning of 18:24. Likewise in 18:28 the action shifts from Peter as agent (18:27) to those guarding Jesus (18:28). 18:29, 31, 33, 19:5, and many other cases are of this type.42 This use also includes many of the instances of the use of oun to mark the end of background material. But the latter use cannot be collapsed into the former. Oun sometimes marks the end of background material even when a shift of agent is not involved (e.g. 4:4).
Nevertheless, oun by no means occurs at every point where there is a shift of agent. I have said that oun is the unmarked conjunction to indicate such a shift. It is used only when particular circumstances do not invite the use of some other conjunction. De, for example, can be expected to occur instead of oun in cases where there is a notable element of contrast (7:31), and can be expected to occur with the construction ho de, hoi de. Similarly, kaiis used instead of oun when it connects closely following events (e.g. 1:21(1), 21(2), 21(3), 2:7, 8). And kai can be used to introduce an additive comment, even when there is a shift of agent (e.g. 9:39).
There are some other notable circumstances where oun does not occur along with a change of agent, (a) When the first action of a new agent is introduced with a verb in the historical present (especially legei), asyndeton is the usual (unmarked) conjunction
(e.g. 1:39, 41, 45, 48, 2:5). But this rule does not apply when the verb in historical present comes after an introductory participle (e.g. 6:5, 19, 18:3, 19:26). When the historical present closes a parenthesis, the use of oun appears to be optional (4:9; but cf. 1:45, 2:7).
(b) When a dialogue reply is introduced with the verb apokrinesthai,43 or when it is an answer to an immediately preceding direct question, the usual (unmarked) conjunction is asyndeton (e.g. 1:23, 8:25, 18:25). But when the question does not immediately precede the response, or when the response is not a direct quotation of an answer, or when the response does not really answer the question, oun begins the response (6:32, 53). Even as thus qualified, rule (b) has exceptions, namely when apokrinesthai begins a long, major discourse (5:19, 7:16), or when the full meaning “therefore” is desired at the beginning of the sentence. In these cases oun is used withapokrinesthai.
(c) When a shift of agent immediately follows a command, the >conjunction can be asyndeton (4:50, 5:15, 6:35; but cf. 6:13,6:21).
(d) When a sentence begins with an anaphoric use of the demonstrative pronoun houtos or ekeinos, asyndeton is the unmarked conjunction (e.g. 1:41, 3:2, 22, 4:47, 5:1, 6, 14).
(e) In the construction with temporal hōs clause beginning a sentence hōs de occurs (§ 10(4)).
The above analysis certainly presents us with an odd collection of patterns for the use of oun. What rational processes could possibly motivate the development of such patterns? At first glance the use of oun to mark the close of a parenthesis and the resumption of the main line of narrative events appears particularly paradoxical. Such an occurrence of oun marks a break rather than a connection. Isn’t such a use at the diametrically opposite extreme from the familiar causal and inferential meaning of oun? So it might seem. But I think that deeper reflection can help to reverse the picture.
1 maintain that, at least in a general way, oun is used by John to suggest or intimate that there is a quasicausal relationship involved. Oun occurs at just those points in the narrative where, without such a suggestion, the narrative would be in some danger of falling into pieces. For example, the end of a parenthesis, or the end of a piece
of text giving background information, often constitutes a point where there occurs a wide shift in topic. The presence of oun can assure the reader. Of course, oun does not assure him that what follows is directly related to what immediately precedes. But it assures him that it is directly related to something preceding, whether immediate or further back. In such a way it might be seen as appropriate at the close of a parenthesis. It says, as it were, “Now we take up again on the same narrative, not an independent one.”
The same reasoning can at least begin to account for the patterns with respect to the occurrence of oun with a shift of agent. Any shift of agent is at least a mild break in the narrative, a break over which a bridging word might be useful. But in certain cases, like answers to questions, responses to commands, and in the occurrence of an anaphoric pronoun, the connection is already strong enough without oun. Hence oun is omitted.44
12. Kai in narrative discourse
Now let us move on to a consideration of the occurrences of kai in Johannine narrative. (1) First, kai connects two successive sentences in narrative when the agent or most prominent participant in the two main actions is the same (usually, this means when the subjects of the two main verbs are the same, e.g. 2:15(1), 15(3), 15(4), 16). The agent remains basically the “same” even when
there are slight shifts in the number of participants included. For instance, Jesus is the agent of katebē in the first part of 2:12, while the disciples and Mary are included with him in the second part of 2:12. This is a case with the “same” agent, and so kai connects katebē and emeinan in 2:12. (See also 7:32.)
(2) Second, kai occasionally occurs even in cases where the agent or most prominent participant shifts (e.g. 1:20, 21(1), 21(2), 21(3), 2:11(2)). 1:20(1) and 1:25(1) are probably both cases where the first kai is part of a both-and construction, already discussed in 2 above. But most other instances are cases of what I would like to call “close narrative continuation.” By using kai‘s the effect is created of a single whole block of events following in simple fashion directly one after the other. In almost all cases of such use of kai, the individual sentences so combined are very short. But it appears difficult if not impossible to give hard and fast rules in such cases to specify at exactly what point kai will cease to be used and oun used instead.
13. Asyndeton in narrative discourse
Finally, what about the use of asyndeton in narrative discourse? The fundamental rule for asyndeton can be formulated the same way for narrative discourse as for expository discourse. In both cases, the most rationalized account of the occurrence of asyndeton explains it as a “default” conjunction. It appears whenever the patterns pertinent to all the other conjunctions do not specify their appearance. The situation with regard to asyndeton in narrative is nevertheless materially quite different from the situation with regard to expository discourse. In expository discourse the other conjunctions taken together do not “capture” nearly all the possibilities for sentence relationships. A great deal is therefore left for asyndeton, and it is statistically quite frequent. In narrative, on the other hand, oun and kai in the nature of the case take up almost all the “space” between them. In consequence, much less is “left” for asyndeton, and it is statistically less frequent. In the case of narrative, therefore, it becomes feasible to list one by one rules covering the entire range of usage of asyndeton.
For the most part, the results have already been stated obliquely in the discussion above (§ 11) concerning exceptions to the use of oun. For the sake of clarity, I restate the patterns here. I also add some further minor uses of asyndeton that it was not convenient to
mention up to this point (see (5) and (6) below). Asyndeton occurs between sentences in narrative in the following cases.
(1) Following a shift in agent, when the first verb in a sentence is in the historical present, the usual (unmarked) conjunction is asyndeton (e.g. 1:39, 41, 45, 48, 2:5).45 When the historical present appears just after the end of a parenthesis or of background material, the use of asyndeton (rather than the ordinary oun) appears to be optional.
(2) When a dialogue reply is introduced with the verb apokrinesthai, or when a dialogue reply is an answer to an immediately preceding direct question, the usual (unmarked) conjunction is asyndeton (e.g. 1:23, 18:25). But when the question does not immediately precede the response, or when the response is not a direct quotation of an answer, or when the response does not really answer the question, oun begins the response (6:32, 53). Even as thus qualified, rule (2) has exceptions, namely when apokrinesthai begins a long, major discourse (5:19, 7:16), or when the full meaning “therefore” is desired at the beginning of the sentence (9:25?). In these cases oun is used with apokrinesthai.
(3) When a shift of agent immediately follows a command, the conjunction can be asyndeton (1:47, 4:50, 5:15, 6:35). But oun occasionally occurs as well (6:13, 6:21).
(4) When a sentence has near its beginning an anaphoric use of the demonstrative pronoun houtos or ekeinos, asyndeton is the unmarked conjunction (e.g. 1:41, 3:2, 22, 4:47, 5:1, etc.).46 But over against this one must remember that there is a (marked) use of de when touto begins a sentence explaining the significance of some previous event. In such cases the de rather than asyndeton usually occurs (but cf 9:22, 12:41). Moreover, kai can occur in cases where here is close narrative continuation (kai touto eipōn 11:28, 18:38, 20:20, 22, cf 11:43).
(5) In a few instances, it appears to me that asyndeton occurs in connection with a shift in agent, instead of the normal (unmarked) occurrence of oun, for still another reason. Namely, it indicates that here is a major upset or disturbance of some kind in the causal connection of events along the main line of narrative. The upsets are of
at least two kinds: interruptions to the main causal sequence, and events in the sequence but motivationally alien to what precedes.
First, there are interruptions. In 2:17 the remark about the disciples’ remembering interrupts the main narrative concerning the conflict between Jesus and the Jews. Similarly, 4:31 introduces a rather long interrupting episode having to do with a dialogue between Jesus and his disciples (4:31-38). It comes in the midst of episodes concerning Jesus and the Samaritans. 8:27 constitutes an obviously interruptory or parenthetical remark about the Jews in the midst of exposition by Jesus.
Second, there are events motivationally alien to what precedes. I include 8:42, 52, 58, 11:25, 35. Perhaps 1:42(2) is a further case. It is perhaps significant that there is a cluster of such instances in the latter part of John 8, perhaps the number-one place in the Gospel where the utterances of the Jews are most “alienated” from the utterances of Jesus next to them. The use of asyndeton in such cases becomes intelligible if we recall the hypothesis above (§ 11) that oun is used to “bridge over” places in the narrative where discontinuity might be felt. To leave out the oun is therefore to emphasize the discontinuity.
(6) As an exceptional case, asyndeton joins lists such as 7:41, 9:9(1), 9(2), 9(3), 10:21, 12:22(2), 29, and the textual variants of 7:41(2), 7:12, 9:16. This is probably the best way of explaining 1:21(2) (the asyndeton before ho prophētēs) as well.
14. Use of conjunctions between larger narrative episodes
Between larger narrative episodes, the patterns of use of the conjunctions appear to be the same, except in one important family of cases. The rules change whenever there is an explicit causal gap between the episodes. Such a gap can be indicated by the introduction of a new participant, by a note as to a new locale (but not one reached by a journey), or by a chronological phrase or clause indicating a new time frame. Hote clauses, hōs clauses, and genitive absolute don’t in most cases count for such purposes, because they don’t generally indicate a causal break or gap (e.g. 4:40, 8:30). But sometimes they do indicate a break across which there is little causal connection (e.g. 7:14).
In the nature of the case, we might expect a change in rules to occur in situations where there are causal gaps. Episodes separated
by time gaps or space gaps don’t usually exhibit the same causal relations as events in the same time-and-space frame. Therefore, conjunctions indicating quasicausal relationships can be expected not to function in quite the same way. What, then, are the patterns for use of conjunctions between such episodes? They can be summarized in a number of rules.
(1) Preceding the phrase tē epaurion use asyndeton. Perhaps the phrase itself is regarded as forming a sufficient equivalent to a conjunction.
(2) With meta tauta or meta touto use asyndeton. This agrees with the earlier rule concerning the use of asyndeton with anaphoric demonstrative pronouns. (There is one exception, 19:38. This seems impossible to explain unless the occurrence of de is interpreted as a contrastive use—de indicates a contrast between the soldiers’ and Joseph of Arimathea’s treatment of the body of Jesus.)
(3) Apart from cases (1) and (2), use de.
There are two exceptions (2:1, 13) where kai occurs instead of de, apparently contrary to (3). The kai at 2:13 is appropriate because of the unusually close connection with the preceding ekei emeinan ou pollas hēmeras. The last sentence of 2:12 really takes the action right up to the time period specified in 2:13. Hence, despite the presence of an explicit sentence about the time, there is no causal gap. On the other hand, 2:1 appears to be a genuine exception. I have no explanation for it.
The oun at 20:19 is also difficult to explain. Its effect, we may say, is to bind together more closely than normal the two sets of resurrection appearances 20:1-18, 20:19-29. (20:19-23 and 20:24-29 should already be seen as bound together by their relationships in content; the key figure of Thomas serves as transition in 20:24.) Evidently the author desired to connect closely the resurrection appearances. Perhaps this is in itself sufficient explanation.
We must also add a fourth rule with regard to background material occurring at the beginning of new episodes.
(4) When the first sentence (or sentences) of an episode is devoted wholly to giving information about space, time, or participants, it means that such a sentence is not part of the main line of narrative events. However, because the sentence is at the beginning as a setting, it does not count as a parenthesis and is not subject to the rules for parentheses introduced by de and closed by oun. After the sentences giving the setting, the narrative usually picks up
with kai, since there is not yet any shift of prominent participants (2:3, 2:13(2), 5:1, 10:23, 11:55).
Note, however, the exceptions. Oun occurs instead of kai at 12:21 and 20:25. In both of these cases, however, the participants introduced in the preceding sentence are not completely new. The Greeks of 12:20 are a subgroup out of the group already introduced in 12:18-19. Thomas of 20:24 is part of the group of disciples in 20:19-23, though he happens not to have been present. Hence it is at least doubtful whether the new episodes beginning in 12:20 and 20:24 respectively are episodes following a causal break. It is probably better, then, to treat 12:20 and 20:24 as parenthetical or background material. As usual they are introduced by de and closed byoun.
15. Global distinctiveness of John’s use of conjunction
I have now completed my basic analysis and classification of John’s use of de, oun, kai, and asyndeton. How does the Gospel of John differ from the non-Johannine writings of the New Testament? It differs, for one thing, in the statistical frequency of occurrence of de, oun, and asyndeton. De has unusually low frequency, while oun and asyndeton have unusually high frequency. But mere difference of frequency by itself need not be so important. To a certain extent, variations in frequency of use between two different texts may be accounted for in terms of variations in content and in choices about sequential and logical organization of that content. Even within the Gospel of John alone we have seen that there is a marked difference between the use of conjunctions in expository discourse and narrative discourse. Even within a single discourse type, say expository discourse, we might expect conjunctions like gar, oun, dio, ara, and dia touto to occur more frequently in a text organized in terms of a tightly knit argument; conversely, de, kai, alla, and asyndeton would occur more frequently in a discourse with looser logical connections.
But such factors are not enough to account for all the distinctiveness in John’s use of conjunctions. Expository discourse in John does indeed give an over-all impression of having a somewhat “loose” logical organization. Topical or thematic unity is more important than tightly knit argument. But this does not wholly account for the very high frequencv of asyndeton in expository
discourse in John. One must say in addition that the range of circumstances in which asyndeton would ordinarily be used has been expanded beyond the range present in other New Testament writers. Correspondingly, the use of de has been greatly restricted. But these altered ranges of use still make some contact with the range of use familiar to the average user of Greek. Hence the reader encountering John for the first time is not totally at sea. He just gradually makes unconscious adjustments to the new ranges of use.
Something similar can be said with respect to narrative discourse in John. Here the range of use of oun has been vastly expanded, so that it absorbs nearly all the territory normally occupied by de. A little of the remaining territory is absorbed by asyndeton, and the boundaries for when kai is used are also somewhat adjusted.
Such are the patterns for using de, oun, kai, and asyndeton in John. Because these patterns are distinctive, they can now be used as a test of common authorship. This will be the subject of a sequel article.
16. Problem cases in the Gospel of John
For the sake of comparison, we should make some note of how many problematic or exceptional occurrences of intersentence conjunctions there are in the Gospel of John. And we should try to see what kind of exceptions there are. Further research could focus on such problem cases.
We should not be surprised that there are some exceptions to the rules above. The rules have deliberately been formulated in the most specific, “mechanical” form possible, in order that they may be testable. But the actual practice in use of intersentence conjunctions is likely to be less rule-bound than grammar within the sentence. It is likely to be affected in subtle ways by a host of semantic factors. Hence no set of “mechanical” rules can do the job perfectly.
The analysis of difficulties is summarized in diagram 2. The first column of the diagram lists small sections of the Gospel of John. The second column gives the total number of intersentence conjunctions in the section. This number is in many cases only approximate, because of uncertainties at points over whether kai connects clauses or sentences. The third column gives the number of exceptions and, in parentheses, the percentage of exceptions out of the
total number of cases. The fourth column does the same for the cases that were judged difficult but probably not genuine exceptions.
Note also that several difficulties or exceptions (2:1, 4:1, 18:28, 19:38) occur at the beginning of pericopes. We already noted that the rules at the beginning of pericopes are different (§ 14). We should also say that such rules are less well confirmed, because there is only a small amount of evidence to use in trying to construct the rules in the first place. If we had a larger quantity of data as a base, we might be able to produce rules expressing more exactly the actual patterns. Hence exceptions to these rules of § 14 should be viewed with more tolerance than exceptions to the main set of rules (§§ 4-13).
In a sequel article I will analyze the passages 1:1-18, 7:53-8:11, and 21:1-25 and discuss the implications for unity of authorship.
Statistics on the Number of Difficulties and Exceptions to the Rules
of test in-
|exceptions (%)||difficulties (%)|
|1:19-28||18||1 (6%) (kai at 1:24)||1 (6%)|
|1:43-51||18||1 (6%) (kai at 1:46(1))||0|
|2:1-11||18||1 (6%) (kai att 2:1(1))||0|
|4:1-26||42||1 (2%) (oun at 4:1)||0|
Statistics on the Number of Difficulties and Exceptions to the Rules
|7:53-8:11||20||3 (15%) (de at 8:5, 10, 11(2))||2 (10%)|
|9:35-41||12||1 (8%) (asyndeton at 9:35)||0|
Statistics on the Number of Difficulties and Exceptions to the Rules
|18:28-32||9||1 (11%) (oun at 18:28)||0|
|19:8-16a||22||1 (5%) (kai at 19:14)||0|
|19:28-30||5||1 (20%) (asyndeton at 19:29)||0|
|19:38-42||9||0||1 (11%) (de at 19:38)|
|21:1-14||33||0||1 (3%) (de at 21:1)|
Statistics for larger groups
|1:19-51||64||2 (3%)||5 (8%)|
|4:1-54||91||1 (1%)||2 (2%)|
|7:53-8:11||20||3 (15%)||2 (10%)|
|9:1-41||87||1 (1%)||1 (1%)|
|18:1-40||83||1 (1%)||2 (2%)|
Statistics on the Number of Difficulties and Exceptions to the Rules
|19:1-42||85||2 (2%)||1 (1%)|
|1572||8 (0.5%)||32 (2%)|
2 Robert E. Longacre, Grammar Discovery Procedures (The Hague/Paris: Mouton, 1968); idem. Discourse, Paragraph, and Sentence Structure in Selected Philippine Languages, 3 vols. (Santa Ana: Summer Institute of Linguistics, 1968); idem, Hierarchy and Universality of Discourse Constituents in New Guinea Languages, 2 vols. (Washington, D.C.: Georgetown University, 1972); idem, An Anatomy of Speech Notions (Lisse: de Ridder, 1976); idem, “Discourse,” in Tagmemics, ed. Ruth M. Brend and Kenneth L. Pike (The Hague/Paris: Mouton, 1976) I 1-44; idem, Discourse Grammar: Studies in Indigenous Languages of Columbia, Panama, and Ecuador, 3 parts (Dallas: Summer Institute of Linguistics, 1976-77); idem, “The Discourse Structure of the Flood Narrative,” in Society of Biblical Literature 1976 Seminar Papers, 112th annual meeting (Missoula: Scholars, 1977) 235-62.
4 De and oun are, of course, postpositive; kai is not. But this difference in word order does not affect the fact that their occurrences are to be defined in opposition to one another and to asyndeton.
6 Longacre, Grammar Discovery Procedures 125-53; idem, Anatomy 274-76; Kenneth L. Pike and Evelyn G. Pike, Grammatical Analysis (Dallas: Summer Institute of Linguistics, 1977) 22, 255-62. For an attempt to define the sentence in Greek, cf. the “colon” of Johannes P. Louw, “Discourse Analysis and the Greek New Testament,” Bible Translator 24 (1973) 104; idem, Semantiek van Nuwe Testamentiese Grieks (Pretoria: Universiteit van Pretoria, 1976) 95; idem, A Semantic Discourse Analysis of Romans (Pretoria: University of Pretoria, 1979) II 1-31. But I am introducing some variation on the above approaches, because I have been influenced by Eunice V. Pike’s theory of hierarchy as developed in the area of phonology (“Phonology,” in Tagmemics, ed. Brend and Pike I 45-83, especially 45-47).
11 Even after all this has been said, there are still some remaining cases of ambiguity. But such problems cannot be explored here. They can be properly handled only in connection with a more technical discussion of the theory of grammatical hierarchy and the nature of “sentence” within this hierarchy.
13 The meaning-relationships signified by de, kai, and gar, etc., are closely related to the prepositional relations, which Longacre now classifies as part of “deep grammar” (Anatomy 98-164). I prefer to call them semantic or semological. The dispute is partly a matter of differing terminology. But I think that labeling proportional relations as deep grammar rather than semantics breeds confusion. My general theoretical defense is given in Vern S. Poythress, “Thirteen-Box Tagmemic Theory as a Method for Displaying Semi-independent Language Variables,” Studies in Language 2:1 (1978) 71-85; idem, “A Framework for Discourse Analysis ” Semiotica, 38-3/4 (1982) 293-95. John Beekman and John Callow take a similar approach in Translating the Word of God (Grand Rapids: Zondervan, 1974) 267-31; and in “The Semantic Structure of Written Communication,” Summer Institute of Linguistics (Dallas: 1981).
18 Cf. Vern S. Poythress, “Prepositional Relations,” in The New Testament Student and His Field (Vol. 5), ed. John H. Skilton and Curtiss A. Ladley (Phillipsburg, NJ: Presbyterian and Reformed, 1982) 159-212; Beekman-Callow, Translating 301, 306-307; William F. Arndt and R. Wilbur Gingrich, A Greek-English Lexicon of the New Testament and Other Early Christian Literature (2nd ed.; Chicago: University of Chicago, 1979) 151a on gar, part la (hereafter cited as Bauer).
20 Bauer, alla, la; see also Smyth, Grammar 2776; F. Blass and A. Debrunner, A Greek Grammar of the New Testament and Other Early Christian Literature (Chicago: University of Chicago, 1961) § 448 (1).
24 Bauer, alla, 5 (“rhetorically ascensive”); Blass-Debrunner, Grammar 448(6); Nigel Turner, A Grammar of New Testament Greek. Vol. III. Syntax (Edinburgh: T&T. Clark, 1963) 330 (“a strong addition”).
32 Poythress, “Prepositional Relations”; Beekman-Callow, Translating 297, on “equivalence”; Longacre, Anatomy 133-34. The preference for asyndeton with anaphoric houtos can, however, be overruled by the desire to use a kai in the “additive” sense (§ 8(2); cf. Bauer, kai, 3, “explicative” use). In the case of metalinguistic statements with anaphoric houtos, kai may or may not occur, depending on the closeness of the link (cf. 15:11, 16:1, but kai at 17:13).
33 Frank Stagg, “The Abused Aorist,” Journal of Biblical Literature 91 no. 2 (1972) 222-31; Stephen M. Reynolds, “The Zero Tense in Greek,” Westminster Theological Journal 32 (1969-70) 68-72; K. L. McKay, “Syntax in Exegesis,” Tyndale Bulletin 23 (1972) 46.
36 Abbott (Johannine Grammar § 2070) discusses this particular use of de, but does not define it as clearly and accurately as Nathan Walz, “Discourse Patterns in John II,” Notes on Translation no. 59 (3-1976) 2-8.
38 “After parenthetical remarks oun indicates a return to the main theme (resumptive)” (Blass-Debrunner, Grammar § 451(1)). Cf. Denniston, Particles 428; Bauer, oun, 2a; Joseph H. Thayer, Greek-English Lexicon of the New Testament (Edinburgh: Clark, 19014), oun, c.
39 Outside the New Testament, something similar already appears in classical Greek: Smyth notes that oun “resumes an interrupted narration” (Grammar, § 2964). Some of the cases that Bauer lists under oun 2a are of this type. Cf. also Johann A. Hartung, Lehre von den Partikeln der griechischen Sprache (Erlangen: Palm & Enke, 1832-33), II 22.
44 This account of oun should be compared with that in Abbott, Johannine Grammar §§ 2191-2200. Abbott’s position illustrates how far astray even a careful worker can go under the influence of an inadequate method. Concerning oun, Abbott summarizes, “Setting aside instances where oun introduces words of the Lord, we find that it either introduces an act [of Christ] of special solemnity, or else—as is most frequently the case—it is applied to His various journeys” (§ 2198). This is completely fanciful. It is of no explanatory value at all in trying to elucidate why oun is used at one point and not used at another. Nearly every act of Christ in the whole Gospel of John could be seen as either part of a journey or as an act “of special solemnity.” On a theological level, the Gospel, by mentioning only some acts of Christ (20:30), selects what it regards as important, and invests it with “solemnity.” Acts of Christ are bound to be significant. They will be so in cases where oun introduces them and in cases where it does not introduce them. It is therefore easy to read in a special solemnity at just those points where oun occurs. After one has done that, it superficially seems that one has an explanation for oun. But this is a circular explanation only, an explanation after the fact. It cannot predict beforehand where oun occurs, because “solemnity” is such an untestable feature. In fact, then, the feature of “solemnity” has nothing to do with oun as such. It is a feature of the Gospel as a whole, in its delineation of the acts of Christ. Abbott has, without knowing it, transferred to oun this general literary property of the Gospel of John.