Testing for Johannine Authorship by Examining the Use of Conjunctions

Vern S. Poythress

[Published in Westminster Theological Journal 46/2 (fall 1984) 350-369. Used with permission.]


This article is concerned with new evidence for the unity of the Gospel of John. In a previous article I delineated rough rules for the use of the principal conjunctions deounkai, and asyndeton in the Gospel of John.1 Now we can use those rules as a test for common authorship. There are some obvious limitations to a test of this type. The test can only be used effectively when we are examining a piece of text large enough to include a significant number of intersentence conjunctions. For example, a hypothesis that a single sentence is a gloss cannot be confirmed or disconfirmed with much confidence. Moreover, the rules are not absolutely rigid. Within the body of the Gospel of John, there are a number of exceptions and doubtful cases (for a fuller survey of the problem cases, see the previous article).2 To pronounce against common authorship with any confidence, we would need to find within a given text a significant frequency of apparent violations of the rules (say, above 10% in a shorter passage, or 5% in a longer passage). Finally, we must bear in mind that occurrences of dekaioun, and asyndeton are subject unusually frequently to corruption in the course of textual transmission. We should treat a given occurrence as a clear violation of a rule only when the external text-critical evidence is strong.

With these qualifications and reservations, I nevertheless believe that we have here a fairly strong test. To show this, let us apply the test in some known cases of different authorship, in order to see how well it works there. We will show that the test easily confirms that Matthew, Mark, and Romans are not by the author of the Gospel of John.


p. 351

1. The Authorship of Matthew

Let us use the first part of Matthew 2 as a test case. Does this text conform to the same patterns for deounkai, and asyndeton as does the Gospel of John? I will analyze in order each occurrence of the intersentence conjunctions deounkai, and asyndeton, beginning in Matt 2:1. Section references will be to the rules delineated in my previous article.

de at 2:1 introduces a new major episode with new participants and a new location. de conforms to the Johannine pattern (§14).

kai at 2:2 introduces a sentence without shift in agent. Again this is in conformity with John (cf. §12(l)).

de at 2:3 should be contrastive,according to Johannine patterns (§10). It does not look that way, but it conceivably could be. It is suspicious, but not necessarily a disconfirmation.

kai at 2:4 is in conformity with Johannine pattern. There is no shift of agent (cf. §12(l)).

de in the expression hoi de in 2:5 is in conformity. hoi de is a permitted special construction with de (§10(1)).

de at 2:8 should be contrastive, but is not. This is not in conformity.

de at 2:9 is ambiguous.3 If hoi de is the subject, de is part of the construction type ho de, in conformity with the pattern (cf. §10(1)). But if hoi de akousantes as a whole is the subject, de is out of conformity with pattern.

kai at 2:9 goes with a shift in agent (from the wise men to the star). Normally we would expect oun in the Johannine pattern (§1l). But kai can occur in rapid narrative sequences (cf. §12(2)). Hence there is some question about the occurrence of kai here, but no definite violation of pattern.

de at 2:10 should be contrastive but is not.

kai at 2:11 goes together with no shift in agent. It is in conformity with pattern (cf. §12(l)).

kai at 2:11(2) and 2:11(3) are similarly in conformity.

kai at 2:12 is in conformity.

de at 2:13 is probably contrastive, and in conformity.

The kai’s in 2:13 are embedded in a command (hortatory discourse). Hortatory discourse has not appeared in purity in the


p. 352

Gospel of John, so it is hard to say how far the rules for expository discourse would hold.

de at 2:14 is apparently not contrastive, and so in violation of the pattern.

We have gone far enough already to say with some confidence that this material is not by the same author as the Gospel of John. The frequency of exceptions is 3/18 = 17%. Going further on would uncover still further disconformities to Johannine patterns, and would strengthen the conclusion. The conclusion follows even apart from many other types of observation that we might make about non-Johannine features of the narrative: fairly frequent use of tote (2:7, 16 ), heavy use of adverbial participles, the construction with par autōn (2:4, 7 ), and so on.

2. The Authorship of Mark

Very briefly, let us perform a similar test with the first part of Mark 2. Mark 2 is to be compared with the patterns for narrative discourse in John.

kai at 2:1 introduces a new episode. But I judge that there is no causal gap between the episodes. Hence the same rules should apply as those within episodes. There is a shift of agent here, from the crowds (ērchonto in 1:45 ) to Jesus (2:1 ). Since, moreover, 2:1 is not the middle of a rapid narrative sequence, we expect oun to begin 2:1 . The use of kai is an apparent disconformity.

kai at 2:2(2) is in rapid narrative sequence. It is in conformity (cf. §12(2)).

kai at 2:3 is not in conformity. It goes together with a shift in agent and the subsequent use of historical present. We should have asyndeton (cf. §13(l)).

kai at 2:5 goes together with a shift of agent. We would normally have oun. However, in the exceptional case, kai may be used as part of a “rapid narrative sequence.” This looks questionable here. Jesus is responding to the whole set of events in 2:4 . Because of the complexity of those events, we would expect continuation with oun.4

The cumulative force of these observations is already enough


p. 353

to say that 2:1–5 is probably not by the same author as John. Again, further confirmation comes from examining more text.

3. The Authorship of Romans

As a final “easy” case, let us take the first part of Romans 2. This is a piece of expository discourse. We will compare it with the patterns for expository discourse in John.

Dio at 2:1 is not in conformity. We expect instead oun or dia touto.

de at 2:2 should be contrastive, but does not look it. But there is textual variation, so it is best to ignore this instance.

de at 2:3 should be contrastive, but is not. It is out of conformity.

de at 2:5 should be contrastive, but is not. Again it is out of conformity.

Already it is apparent that Romans 2 is not Johannine in its use of intersentence conjunctions.

The use of this test on Matthew, Mark, and Romans is enough to show that in an ordinary instance 10 verses is enough to supply a rough test of Johannine authorship. Even as few as 5 verses is sometimes enough; 20–30 verses would be enough to give us comfortable confidence.

4. The Redaction of the Bulk of the Gospel of John

Now let us try to apply the test to the question of the literary and redactional history behind the extant text of the Gospel of John itself. Can we detect portions of the Gospel that deviate from the “bulk” pattern of the Gospel as a whole? Table I lists deviations from the rules tabulated by pericopes.5 (In later sections we will confirm these results in detail for the key passages 1:1–18 ; 7:53–8:11 ; 21:1–23 ; 21:24–25 .) The table shows that, in the whole Gospel, there is only one “deviant” portion, namely 7:53–8:11 ; the famous pericope on the adulterous woman). There are also a disturbing number of difficulties, though no definite violations of the rules, in 1:35–42 . All the rest of the Gospel conforms


p. 354

impressively well to the bulk pattern. How can we account for this? What are the implications for unity of authorship and for the redactional history?

Consider: the “bulk” patterns we are talking about are patterns for the use of intersentence conjunctions in John, patterns delineated at length in my previous article. They are characteristic of John alone, and not of the rest of the New Testament. We have established this briefly by comparison with Matthew, Mark, and Romans (§§2–4). In addition, my own brief checks of portions of the other non-Johannine literature of the New Testament and the Apostolic Fathers confirm this. Hence it is reasonable to postulate that this pattern has its source in a single human being standing at some stage in the production of the Gospel of John. Let us call this individual the Pattern-producer. There may have been others involved besides this Pattern-producer. But those before him in the process of production had their product revised by him so as to conform to his pattern. Those after him did not upset the pattern already produced (except for 7:53–8:11 ).

Now let us ask what, in all probability, could and could not have happened after the stage in composition at which the Pattern-produced worked. The Pattern-producer produced a written text. Only a written text, not an oral one, could be easily transmitted in such a way as to preserve the pattern. Now, suppose that a given piece of text from the Gospel of John, a piece of substantial length (10–20 verses or more), conforms to the pattern. Any such piece of text must have been present, at least in bulk, in the written document produced by the Pattern-producer. It would have been difficult if not impossible for a later editor to introduce whole blocks of new material, because they would not conform to the pattern. Of course, a later editor could have deleted blocks of material without substantially changing the pattern. But in the nature of the case such deletion is far less likely than addition of new material.

What about the possibility that we are dealing with a Johannine school? Some have argued that a Johannine school or milieu could have produced many of the similarities.6 This is true, perhaps,


p. 355

for the theological themes. It is less likely to be true of stylistic features. It is least likely to be true, it seems to me, for a factor as subtle as the use of intersentence conjunctions. Only a deliberate imitator could hope to conform to the pattern. And even a deliberate imitator would concentrate on harmony of content and of over-all style, not on the details of sentence-level connectives.

Hence the written text of the Pattern-producer is substantially the Gospel of John as we have it. The Pattern-producer’s text even included the disputed sections 1:1–18 and 21:1–23 , since these too conform to the pattern. The Pattern-producer’s text did not include 7:53–8:11 , since this does not conform to the pattern.

Editors coming after the Pattern-producer may have introduced glosses of short length (say 1–3 verses). 21:24–25 may be such a gloss. The test pattern is of such a character that short glosses cannot be detected. However, no large blocks made up of “whole cloth” were inserted after the stage of the Pattern-producer. Some editors might conceivably have reworded the text received from the Pattern-producer, perhaps in the way that many scholars think Luke dealt with the Gospel of Mark. But reworking to any considerable extent tends to disrupt the grammatical form of the original. Matthew and Luke, for example, do not use kai with nearly the same frequency as does Mark. Hence, at most a very small amount of reworking separated the Pattern-producer’s text from the “autographic” text, the initial text available to the “public” from which copies started multiplying. The absence of the test pattern from 7:53–8:11 , taken together with the external manuscript evidence, shows that 7:53–8:11 was added after the autographic stage.

5. The Composition of the Bulk of the Gospel of John

Consider now what is most likely to have happened before the stage at which the Pattern-producer wrote. The Pattern-producer may have used one or more sources, oral or written. But whatever sources he used, he digested them; he conformed them to his own style. No substantial block of material from his sources did he simply take over verbatim. Moreover, it is reasonable to assume that in digesting his material and altering it, he did notsimply alter it in its conjunctions or in other features of surface grammar


p. 356

(for an examination of such distinctive grammatical features, see Abbott, Schweizer, and Ruckstuhl).7 The use of intersentence conjunctions is connected in subtle ways with the propositional relations between sentences of discourse, and these interlock with an author’s preferences for the over-all organization of arguments and stories. Hence the Pattern-producer is likely to be also the person who is most responsible for the remarkable unifying theological and thematic features so familiar and characteristic in the Gospel of John.

But this in turn means that the Pattern-producer digested his sources thoroughly. He digested them grammatically, rhetorically, thematically, theologically. If so, he destroyed the most reliable means by which we might hope to make any kind of reasonable guesses about a diversity of sources. To try to separate sources behind the Pattern-producer remains quite speculative.8


p. 357

Of course, it might still be possible to detect statistical variations in the frequency of distinctively Johannine characteristics as Fortna and Nicol attempt to do.9 Such variations might be due to the influence of a source. But they might also be due to differences in form and subject-matter. The differences between different source hypotheses do not inspire confidence that there is reasonable noncircularity left.10

But there is still one possible distinction of sources that might appear to be justified on the basis of the pattern itself. The distinction between expository and narrative discourse is a necessary feature in the definition of the pattern. This distinction is not so far away from the idea of a distinct source for the “discourses” and for the “signs.”11 Does this confirm the existence of a “signs” source? It does not, for several reasons.

(1) Expository discourse and narrative discourse are bound to show some distinctions in their use of connectives in any author.12 The dominant mode of connection between sentence contents in expository discourse is logical, topical, even sometimes argumentative. In narrative the dominant mode of connection is temporal and causal. Hence the intersentence conjunctions are bound to show a somewhat different distribution of use.

(2) In John all narrative shows the same features, not only the narratives of the “signs,” but every pericope in the Passion narrative. The totality of narrative discourse is not what most people have in mind by a “signs” source.13


p. 358

(3) In John all expository discourse shows the same features. Not only the “Farewell Discourse” of John 14–17 , but the Evangelist’s expository comments, the replies of the Jews, and the speeches of John the Baptist conform to the pattern. Admittedly there is only a comparatively small sample of expository material apart from Jesus’ speeches. But it is enough to give this theory difficulty. Our impressions will, moreover, be confirmed when we see that 1–2-3 John show the same pattern as the expository discourse of the Gospel.

(4) In John narrative and expository discourse are closely intertwined. The expository discourses frequently occur as short or long speeches in the over-all framework of dialogue, which is a species of narrative. And narratives gain deeper significance from accompanying expository material. (For example, the healing of the blind man in John 9 gains significance from Jesus’ teaching on light, mentioned briefly in 9:5 and developed more fully in John 8.)

(5) The patterns with respect to conjunctions in expository discourse and narrative show at least two common features. (a) In both expository discourse and narrative, de exhibits a restricted range of use in comparison to its use outside of John. Its main use is the contrastive one. (b) In both expository discourse and narrative, asyndeton seems to function as the “default” option. It is used for the most part simply when other conjunctions are not appropriate. This is, perhaps, not so surprising, because at the most naive level asyndeton is merely an absence. But in John it means that asyndeton has unusually high frequency.

Here I conclude that it is better not to postulate two distinct sources on the basis of the distinction in discourse features between expository and narrative discourse in John.

6. The Authorship of the Prologue of John

Now let us examine in detail the Prologue of the Gospel of John, John 1:1–18, in order to check whether its use of conjunctions is in conformity or disconformity with the bulk of the Gospel of John.

Though the Prologue has some embedded “mini” narratives (1:11, 14 ), it is basically expository discourse. At root, the content of the Prologue is organized topically and logically, not by sheer


p. 359

temporal sequence. Hence, the patterns for Johannine expository discourse (§§4–9) are to be applied. (Section numbers refer to the earlier article on intersentence conjunctions.)14

The kai’s at 1:1(1) and 1:1(2) coordinate closely related statements. Their use is in conformity with the Johannine pattern (§8(1)).

Asyndeton at 1:2 is to be expected preceding the anaphoric pronoun houtos (cf.§8(1)). It is in conformity.

Asyndeton at 1:3 is the unmarked conjunction, connecting vaguely related statements. It is in conformity (cf. §9).

kai at 1:3 coordinates closely related statements, in conformity with the pattern (§8(l)).

Whichever way one punctuates the end of 1:3 , the asyndeton is in conformity (unmarked conjunction, §9).

The kai’s of 1:4 , 1:5(1) , and 1:5(2) link closely coordinated statements with common words. (In conformity, §8(l)).

Asyndeton at 1:6 is regular unmarked use (§9).

Asyndeton at 1:7 is to be expected before houtos (§8(1)).

Asyndetons at 1:8 , 1:9 , and 1:10 are unmarked uses (§9).

The kai’s at 1:10(1) and 10(2) connect closely coordinated statements (§8(1)).

Asyndeton at 1:11 is an unmarked use (§9).

kai at 1:11 connects events in a short quasinarrative sequence (§8(3)).

de at 1:12 is contrastive: it contrasts “his own” with “those who received him.” (In conformity, §5.)

kai at 1:14(1) appears at first glance not to be in conformity with the pattern. We would expect asyndeton at the beginning of a new paragraph loosely connected with what precedes. However, the kai here is probably part of a both-and construction, or possibly has the sense “also” (§2). It would be unwise, therefore, to say that it is not in conformity with the pattern.

The kai’s of 1:14(2) and 1:14(3) connect events in a quasinarrative, and are in conformity (§8(3)).

Asyndeton at 1:15 is an unmarked use, in conformity (§9).

kai at 1:15 probably connects clauses rather than sentences. Legōn probably modifies marturei as well as kekragen, indicating that we are dealing with a single sentence here (cf. §1).


p. 360

The hoti’s of 1:16 and 1:17 are actually typically Johannine in character, according to Abbott’s investigations.15 But my own work says nothing about hoti.

Asyndetons at 1:18(1) and 1:18(2) are unmarked uses (§9). We might have expected alla at 1:18(2) . But the line between types of contrast and tension requiring allade, and asyndeton in John is somewhat fluid. The two parts of 1:18 are in tension; but, viewed from a certain standpoint, they are not in virtual contradiction. Hence asyndeton might well be preferred to alla. Anyway, the distinctiveness of John lies in the higher, not lower, frequency of use of asyndeton. Hence this occurrence of asyndeton is not at all against Johannine authorship.

All in all, there is not a single disconformity to the Johannine pattern in the whole eighteen verses. This is quite impressive, when compared with the sample results on Matthew, Mark, and Romans. It argues strongly that the substance of 1:11–18 derives from the Pattern-producer. In other words, it has the same author as the bulk of the Gospel.

7. The Authorship of John 5:4

Next, let us try out the same test on the text-critically problematic passage John 5:4.

The gar of 5:4 introduces a reason why the multitude of invalids was gathered at the pool. It is in conformity.

oun at 5:4 introduces a sentence continuing a short narration of what customarily happened at the pool. Possibly a narration of customary (repeated) events might exhibit different patterns than the narration of once-for-all events. But in this case a shift of agent (from the angel to the invalids) goes together with oun, in agreement with the Johannine pattern.

Hence there are no disconformities to the Johannine pattern in 5:4 . However, the text is too short to provide opportunity for a significant test. Whoever the author was, we might have expected the gar in 5:4 ; this is a pattern common to all of Greek, not merely to John. Only the use of oun at 5:4 is distinctive to John. But a single instance like this is too little to go on. The test is inconclusive with respect to the authorship of John 5:4.


p. 361

8. The Authorship of John 7:53-8:11

The text John 7:53–8:11 is, however, long enough to provide some scope for the detection of conformity or disconformity to the Johannine pattern. Let us examine the use of intersentence conjunctions in 7:53–8:11 , one by one.

kai at 7:53 is a continuation of the narrative with no shift in agent. The use of kai is in conformity to the Johannine pattern (§12(1)).

de at 8:1 is contrastive, in conformity with the pattern.

de at 8:2 , according to the rules of §10, can be accounted for if it is contrastive. It does not look strongly contrastive. But the author might have wished to draw attention to reverse movements of Jesus, and might have inserted a contrastive marker for this purpose. More likely, 8:2 is to be regarded as the beginning of a new episode. In that case, de is used noncontrastively according to §14(3).

The kai’s at 8:2(1) and 2(2) must be interpreted as connecting sentences in a “close narrative continuation,” a rapid sequence of events. Under such conditions, kai can occur together with a shift in agent (cf. §12(2)).

de at 8:3 , in order to be in conformity, must be interpreted as contrastive. This is suspicious, but might possibly be. The author might be saying, “But the scribes and the Pharisees had different plans, contrasting with Jesus’ teaching plans.”

kai at 8:3 is a narrative continuation with no shift in agent. It is in conformity with the pattern (§12(1)).

de at 8:5 is not contrastive, hence not in conformity (§10).

oun at 8:5 comes in the middle of an embedded expository discourse uttered by the scribes and Pharisees. Evidently oun here has its ordinary inferential sense, and so should be reckoned as in conformity.

de at 8:6 is part of the construction touto de explaining the significance of the event of 8:4–5 . It is in conformity (cf. §10(3a)).

de at 8:6(2) ought to be contrastive, but does not look it. It is suspicious. We would normally expect oun here.

de at 8:7 is the regular use of de with a hōs clause (cf. §10(4).

The kai’s of 8:7 and 8:8 are a continuation with no shift in agent, in conformity with the pattern (§12(1)).


p. 362

de at 8:9 is probably16 to be interpreted as part of the construction hoi de, and so in conformity with the pattern (cf. §10(1)).

kai at 8:9 is a close narrative continuation (§12(2)), perhaps without any significant shift of agent (since the “real” agent of kataleipō is the Jews).

de at 8:10 is clearly not contrastive, out of conformity to the pattern.

Asyndeton at 8:10 is unmarked continuation in expository discourse (§9).

De at 8:11 is part of the pattern ho de (in this case hē de).

de at 8:11(2) is not contrastive and not in conformity.

Asyndeton at 8:11 is unmarked continuation in expository discourse, in conformity.

On the basis of a significant number of disconformities to the Johannine pattern (de at 8:5 , 8:10 , and 8:11(2) , plus some “suspicious” cases), we can confidently conclude that 7:53–8:11 does not derive from the Pattern-producer. When we take into account the external text-critical evidence, we can conclude that in fact 7:53–8:11 was not part of the autographic text.

9. The Authorship of John 21:1-23

Now let us apply the test to the last chapter of John, omitting the final comments in 21:24–25 . John 21:1–23 is too lengthy a text for it to be worthwhile to discuss individually each occurrence of an intersentence conjunction. I will confine my explicit remarks to the instances which are somehow noteworthy or possible disconformities to the Johannine pattern.

Meta tauta at 21:1 introduces a new episode in a manner in conformity with the pattern (cf. §14(2)).

de at 21:1 is problematic. It seems not to be in conformity, but the added sentence is quite unusual in content. Is it quasiparenthetical? Is it a second introduction to the episode, and hence in conformity with the unmarked use of de in introducing new episodes? Or, more likely, is de functioning with kataphoric houtōs similar to the pattern of §6(3)? It is probably best to say that


p. 363

the de here is a bit suspicious, but not a definite case of disconformity.

In 21:2 , following the houtōs of 21:1 , asyndeton is to be expected. This use of asyndeton was not listed separately in my discussion of asyndeton in John, because the pattern is not common enough in John to merit separate mention. But it is a common feature in Greek in general.17

Asyndetons at 21:3(1) , 3(2) , and many other places in the chapter conform to the rule that asyndeton is normally used preceding a verb in the historical present.

Asyndeton at 21:3(3) is not in conformity. However, there are two textual variants with considerable support. The textual variant with kai is in conformity. kai here would continue the narrative with no shift in agent. The textual variant oun is not what we would expect. But it might conceivably be interpreted as having here the full sense “therefore.” Because of the text-critical uncertainty, this cannot be said to be a disconformity.

de at 21:4 should be interpreted as marking the beginning of a new narrative episode in a different time frame. It need not be contrastive. But there also is a possibility of contrast between prōias and the preceding nukti. Hence, either way, this de is in conformity with the pattern.

oun at 21:5 is problematic. Normally we expect asyndeton with the historical present legei. But this oun picks up after a deviation from the main line of narrative events in 21:4b (“the disciples did not know that it was Jesus”). The use of oun is optional when the historical present follows the close of a parenthesis (cf. §13(1)).

oun at 21:7 is the full inferential sense of oun.

oun at 21:9 is a use of oun after parenthetical material (§11(2)).

kai at 21:11 should be seen as a “close narrative continuation,” in conformity with the pattern (§12(2)).

de at 21:12 introduces parenthetical information in the imperfect.


p. 364

Asyndeton at 21:14 is preferred with anaphoric touto (cf. §13(4)).

Asyndeton at 21:17(2) appears to be in disconformity. We expect oun with the change of agent. However, this may be one of the cases where asyndeton is used to mark a disturbance in the events along the main line of the narrative (cf. §13(5)).

kai at 21:19 appears where, preceding touto, we might expect asyndeton. Possibly the preceding touto de eipōn at the beginning of the verse, in parallel with the second touto eipōn, has led to the preference for kai. More likely, this is just a case of the use of kai with close narrative continuation (cf. kai touto eipōn in 11:28 ; 18:38 ; 20:20 ; 20:22 ; §13(4)).

Asyndeton at 21:20 follows a command. It is in conformity with the pattern (cf. §13(3)).

Instead of oun at 21:21 we expect asyndeton with touton. The textual variant does have asyndeton.

oun at 21:23 has the full inferential sense.

All in all, for as long a piece of text as this, the degree of conformity to the Johannine pattern is impressive. There is no definite case of disconformity. There is only some question about the de of 21:1 and the asyndeton of 21:17(2). The impressive degree of agreement supports the hypothesis that 21:1–23 derives from the Pattern-producer.

10. The Authorship of John 21:24-25

Finally, we can apply the test on conjunctions to John 21:24–25.

Asyndeton at 21:24 preceding anaphoric houtos is in conformity with the pattern (§13(4); or §9).

kai at 21:24(1) joins clauses, not sentences.

kai at 21:24(2) introduces a metalinguistic comment on the first part of vs. 24 . This is in conformity (§8(2)).

de at 21:25 introduces another “additive” type of comment on one aspect of 21:24 . We would normally expect a kai. But the presence of another kai with the meaning “also” has converted the intersentence conjunction to de(cf. §6(2)).

Thus 21:24–25 is in conformity with the Johannine pattern. However, it is too short a piece of text to provide a good test. The test is inconclusive.


p. 365

11. The Authorship of the Johannine Epistles

The Johannine epistles, 1–2-3 John , are instances of expository discourse. The hypothesis of common authorship with the Gospel of John can therefore be tested using the patterns found to hold with respect to the expository discourses in the Gospel. It would be too tedious to evaluate in this article each use of intersentence conjunctions in 1–2-3 John . I will confine myself to a summary of the results.

In all of 1–2-3 John , as far as I can see, there is only one clear exception to the Johannine pattern for use of conjunctions. We find de in 1 John 4:18 where we would expect kai. But it is possible that the final clause of 4:18 , introduced by de, is subordinated to the earlier hoti. In that case, de would be functioning as an interclausal rather than an intersentence conjunction. But even if the connection is interclausal, we would expect a kai. The dehere constitutes a definite difficulty. But one such difficulty out of a total of seven chapters of text still constitutes impressive agreement with the Gospel of John.

In addition, there are some problems with a few cases of the use of kai in 1–2-3 John . kai occurs in 1 John 1:5, 2:3 , 2:20 , 2:27 , 2:28 , 3:23 , 4:3(2) , 4:16 , 5:11 , 5:14 , and 2 John 5, all cases where we might expect asyndeton instead. But upon closer inspection all of these problem cases have plausible explanations. Some of them (1:5 , 2:3 , 3:23 , 4:3(2) , 5:11 , 5:14 ) are instances of kai with kataphoric demonstrative pronoun houtos, already noted as an exceptional case (§6(3)). The kai of 4:16(1) can easily be interpreted either as part of a both-and construction or as a further intensification of the emphatic hemeis (cf. also emphatic pronouns after kai in 1 John 2:20; 2:27 ). kai could also be seen as appropriate for introducing the “additive” statement of 4:16 .

The other instances from among this list of kai’s should probably be handled by means of a separate rule. The rule is this: when a “nontemporal” nun (not meaning literally “at the present moment”) introduces a new sentence, it is preceded by kai. The construction kai nun is in fact a fixed usage recognized by the standard works.18 It occurs at least once in the Gospel of John,


p. 366

at John 14:29, though the kai there can also be interpreted by other means (as introducing a metalinguistic comment, cf. §8(2)). 1 John 2:28 and 2 John 5 are the instances in 1–2-3 John .

In conclusion, the test pattern for intersentence conjunctions supports the idea of unity of authorship of all of the Gospel of John and of 1–2-3 John .

Westminster Theological Seminary

p. 367

Table 1

Statistics on the Number of Difficulties and Exceptions to the Rules


passage total number of
test instances
exceptions(%) difficulties(%)
J1:1–13 19 0 0
1:14–18 9 0 1 (11%)
1:19–28 18 1 (6%) (kai at 1:24) 1 (6%)
1:29–34 10 0 0
1:35–40 13 0 3(23%)
1:41–42 5 0 1(20%)
1:43–51 18 1 (6%) (kai at 1:46(l)) 0
2:1–11 15 1 (6%) (kai at 2:1(1)) 0
2:12–22 18 0 0
2:23–25 3 0 0
3:1–21 33 0 0
3:22–36 32 0 0
4:1–26 42 1 (2%) (oun at 4:1) 0
4:27–30 9 0 1 (11%)
4:31–38 13 0 0
4:39–42 6 0 1 (17%)
4:43–45 4 0 0
4:46–54 17 0 0
5:1–9a 12 0 0
5:9b–16 15 0 1 (7%)
5: 17–29 19 0 0
5:30–47 31 0 1 (3%)
6:1–15 21 0 2(10%)
6:16–21 11 0 1 (9%)
6:22–40 24 0 0
6:41–51 17 0 0
6:52–58 10 0 0
6:59–71 21 0 0
7:1–9 16 0 2(13%)
7:10–13 8 0 0
7: 14–24 18 0 0
7:25–36 26 0 0
7:37–44 12 0 0
7:45–52 9 0 0
7:53–8:11 20 3 (15%) (de at 8:5, 10, 11(2)) 2(10%)
8:12–20 18 0 0
8:21–29 20 0 0
8:30–47 39 0 2 (5%)
8:48–59 28 0 0
9:1–7 14 0 0
9:8–12 13 0 0


Table 1 (Continued)


passage total number of test instances exceptions(%) difficulties(%)
9:13–17 11 0 0
9:18–23 13 0 0
9:24–34 24 0 1(4%)
9:35–41 12 1 (8%) (asyndeton at 9:35) 0
10:1–6 12 0 0
10:7–18 26 0 0
10:19–21 6 0 1(17%)
10:22–30 19 0 0
10:31–42 16 0 0
11:1–16 22 0 0
11:17–31 21 0 0
11:32–44 24 0 1(4%)
11:45–54 16 0 0
11:55–57 6 0 0
12:1–8 11 0 0
12:9–11 3 0 0
12:12–19 12 0 1(8%)
12:20–36 37 0 0
12:37–43 7 0 0
12:44–50 11 0 1(9%)
13:1–11 17 0 1(6%)
13:12–20 15 0 1(7%)
13:21–30 14 0 1(7%)
13:31–38 16 0 0
14:1–7 13 0 0
14:8–21 27 0 1(4%)
14:22–31 23 0 0
15:1–17 30 0 0
15:18–27 15 0 0
16:1–11 16 0 0
16:12–16 11 0 0
16:17–24 23 0 0
16:25–33 20 0 0
17:1–8 12 0 0
17:9–19 21 0 0
17:20–26 8 0 0
18:1–11 19 0 0
18:12–14 5 0 0
18:15–18 12 0 0
18:19–24 11 0 1(9%)
18:25–27 6 0 0
18:28–32 9 1 (11%) (oun at 18:28) 1(11%)
18:33–40 21 0 0
19:1–7 18 0 0


Table 1 (Continued)


passage total number of test instances exceptions(%) difficulties(%)
19:8–16a 22 1 (5%)(kai at 19:14) 0
19:16b–24 16 0 0
19:25–27 4 0 0
19:28–30 5 1 (20%)(asyndeton at 19:29) 0
19:31–37 11 0 0
19:31–37 11 0 0
19:38–42 9 0 1(11%)(de at 19:38)
20:1–10 21 0 0
20:11–18 19 0 0
20:19–23 11 0 0
20:24–29 15 0 0
20:30–31 2 0 0
21:1–14 33 0 1(3%)(de at 21:1)
21:15–23 21 0 0
21:24–25 3 0 0


Statistics for larger groups


1:1–18 28 0 1(4%)
1:19–51 64 2 (3%) 5(8%)
2:1–25 39 1 (3%) 0
3:1–36 65 0 0
4:1–54 91 1 (1%) 2(2%)
5:1–47 77 0 2(3%)
6:1–71 104 0 3(3%)
7:1–52 89 0 2(2%)
7:53–8:11 20 3 (15%) 2(10%)
8: 12–59 105 0 2(2%)
9:1–41 87 1 (1%) 1(1%)
10:1–42 79 0 1(1%)
11:1–57 89 0 1(1%)
12:1–50 81 0 2(2%)
13:1–38 62 0 3(5%)
14:1–31 63 0 1(1.6%)
15:1–27 45 0 0
16:1–33 70 0 0
17:1–26 41 0 0
18:1–40 83 1 (1%) 2(2%)
19:1–42 85 2 (2%) 1(1%)
20:1–31 68 0 0
21:1–25 57 0 1(2%)
1:1–21:25 without7:53–8:11 1572 8 (0.5%) 32(2%)



1 Vern S. Poythress, “The Use of the Intersentence Conjunctions deounkai, and Asyndeton in the Gospel of John,” Novum Testamentum, to appear.

2 Ibid., §16.

3 Cf. C. F. D. Moule, An Idiom Book of New Testament Greek (2d ed.; Cambridge: Cambridge University, 1971) 107.

4 The textual variant with de is also not in conformity with Johannine patterns.

5 This table is further explained in my previous article, “Use of Intersentence Conjunctions,” §16.

6 E.g., cf. Robert Kysar, The Fourth Evangelist and His Gospel (Minneapolis: Augsburg, 1975) 20–23; Eugen Ruckstuhl, Die literarische Einheit des Johannesevangeliums (Freiburg: Paulusverlag, 1951) 186; Raymond Brown, The Community of the Beloved Disciple (New York: Paulist, 1979).

7 Edwin A. Abbott, Johannine Grammar (London: Adam and Charles Black, 1906); Eduard Schweizer, Ego Eimi (Göttingen: Vandenhoeck & Ruprecht, 1939) 87–99; Ruckstuhl, Einheit, 193-205. Note that my test reinforces the conclusions of Schweizer and Ruckstuhl concerning stylistic unity of the Gospel of John. Schweizer and Ruckstuhl include John 21 but not 7:53–8:11 .

8 “We must leave the question [of sources] open, concluding that if the evangelist used written sources, their distinctive character is not discernible through the finishing work which he or a subsequent editor accomplished on his material” (Nigel Turner, A Grammar of New Testament Greek. Vol. IV. Style [Edinburgh: T. & T. Clark, 1976] 66–67).

Some note should be given to the work of A. Q. Morton and J. McLeman on the sources of John. In their work The Genesis of John (Edinburgh: Saint Andrew, 1981) 17–26, they complain about the subjectivity of conventional source criticism. They then attempt to introduce objective criteria by using stylometry and counts of the letter-length of codex columns. With regard to their letter counts, they have indeed found a pattern in the Gospel of John. But they have not yet given adequate attention to the question of whether it may be only a pseudopattern. The probability of finding at least one low-probability pattern in a sequence of random numbers can be high if there are (antecedently) sufficiently many patterns to look for.

Morton and McLeman also claim to separate two main sources J1 and J2 and show that there are stylometric differences between them. But a quick inspection shows that J1 contains proportionally more narrative discourse (e.g., all of 18:28–21:14 ), and J2 contains proportionally more expository discourse (e.g., all of 14:1–17:26 ). There are bound to be differences between these two in sentence length and in use of intersentence conjunctions. Moreover, the predominantly expository material of J2 can be expected to agree more closely with the almost wholly expository content of 1 John. Sure enough, this is what Morton and McLeman find (p. 106). Hence the impressive mass of statistical evidence they accumulate in pp. 95-113 can be accounted for without postulating sources. In fact, it very largely is accounted for simply by the difference in the rules that apply to narrative and expository discourse. (See further discussion in §5 of this article.)

9 Robert T. Fortna, The Gospel of Signs: A Reconstruction of the Narrative Source Underlying the Fourth Gospel (Cambridge: Cambridge University, 1970) 203–18; Willem Nicol, The Sēmeia in the Fourth Gospel: Tradition and Redaction (Leiden: Brill, 1972) 16–27.

10 Cf. Donald A. Carson, “Current Source Criticism of the Fourth Gospel: Some Methodological Questions,” JBL 97 (1978) 411–29.

11 Cf. Rudolf Bultmann, The Gospel of John (Philadelphia: Westminster, 1971); for a survey of source hypotheses after Bultmann, see Carson, “Current Source Criticism”; Robert Kysar, The Fourth Evangelist, 9-81.

12 Cf. Ruckstuhl, Einheit, 187; Schweizer, Ego Eimi, 106-107.

13 But cf. Fortna, The Gospel of Signs.

14 Poythress, “Intersentence Conjunctions.”

15 Abbott, Johannine Grammar §2180.

16 The ambiguity is noted by Moule, Idiom Book, 107.

17 Cf. J. D. Denniston, The Greek Particles (2d ed.; Oxford: Oxford University, 1934) xliii.

18 Cf. BAGD nun, 2; BDF §442(15); A. T. Robertson, A Grammar of the Greek New Testament in the Light of Historical Research (Nashville: Broadman, 1934) 1147 (“non-temporal use of νῦν”).

Sign up to receive new posts via e-mail