Add to My Library 
For the past several decades, exegetes have been engaged in a quiet battle over a locus classicus of dogmatic theology—yet, quite curiously, few theologians have plundered the battlefield for booty.1 I am referring to 2 Tim 3:16, the passage which on any reckoning must be regarded as of central importance to the self-witness of scripture.2 From time to time—though with increasing regularity in conservative Protestant circles—exegetical essays are showing up which address a variety of problems: What is the meaning of grafhv—mere “writing,”3 or “scripture”? And if scripture is meant, does it refer exclusively to the autographs or does it include the copies?4 Does pavsa mean “all” or “every”?5 What is the meaning of qeovpneusto"? 6And, finally, what is the relationship of qeovpneusto" to grafhv—i.e., is it attributive (thus, “all/every inspired scripture is also profitable. . .”) or predicate (“all/every scripture is inspired and profitable”)?
It is the conviction of many exegetes that this last question is the most critical theologically—as well as the most difficult to resolve exegetically. Without entering into the theological discussion per se, though hopefully contributing to it, it is my purpose merely to address the issue of the grammatical relation of qeovpneusto" to grafhv. Further, I wish to restrict my discussion to what I consider to be the primary syntactical evidence,7 for I believe that we can come down fairly decisively on one side of the fence by standing on this leg of evidence alone.8
In arguing for the primacy of syntax in resolving the issue, I am cognizant of going out on a limb not often taken in recent studies. For example, House suggests that since attributive and predicate nuances of qeovpneusto" are both “grammatically permissible,” “the decision ultimately must be made by determining how this word relates to its context.”9
Goodrick goes further in deprecating the feasibility of an objectively verifiable solution. He writes:
I have chosen [the translation] “All Scripture is God-breathed,” but I have little to defend what I have done. . . .
. . . I am reluctantly being dragged to the conclusion that an exegete of quality is one who has, by much exposure to the text and to the language in which it is written, developed a strong and reliable intuition.
I place the verb where I do because “it scans that way.”10
The problem with this approach is that the text does not ‘scan’ that way for several scholars who are equally proficient in their reading and exegetical skills.11 Ultimately, the intuitive approach which Goodrick is advocating is usually quite serviceable for one’s first impression of a passage, but when he faces a crux interpretum such as 2 Tim 3:16, there is a high antecedent probability that his theological presuppositions will override, or at least cloud, his exegetical intuition.12
Nevertheless, the fact that Goodrick makes little effort in the direction of a syntactical resolution is indicative of a majority trend: the grammar of 2 Tim 3:16 appears to be sufficiently unique, or perhaps sufficiently amorphous, to create something of a syntactical impasse.13
In 1961, J. W. Roberts went against the tide and argued14 for an attributive qeovpneusto" by basing his argument squarely on the linguistic structure of the text:
There are twenty-one instances in the New Testament in which pas is used to modify a noun which is immediately followed by another adjective as in 2 Tim. 3:16. In every case the Greek order of words is (1) pas, (2) the noun, and (3) the adjective. Typical examples are “every good tree” (Matt. 7:17); “every idle word” (Matt. 12:36); “every spiritual blessing” (Eph. 1:3); “every good gift” (James 1:17). . . . In no case of this usage is the adjective separated from the noun so as to be taken as a predicate.15
Roberts is to be commended for attempting to wrestle seriously with the syntactical phenomena of the text. In particular, he has clearly shown the error of arguments such as Dornier's: “Si l’adjectif qeovpneusto" tait une pithte, il devrait normalement tre plac avant grafhv; plac aprs, il se prsente comme un attribut...”16
However, there seems to be a fundamental logical error in Roberts’ presentation: by definition Roberts seems to deny the possibility of the second adjective being predicate. He first declares that the only constructions he is examining are those in which pa'" “is used to modify a noun which is immediately followed by another adjective [italics added].” But he concludes this survey by adding, “In no case of this usage is the adjective separated from the noun so as to be taken as a predicate [italics added].”17 Further, he assumes that the position of the adjective in an anarthrous construction is a reliable indicator of its relation to the noun: “In no case of this usage is the adjective separated from the noun so as to be taken as a predicate [italics added].”18 In actual NT usage, however, the adjective following a pa'"—noun construction may be separated from the noun by an intervening word or phrase without being a predicate adjective,19 and conversely, as we will demonstrate, it may immediately follow the noun, yet not be attributive.20
In addition to being circular, Roberts' argument is also semantically insensitive: all his examples, save perhaps one,21 are from non-equative clauses—i.e., clauses in which the main point (grammatically speaking) is not an assertion about the subject. In such clauses, predicate adjectives are indeed few and far between.22
But the construction in 2 Tim 3:16 belongs to an equative clause—i.e., a clause in which the central point (syntactically at least) is an assertion about the subject. Now a point of clarification is in order. I am not arguing in a circle: I am not saying that since qeovpneusto" is a predicate adjective, 2 Tim 3:16 is an equative clause. That would indeed be putting the cart before the horse! Rather, the construction pa'sa grafhV qeovpneusto" kaiV wjfevlimo" constitutes an equative clause because it requires an implied verb which, in turn, asserts at least that all scripture is profitable. This is entirely apart from the issue of qeovpneusto"’s special relation to grafhv.
A more valid approach than Roberts’, it seems to me, is one which focuses on equative clauses. We need to ask whether the adjective in a pa'"-noun-adjective construction in an equative clause is normally predicate or attributive. However, since pa'" is used in such constructions in the NT only a few times, it is necessary to expand our approach in two directions, though still concentrating on equative clauses: (1) We will touch on the slightly broader phenomenon of adjective-noun-adjective to see if this will help to inform the more specific pa'" -noun-adjective construction. But since there are only six such constructions in the NT, we also should get a representative sampling of usage in extra-NT Greek. (2) We will examine the pa'"-noun-adjective constructions in equative clauses in the LXX. The LXX is targeted for its special contribution because (a) the LXX is both Koine Greek and biblical Greek,23 and (b) the LXX can be examined exhaustively with reference to the pa'"-noun-adjective constructions via Hatch and Redpath’s Concordance.24
Besides an exhaustive study of the general phenomenon of anarthrous noun-adjective constructions in the NT, I have looked at representative portions from Homer, Herodotus, Thucydides, Demosthenes, Polybius, Josephus, select papyri, as well as the LXX.
In the 5,290 lines of text I perused, only three yielded instances of the adjective-noun-adjective construction in equative clauses.25 Two of these, coincidentally, were in Herodotus. In Book 1.8 we see tivna... lovgon oujk uJgieva. This, however, is not an ideal parallel for there is an intervening word between the first (pronominal and attributive) adjective and the noun and a negative particle separating the second (predicate) adjective from the noun. In 1.6, however, we see a clearer example: pavnte" {Ellhne" hsan ejleuvqeroi. Here the pronominal adjective is attributive while the second adjective is predicate. This affords an excellent parallel with 2 Tim 3:16, for those who affirm that qeovpneusto" is a predicate adjective are also convinced that the location of the implied equative verb or copula (at least, as far as the translation is concerned) is between grafhv and qeovpneusto". Herodotus’ example offers proof that such an understanding can be legitimate in Greek.26 The other reference is Zech 14:21 in which pa'" is attributive and a{gion (a{gio" in some witnesses) is predicate (e[stai pa'" levbh"...a{gion).
Although no other examples were found in equative clauses, I did find two more precise parallels to the total construction in 2 Tim 3:16 in the LXX. By ‘precise’ I mean adjective-noun-adjective-kaiv-adjective (which, in 2 Tim 3:16, is pa'sa grafhV qeovpneusto" kaiV wjfevlimo"). In Deut 7:1 the construction is eJptaV e[qnh pollaV kaiV ijscurovtera with eJptav functioning as an attributive and pollaV kaiV ijscurovtera functioning as predicates. In Gen 2:9 the adjective preceding the noun is pa'" (pa'n xuvlon wJrai'on eij" o{rasin kaiV kaloVn eij" brw'sin)—again, the adjectives following are predicate.27
Therefore, in the only three parallels to 2 Tim 3:16 in equative clauses I discovered in extra-NT literature, the second adjective was always predicate and the first adjective was attributive. And in the only two constructions in extra-NT literature which paralleled the complete construction of adjective-noun-adjective-kaiv-adjective in 2 Tim 3:16, even though both parallels were in non-equative clauses, the adjectives following the noun were predicate while the adjective preceding the noun was attributive. Although the examples are not numerous, it might be significant that they all point in one direction.
Adjective-noun-adjective constructions in equative clauses. I discovered only six instances in the NT, apart from those involving pa'", in which the construction in equative clauses was adjective-noun-adjective. In Matt 22:36 and Mark 12:28 (parallel passages) there may be ambiguity as to which adjective is predicate and which is attributive. However, in neither instance can both adjectives be construed as attributive. Various strands of grammatical, lexical, and historical evidence, in fact, suggest that in each case the preceding adjective is attributive and the following adjective is predicate.28 In Jas 3:8 kakovn is a substantival adjective with ajkatavstaton preceding it and serving in an attributive role. However, these two words form the main body of the clause, with mesthv serving in an appositional capacity (in a sense) to kakovn. Technically, kakovn is in the predicate with an implied subject, ajkatavstaton is an attributive adjective modifying this substantival adjective, and mesthv, though functioning as a predicate adjective, is functioning thus in its own appositional phrase, not in the main clause. Rev 16:18 affords a similar example (thlikou'to" seismoV" ou{tw mevga").
The gospel of Luke furnishes better examples. In 19:17 we see ajgaqeV dou'le... pistov". The first adjective is attributive and the second adjective is predicate. However, the second adjective is in the following clause (though the noun is only implied in the verb). But in 4:24 we have oujdeiV" profhvth" dektov" ejstin. Here the first (pronominal) adjective is attributive and the second adjective is predicate.
Thus in constructions not involving pa'" no example had both adjectives functioning as attributives. However, there was ambiguity in two texts as to which adjective was attributive and which was predicate; two others had the second adjective outside of the main clause; and only one was a clear instance of attributive-noun-predicate. The pa'" constructions hopefully will give us a clearer picture.
Pa'"-noun-adjective constructions in equative clauses. Besides 2 Tim 3:16, there are at least four more similar constructions in the NT. In Luke 2:23 the construction is pa'n a[rsen...a{gion...klhqhvsetai. Here the pronominal adjective is attributive and the following adjective is predicate. In Jas 1:19 the construction is e[stw deV pa'" a[nqrwpo" tacuv"... braduv"...braduv". Here the pronominal adjective is attributive and all three adjectives following the noun are predicates. This, then, is a step closer to the 2 Tim 3:16 construction, for it too involves more than one adjective following the noun. In Jas 4:16 we see pa'sa kauvchsi" toiauvth ponhrav ejstin. This is the first clear instance in either the NT or extra-NT literature which we have examined in which both the preceding and trailing adjectives are attributive. Yet in this isolated example, the linguistic situation veers off from what we see in 2 Tim 3:16 in one very important point, viz., the presence of toiauvth as the trailing adjective. toiouvto", as other pronominal adjectives, may stand outside of the article-noun group but still have an attributive relation to the noun (cf. aiJ dunavmei" toiau'tai in Mark 6:2). In fact, with this in mind, we could well argue that Jas 4:16 fits neatly with the “predicate qeovpneusto"” view, for the following adjective, ponhrav, is predicate. In 1 Tim 4:4 the structure is even closer to that of 2 Tim 3:16: pa'n ktivsma qeou' kalovn, kaiV oujdevn... Here it is obvious that the first adjective is attributive and the second is predicate. There is the further parallel in that the second adjective is joined by kaiv to the word in the predicate, oujdevn. It might be objected that oujdevn here is used substantivally and therefore does not afford an exact parallel with 2 Tim 3:16. However, the parallel is not at all diminished for wjfevlimo" in 2 Tim 3:16, as oujdevn here, could grammatically stand in the predicate alone. This text, then, is the closest parallel to 2 Tim 3:16 in the NT. The fact that it, too, is in a pastoral epistle adds weight to the view that qeovpneusto" in 2 Tim 3:16 is a predicate adjective.
As with the extra-NT evidence, and the adjective-noun-adjective constructions within the NT, the pa'"-noun-adjective construction suggests the same semantics: the first adjective is attributive and the second is predicate. We have found no clear exceptions to this principle.29 But admittedly, the examples are few: altogether only fourteen adjective-noun-adjective constructions were found in the extra-NT and NT literature. Nevertheless, this monolithic trend can hardly be used in support of an attributive qeovpneusto" in 2 Tim 3:16.
The singular semantic path down which our construction has gone has led me to propose, as a working hypothesis, the following ‘rule’: In pa'"-noun-adjective constructions in equative clauses the pa'", being by nature as definite as the article, implies the article, thus making the adjective(s) following the noun outside the implied article-noun group and, therefore, predicate.30 This hypothesis can be put through a rigorous test which is inductively complete and self-contained.31
In perusing the more than 6,000 entries on pa'" in Hatch-Redpath, I culled from them (what I believe to be) all the pa'"-noun-adjective constructions in equative clauses. Altogether, I discovered thirty-six such constructions. Remarkably, in thirty-five instances the pa'" was definitely attributive and the adjective(s) following the noun was/were definitely predicate. For example, pa'sai yucaiV eJptav in Gen 46:25; pa'sa qusiva iJerevw" oJlovkrauto" e[stai in Lev 6:23(16); pa'" ajnhvr ...divkaio" in Prov 21:2; pa'sa kefalhV falakrav in Ezek 29:18.32
On only one occasion was there ambiguity. In 2 Kings (4 Kingdoms) 19:35 (pa'nte" swvmata nekrav) it was questionable as to which adjective was attributive and which was predicate.33 But even here it was not possible to construe both adjectives as attributive. Thus this one possible exception to the ‘rule’ in no way supports an attributive qeovpneusto" in 2 Tim 3:16.
On six occasions I discovered the construction pa'"-noun-adjective-kaiv-adjective, which is an even stronger parallel to the construction in our target passage: hsan deV pa'sai yucaiv... pevnte kaiV eJbdomhvkonta in Exod 1:5; pa'n ajrsenikovn...ojktakiscivlioi kaiV eJxakovsioi in Num 3:28; pa'n ajrsenikovn... eJxakiscivlioi kaiV penthvkonta in Num 3:34; pa'n ajrsenikovn duvo kaiV ei]kosi ciliavde" in Num 3:39; h pa'" Israhl civliai kaiV eJkatoVn ciliavde" in 1 Chron 21:5; pa'" deV tovpo"... a[bato" kaiV purifleghV" ginevsqw in 3 Macc 3:29. On each of these occasions both adjectives following the noun were predicate. Hence, these verses add substantial weight as fairly precise parallels,34 along with 1 Tim 4:4, to 2 Tim 3:16.
The totality of this septuagintal evidence was so overwhelmingly in support of the ‘rule’ suggested in this paper that I felt compelled to pursue one more validation process. If it is true that the article is implied in the pa'" in pa'"-noun constructions in equative clauses, and that any adjective following the pa'"-noun construction would be considered in the predicate, then it ought also be true that any adjective preceding the pa'"-noun construction would be in the predicate. After all, if the pa'" in such constructions implies the article, then it should not matter, ex hypothesi, which side of the article-noun group the adjective falls: either way, it should still be predicate.
I tested this hypothesis by again examining the entries on pa'" in Hatch-Redpath. I discovered ten adjective-pa'"-noun constructions in equative clauses. In each instance the adjective preceding the pa'"-noun group was clearly predicate.35
To summarize the septuagintal evidence: thirty-five of thirty-six pa'"-noun-adjective constructions in equative clauses definitely supported the ‘rule.’ One was questionable, though it in no way viewed both adjectives as attributive. All ten adjective-pa'"-noun constructions supported the ‘rule.’ Altogether, in forty-six syntactical parallels to our passage, at least forty-five support a predicate qeovpneusto" and the one possible exception does not support an attributive qeovpneusto".
In this paper I have sought to demonstrate that the structural phenomenon of 2 Tim 3:16 does not create a grammatical impasse. That is to say, we do not need to rely solely on intuition nor quickly move on to contextual factors to understand the relation of qeovpneusto" to grafhv. There is a wealth of information provided by syntactical parallels which bring into sharp relief what appears to be the truly idiomatic nature of the construction.
A pioneer in giving priority to the syntax in 2 Tim 3:16 was J. W. Roberts. Yet, as significant as his study was, it involved certain weaknesses in method which resulted in a rather distorted view of the nuance of qeovpneusto".
By restricting our study to equative clauses—of which there was a fairly ample supply—we believe that we have taken a more valid approach. The results, therefore, are more sure-footed. And, it bears repeating, the semantic force of our construction was so one-sided that we could suggest a principle based on it: In pa'"-noun-adjective constructions in equative clauses, the pa'", being by nature as definite as the article, implies the article, thus making the adjective(s) following the noun outside the implied article-noun group and, therefore, predicate.36
As applied to 2 Tim 3:16, this principle indicates that a predicate qeovpneusto" is certainly a valid—and perhaps the only—option. Hence, we translate the passage, “All/every scripture is inspired and profitable. . .”37 In the least, our study suggests that the REB’s rendering “Every inspired scripture has its use” should probably be relegated (in our present state of knowledge) to the margin.
1 I do not here mean to make a sharp dichotomy between exegete and theologian, for it is difficult (and, in my view, undesirable) to refrain from trespassing into the other’s realm from time to time (see S. L. Johnson, “Romans 5:12—An Exercise in Exegesis and Theology,” in New Dimensions in New Testament Study, edd. R. N. Longenecker and M. C. Tenney [Grand Rapids: Zondervan, 1974] 298-316, for a good that exegetes—if they are properly to understand the text—must also be sympathetic with the theological concerns of the primitive church).
2 Of course, as such this is typically a concern—almost exclusively—of theological conservatives, especially in American evangelical circles, as may be seen by the bulk of the references in n. 11 (and passim). There are at least two reasons why work on this passage is usually the domain of evangelicals: (1) concern over the self-witness of scripture usually implies a high view of scripture—i.e., that it is something more than a mere human product; (2) those scholars who affirm the pastorals’ authenticity are finding themselves in an ever-shrinking circle. In spite of one’s estimation of scripture in general or the authorship of the pastorals in particular, all should recognize that these epistles constitute a significant part of the religio-literary records of the primitive church. And the fact that they eventually merited membership in the canon indicates something of their strategic role in either shaping or affirming the beliefs of nascent Christianity. Purely for the sake of a properly informed historical exegesis, therefore, New Testament scholarship as a whole should not neglect such a text.
3 C. K. Barrett, The Pastoral Epistles in the New English Bible (Oxford: Clarendon, 1963) 114.
4 G. L. Bahnsen, “Autographs, Amanuenses and Restricted Inspiration,” EvanQ 47 (1975) 162-67; E. W. Goodrick, “Let’s Put 2 Timothy 3:16 Back in the Bible,” JETS 25 (1982) 481-83.
5 This is one stone rarely left unturned by exegetes.
6 See M. R. Austin, “How Biblical is ‘The Inspiration of Scripture’?”, ExpTi 93 (1981-82) 76-79; H. W. House, “Biblical Inspiration in 2 Timothy 3:16,” BSac 137 (1980) 57-58; Goodrick, 484-85, for some recent studies. For a now classic treatment of this text, see B. B. Warfield, Revelation and Inspiration (Oxford: University Press, 1927) 229-59. Curiously, BAGD cites Warfield under qeovpneusto", while the first edition of BAG lacks this citation.
7 Secondary syntactical evidence would include the force of the asyndetic clause, the implied placement of the missing copula, and the normal, even routine, use of kaiv when joining two adjectives. Besides these grammatical clues as to the author’s meaning, however, there are contextual indicators which, though less decisive here (for they usually degenerate into a presuppositional and, therefore, circular form of argument), are frequently trumpeted as the coup de grce in the light of the alleged ambiguity of the syntax of the clause.
8 I do not regard grammar as a panacea for all of our exegetical ills, of course, but in this particular text I do think that the syntax of the construction has been too quickly passed over.
10 Goodrick, 483. To be sure, I have little doubt that Goodrick would appreciate some hard data which would corroborate how he reads the text (so, perhaps, ‘deprecate’ is too strong a term).
11 So Barrett, Schweitzer, Leaney, Dibelius, Spicq, et al. Here is a good place, it seems, to catalog some of the specialized studies done in the last few decades over this issue.
On the side of an attributive qeovpneusto": J. W. Roberts, “Every Scripture Inspired of God,” RestQ 5 (1961) 33-37; J. W. Roberts, “Note on the Adjective after Pa'" in 2 Timothy 3:16,” ExpTi 76 (1964-65) 359; R. J. A. Sherriffs, “A Note on a Verse in the New English Bible,” EvanQ 34 (1962) 91-95; E. L. Miller, “Plenary Inspiration and II Timothy 3:16,” LuthQ 17 (1965) 56-62.
For a predicate qeovpneusto": J. H. Bennetch, “2 Timothy 3:16a: A Greek Study,” BSac 106 (1949) 187-95; D. C. Oakley, “The Contribution of the Greek Text to the Doctrine of Inspiration” (Th.M. thesis, Dallas Theological Seminary, 1955) 22-26; J. A. Witmer, “The Biblical Evidence for the Verbal-Plenary Inspiration of the Bible,” BSac 121 (1964) 243-52; J. P. Eidsness, “An Exegesis of Important Passages Relative to the Doctrine of Inerrancy,” (Th.M. thesis, Dallas Theological Seminary, 1968) 54-55; T. P. McGonigal, “‘Every Scripture is Inspired’: An Exegesis of 2 Timothy 3:16-17,” Studia Biblica et Theologica 8 (1978) 53-64; House, 54-63; Goodrick, 479-87.
12 On this score, the analogy that Goodrick draws concerning Günther Zuntz’s intuitive interpretation (483) breaks down, for Epictetus (where Zuntz, in Goodrick’s analogy, applied his intuition) is a far cry from the NT which respect to the presuppositional (and emotional) baggage that one brings to the text.
13 In particular, both Sherriffs and House staunchly affirm the structural ambiguity of the passage, though they both come down (quite tentatively) on opposite sides of the fence.
14 Though based, in part, on an earlier study by R. M. Spence (ExpTi 8 [1896-97] 563-65).
15 Roberts, “Every Scripture,” 35 (all references to Roberts will be to this article as his later essay in ExpTi was merely a summary of this one). “The other examples are Acts 23:1; 2 Cor. 9:8; Eph. 4:29; Col. 1:10; 2 Thess. 2:17; 2 Tim. 2:21; 4:18; Titus 1:16; 2:10; 3:1; Heb. 4:12; James 3:16; Rev. 8:7; 18:2; [sic] 12: [sic] 21:19” (ibid.).
16 P. Dornier, Les pitres Pastorales (Paris: LeCoffre, 1969) 234. Cf. also J. N. D. Kelly, A Commentary on the Pastoral Epistles (London: A & C Black, 1963) 203, et al.
17 Ibid. There are four other, though somewhat less significant, problems with Roberts’ argument. First, his calculations are incorrect for there are more than twenty-one instances of the pa'"-noun-adjective construction in the NT. We should add to the list these seven: pa'n ai|ma divkaion in Matt 23:35; pavnta a[nqrwpon tevleion in Col 1:28; pavnto" ei[dou" ponhrou' in 1 Thess 5:22; pantiV e[rgw/ ajgaqw'/ in 1 Tim 5:10; pa'n e[rgon ponhrav in Jas 4:16. Second, Roberts does not consistently follow his own rule for not all of his examples fit the structural definition of an adjective immediately following the same noun which pa'" precedes (cf. ejn pantiV e[rgw/ kaiV lovgw/ ajgaqw'/ in 2 Thess 2:17 and pa'san pivstin ejndeiknumevnou" ajgaqhvn in Titus 2:10). Third, the fact that Roberts has two examples which do not fit precisely the structural pattern he claims to be examining opens the door for two other examples, both of which alter the picture he has painted (cf. pa'n a[rsen dianoi'gon mhvtran a{gion in Luke 2:23 and pa'n ktivsma qeou' kalovn, kaiV oujdeVn ajpovblhton in 1 Tim 4:4). Finally, rather than intentionally distorting the data, I think that Roberts has contrived something of a mixed-bag method: although he is at first dealing with the relation of qeovpneusto" to grafhv, it is apparent that his discussion subtly picks up, perhaps unconsciously, a secondary issue, viz., the translation of pa'" (34-35: he introduces the paragraph we have quoted by arguing for the lexical force of pa'" [in this context, at least] as meaning ‘every’ rather than ‘all’; then, at the beginning of our paragraph he combines this with the syntactical force of qeovpneusto", a point which occupies the major thrust of the paragraph. But toward the end of the paragraph he curiously switches back to the meaning of pa'", arguing that “the one place where the translations have ‘all’ is Titus 2:10 where the context shows that the word pas means ‘perfect’ or ‘complete’ faith” [35]).
19 Cf. 2 Thess 2:17 and, perhaps (though I would argue differently) Titus 2:10. (Cf. also examples cited in n. 34.)
20 Cf. pavnta a[nqrwpon tevleion in Col 1:28, an object-complement construction (see my article, “The Semantics and Exegetical Significance of the Object-Complement Construction in the New Testament,” GTJ 6 [1985] 99-100 [n. 40] for a discussion of this text), pa'" a[nqrwpo" tacuv" in Jas 1:19. There are other instances outside the NT which we will address shortly.
21 pa'sa dovsi" ajgaqhV kaiV pa'n dwvrhma tevleion ... ejstin katabai'non in Jas 1:17, according to Roberts, forms an equative clause; yet it is possible that ejstin katabai'non is periphrastic, rendering the whole clause as non-equative. On this, see discussion in n. 29.
22 Of the 2,054 anarthrous noun-adjective constructions in non-equative clauses I discovered in the NT, at most only seven percent involved predicate adjectives (typically, in object-complement constructions). See my essay, “The Relation of Adjective to Noun in Anarthrous Constructions in the New Testament,” NovT 26 (1984) 156, 159.
23 By ‘biblical Greek’ I do not intend to convey agreement with Turner’s conclusion “that Bibl. Greek is a unique language with a unity and character of its own” (J. H. Moulton, A Grammar of New Testament Greek, 4 vols, vol 3: Syntax, by N. Turner, 4).
24 For purposes of this study, and the one on which it is based (Wallace, “The Relation of Adjective to Noun,” 132), the standard manual editions of Nestle-Aland25 and Rahlfs (for LXX) were used, though the results were checked against more critical works. It should be noted that the database was compiled before Nestle-Aland26-27, Gramcord/Accordance, or the Thesaurus Linguae Graecae were available. Thus, all the work had to be done manually.
25 The portions examined were: Homer Odyssey 1.1-62, Iliad 18.1-165; Herodotus Book 1.1-22, Book 2.1-6 (Stein’s edition); Thucydides Book 1.1-14.3, Book 2.1.1-2.5 (Hude’s edition); Demosthenes Book 1.1-28 (First Olynthiac), Book 9.1-46 (Third Philippic); Polybius The Histories 1.1.1-11.8; Josephus Jewish Antiquities 15.1-99; Elephantine Papyrus 1,11.1-18; Elephantine Papyrus 2,11.1-18; Cairo Zenon Papyrus 59426.1-8; Cairo Zenon Papyrus 59251.2-22; Elephantine Papyrus 13.1-15; Urkunde der Ptolemerzeit 59.1-33; Urkunde der Ptolemerzeit 62.1-36; Tebtunis Papyrus 110.1-15; Berlin Griechische Urkunde 1103.2-30; and Berlin Griechische Urkunde 1121.1-46. In the LXX, I examined (apart from the specialized study on pa'"-noun-adjectives which was culled, initially at least, from Hatch-Redpath) Gen 1:1-4:26; Exod 20:1-26; Lev 19:1-37; Deut 6:1-8:15; Psalms 1, 2, 17 (18), 21 (22), 31 (32), 38 (39), 50 (51), 70 (71) 118 (199), 149, 150; Isa 40:1-44:28; 49:1-53:12; Jonah 1:1-4:11; Zech 12:11-14:21.
26 The placement of the verb and the sense of the construction, however, do not necessarily have a one-to-one correspondence. Note the example in Zech 14:21 as well as those listed in n. 34.
27 These two texts are listed here for the sake of completeness, though their relevance to our target passage, because they involve non-equative clauses, may be minimal.
28 For a discussion, see my thesis, “The Relation of Adjective to Noun in Anarthrous Constructions in the New Testament” (Th.M. thesis, Dallas Theological Seminary, 1979) 49-51.
29 There is one twofold example which seems, prima facie, to violate this principle. In Jas 1:17 pa'sa dovsi" ajgaqhV kaiV pa'n dwvrhma tevleion are clearly fully attributive constructions. However, what is not so clear is whether the clause is equative or non-equative. In the words following (a[nwqevn ejstin katabai'non) it is possible to treat the verb and the participle in separate clauses or as a periphrastic construction. If they are periphrastic (in which case a[nwqen would qualify pa'sa dovsi" ajgaqhV kaiV pa'n dwvrhma tevleion rather than function predicatively to ejstin) then the whole clause would belong to the non-equative category. In light of the evidence amassed thus far, and in light of this distinct grammatical possibility, we cannot cite Jas 1:17 as a clear exception to the principle we have suggested.
30 One objection to this principle at the outset might be concerned with its confinement to constructions involving pa'". Pa'", unlike most adjectives, does not need the article to make the noun any more definite. Thus it might also be added here that other pronominal adjectives which are equally definite would, in all probability, fit the ‘rule’ as well. We see this with the numeral ei|" in Eph 4:6, for example, in which the wording approximates a “Granville Sharp construction” (for which see my essay, “The Semantic Range of the Article-Noun-Kaiv-Noun Plural Construction in the New Testament,” GTJ [1983] 61-63): ei|" qeoV" kaiV pathvr. Here it is obvious that pathvr refers to the same person as qeov" (note also tou'to in Acts 5:31). But are there examples of pronominal adjectives in adjective-noun-adjective constructions? Significantly, of the six instances of adjective-noun-adjective constructions in equative clauses we examined, the clearest parallel to the structural phenomenon in 2 Tim 3:16 was in Luke 4:24 (oujdeiV" profhvth" dektov" ejstin). Here the pronominal adjective does indeed function attributively while the second adjective is predicate.
31 By this I mean only that an entire body of literature (the LXX) can be examined exhaustively, rather than selectively, through the use of a concordance (Hatch-Redpath).
32 Cf. also Gen 46:22, 26, 27; Exod 1:5; Lev 11:32, 34 (bis); 13:58; 15:4 (bis), 9, 17, 24, 26; 17:15; 27:11, 28; Num 3:28, 34, 39; Josh 21:26; Judg 20:17; 1 Sam 11:8; 1 Chron 2:4; 21:5; Prov 3:15; 8:1; Jer 9:26; Zech 14:21; Sirach 23:17; 3 Macc 3:29.
33 Of course, if pavnte" is not modifying the noun (which is quote probable, due to the lack of gender concord) then we would most naturally treat it substantivally as a pronoun.
34 It might be objected that these examples do not provide precise parallels because (1) in the first five instances, the kaiv joined two numerals in such a way that it would be impossible to treat these adjectives as bearing a different relation to the noun (e.g., if we were to consider pevnte in Exod 1:5 as attributive and eJbdomhvkonta as predicate, we would get the nonsense reading of “all five people were also seventy”!); (2) in the last text cited (3 Macc 3:29), even though this passage does not involve the “numbers idiom,” the verb is expressed (ginevsqw), rendering it more explicit than the construction in 2 Tim 3:16.
In response, one should note that: (1) The very fact that the trailing adjectives in five of the examples can only be taken as predicates is hardly an argument against a predicate qeovpneusto" in 2 Tim 3:16. These instances may, in fact, be merely an extension—to the point of a set idiom—of the semantics of the pa'"-noun-adjective construction we have already seen in the LXX. Further, all grammatical study must proceed on the basis of an indisputable semantic nuance for the particular construction under consideration. That all of the indisputable examples of the construction which is the concern of this paper affirm only that the adjective(s) following pa'"-noun in an equative clause is/are predicate just might indicate that such was part of the warp and woof of hellenistic Greek. (2) Although the verb is expressed in 3 Macc 3:29, its location gives no hint as to whether the trailing adjectives should be treated as attributives or predicates. Hence, it affords a decent parallel to our target passage, for if the author of 2 Timothy had added a verb after wjfevlimo" in 3:16 (paralleling exactly the construction in 3 Macc 3:29), the debate over the relation of qeovpneusto" to grafhv would hardly thereby have been settled. Further even when the verb does stand between noun and adjective, this is not a sure indicator that the adjective belongs in the predicate (cf., e.g., fw'" eiden mevga in Matt 4:16; tufloiv eijsin oJdhgoiv in Matt 15:14; mikraVn e[cei" duvnamin in Rev 3:8). Again, we submit that neither the presence (or absence) of the verb, nor its location in the clause, is the primary factor which determines the relation of adjective to noun in any given instance. (3) Admittedly, most of our parallels from the LXX employ the copula while it is absent from 2 Tim 3:16. But to demand that the parallels be more precise than pa'"-noun-adjective in an equative clause (especially since the presence and location of the verb are not decisive matters) just might define any parallels out of existence. I am reminded here of W. Grudem’s recent insight in connection with the syntax of 1 Pet 3:19 (“Christ Preaching through Noah: 1 Peter 3:19-20 in the Light of Dominant Themes in Jewish Literature,” TrinJ [1986] 22):
... it is exegetically illegitimate to demand parallel examples which are so narrowly specified that one would not expect to find many, if any, examples. (It would be similar to saying that w , “of whom,” in [1 Peter] 3:3 cannot refer to “wives” because there is no other example of a relative pronoun taking as its antecedent an articular feminine plural vocative...!)
Consequently, though the parallels we have uncovered do not qualify for the accolade “exact parallels,” they are as precise as the extant literature we have examined has turned up. In the least, it would appear that the burden of proof rests on the shoulders of the one who wishes to see attributive adjectives in such constructions.
35 Deut 27:26; Prov 3:32; 11:20; 16:5; 22:11; Judith 16:16 (bis); Wisdom of Solomon 13:1; Sirach 25:19; 4 Macc 9:29.
36 One should not be under the delusion that this ‘rule’ is absolute. As we have already pointed out, there may be exceptions to it within even the NT itself (Jas 1:17; 4:16). And I am sure that exceptions from extra-biblical Greek may well be produced. (Indeed, since the writing of the first draft of this paper, I have found one clear exception to this principle. In Didache 13.1 we read that “Every true prophet... is worthy of his food” [Pa'" deV profhvth" ajlhqinov"... a[xio" ejsti th'" trofh'" aujtou']. Here, ajlhqinov" is attributive even though the clause is equative.) After all, a database of about fifty examples is simply too small a foundation from which to build a dogmatic superstructure. Still, it is striking to me that even these already-encountered potential exceptions have grammatical peculiarities which render them debatable. The preponderance of evidence within the pages of the Greek Bible is so staunchly on the side of a predicate qeovpneusto" that, in my opinion, advocates of an attributive qeovpneusto" must either counter with strong grammatical evidence of their own or else make out a better case on non-syntactical fronts. In other words, from all available evidence, the first readers of 2 Timothy would be expected to read qeovpneusto" as a predicate adjective since there is nothing compelling in the context to detour them from this most natural understanding.
We might add here a further clarification of the ‘rule’: By saying that pa'" implies the article, we do not mean that the construction is semantically identical, on all levels, to pa'sa hJ grafhv. For if that were the case we would have to read the text as ‘all scripture,’ rather than ‘every scripture,’ yet the evidence from the LXX mitigates the necessity of the former translation (cf. 1 Chron 21:5; 3 Macc 3:29). The ‘rule,’ then, extends only to defining the relation of the trailing adjective to the noun.
37 The NET Bible NT very accurately has “every scripture is inspired by God and useful for teaching...”