Greek First Class Conditions: Ledgerwood

                                        Grace Theological Journal 12.1 (1992) 99-118.

                         [Copyright © 1992 Grace Theological Seminary; cited with permission;

                                          digitally prepared for use at Gordon College]




                  CLASS CONDITIONAL IMPLY?


                      THE TESTIMONY OF THE



                                                L. W. LEDGERWOOD III


            Debate has been engaged for more than a century over what im-

plications, if any, a Greek First Class Conditional (FCC) has concern-

ing the proposition in its protasis. Some pedagogical grammars claim

that the Greek FCC is well translated with the English causative con-

struction introduced with "since." In this paper a twofold approach is

used to show that this claim is in error.

            First, a methodology for formulating and testing hypotheses con-

cerning historical languages is established. The methodology is based

on a Popperian view of hypothesis testing. In this case a testable hy-

pothesis is formed utilizing the descriptive apparatus of H. P. Grice.

The hypothesis is that the FCC is well translated with English "since"

and it is proven false.

            Second, the testimony of four ancient Greek grammarians is eval-

uated. The grammarians examined are: Dionysius Thrax (1st century

BCE), Apollonius Dyscolus (2nd century CE), Stephanos and Hel-

liodorus (Byzantine period). It is shown that these grammarians agree

with the conclusion that it is not appropriate to translate the FCC with

an English causal introduced by" since."

                                                *     *     *


                                                I. INTRODUCTION

DOES a Koine Greek conditional sentence introduced by ei] ("if")

with the indicative imply the truth of the proposition in its prota-

sis? Debate on this issue has been engaged for over 100 years. In the

19th century two of the major participants in the debate were William




Goodwin1 and Basil Gildersleeve.2 Early in this century, A. T. Robert-

son,3 claiming to be in the Gildersleevian tradition, asserted that the

truth of the proposition in the protasis is implied to be true or at least

assumed true for the sake of argument. Some modern pedagogical

grammars follow Robertson's assertions and carry them to an extreme

that Robertson himself did not.

These pedagogical grammars claim that a Greek conditional intro-

duced by ei] with the indicative should be translated with an English

causal construction. That is, a sentence like:


(1a)  Ei] ou#n sunhge<rqhte t&? Xrist&? ta> a@nw zhtei?te  (Col 3: 1)


should be translated with the causal (lb) below and not with the condi-

tional (lc).


(1b) Since then you have been raised up with Christ, keep seeking the

things above.

(lc) If then you have been raised up with Christ, keep seeking the

things above.


They claim that sentence (la) implies that the proposition in its prota-

sis, namely, "You have been raised up with Christ," is true and for this

reason an English causal sentence should be used. Recently, James

Boyer4 argued that such a claim is in error.

This debate has been clouded by at least two factors: ambiguity of

terms and hypotheses formulated in an untestable manner. For this rea-

son, no one has achieved a level of proof on which all can agree. How-

ever, H. P. Grice5 has developed linguistic theory which provides a

descriptive apparatus in which testable hypotheses concerning implica-

tions can be formulated. Using Grice's descriptive apparatus it is pos-


l Wi11iam Goodwin, "The classification of Conditional Sentences in Greek Syntax,"

in Journal of Philology 15 (1874) 188-205; "'Shall' and 'Should' in Protasis, and Their

Greek Equivalents," in Journal of Philology 18 (1877) 18-38; Syntax of the Moods and

Tenses of the Greek Verb (London: MacMillan, 1889); Greek Grammar (London: Mac-

Millan, 1879, reprinted by St. Martin's, 1878) §§ 1381-1424.

2 Basil L. Gildersleeve, "Studies in Pindaric Syntax," in American Journal of Phi-

lology, 3 (1882) 434-55; "A Reply to E. B. Clapp," in American Journal of Philology 9

(1888) 491-92; "Stahl's Syntax of the Greek Verb," in American Journal of Philology

30 (1909) 1-21.

3 A. T. Robertson, A Grammar of New Testament Greek in Light of Historical Re-

search (Nashville: Broadman, 1934).

4 James L. Boyer, "First Class Conditionals, What Do They Mean?" in Grace

Theological Journal 2.1 (1981) 75-114.

5 R. P. Grice, "Logic and Conversation," in Syntax and Semantics 3, Speech Acts,

ed. P. Cole and J. P. Morgan (New York: Academic, 1975) 41-58; R. P. Grice, "Further

Notes on Logic and Conversation," in Syntax and Semantics 9, Pragmatics, ed. J. M. Sa-

dock (New York: Academic, 1978) 113-27.



sible to define a clear and unambiguous hypothesis to test whether or not

the claim of these pedagogical grammars is indeed sound. In the fol-

lowing paper, the assertions of some grammarians over the past century

are reviewed. The claim of the pedagogical grammars which assert that

a first class conditional should be translated with English "since"is for-

mulated into a testable hypothesis. The methodology employed proves

unambiguously that conditional sentences introduced with ei] plus the

indicative do not imply the truth of the proposition in the protasis.

In the debate over the implications of Greek conditionals, no one

has gone back to examine what ancient Greek grammarians said about

the issue. A second purpose of this paper is to do just that. The relevant

claims of Greek grammarians from 200 B.C. to A.D. 600 are reviewed.

These confirm that conditional sentences introduced with ei with the

indicative do not imply that the proposition in the protasis is true.



There are two conditional particles in Greek: ei] and e]a<n. Readers

of this paper not familiar with Greek may, for the time being, consider

both ei] and e]a<n to mean "if" neglecting any differences in meaning

between them. Greek also has a causal particle e]pei< which is well

translated by the English "since."

Many grammarians categorize the Greek conditionals in different

ways and use different names for their categories. Only two of the

forms of the conditionals will be discussed in this paper: the forms

many grammarians call the first and third class conditionals. The

causal construction will also be discussed. The following notational

shorthand will be used to refer to these constructions.


Shorthand                   Syntactic form                       Common name

ei] p,q               ei] + indicative, indicative                 first class conditional

e]a<n p,q           e]a<n + subjunctive, indicative           third class conditional

e]pei< p,q           e]pei< + indicative, indicative             causal construction


In this notation, "p" and "q" are variables representing clauses in the

protasis and apodosis respectively.




A. William Goodwin

William Goodwin sets forth his claims in no uncertain terms:

(2) Probably no grammarian would now maintain the absurdity that the

indicative in the protasis expresses either "certainty in fact" or

"what is believed by the speaker to be true." . . . Most grammarians



are eager to disclaim any connection between the "certainty" here

intended and the matter of fact or even opinion; and thus they

reduce the "certainty" to a harmless abstraction, which is utterly

valueless as a definition. . .


I have now nothing to change the statement which I made in

1864, . . . Every example that I have met has only confirmed the

opinion, which I now express with the greatest confidence that

there is no inherent distinction between the present indicative [ei]

p,q] and present subjunctive [e]a<n p,q] in the protasis, except that

of time6 (Goodwin's emphasis).


Goodwin spends the bulk of his article on aspectual and temporal

differences between conditionals of the form e]a<n p,q and ei] p,q when

the proposition q is expressed with a future indicative.


B. Basil Gildersleeve

Concerning the first class condition Gildersleeve says:

(3) It is used of that which can be brought to the standard of fact; but

the standard may be for or against the truth of the postulate. All

the logical condition [ei] p,q] asserts in the inexorable connection

of the two members of the sentence. It is the favorite condition in

argument. . . when one wishes to be or seem fair. . . when one is

sure of the premise. . . . But so long as the negative continues to

be mh<, the conditional and the causal do not coincide. . . . In

prose, it is semi-causal.7


An observation to make concerning this passage is that Gildersleeve

does not say that ei p,q implies that the proposition p is true like a

causal e]pei< p,q does. On the contrary, he even says it does not do so.

Robertson claims to be in the Gildersleevian tradition. However, the

terminology he uses is not as concise as Gildersleeve's and he has been

interpreted by some to suggest more than Gildersleeve did, namely that

ei] p,q implies the truth of p.


C. A. T. Robertson

Robertson says concerning these conditionals:

(4) This theory in brief is that there are four classes of conditions

which fall into two groups or types. The two types are the deter-


6 Goodwin, "Conditional Sentences in Greek Syntax," in Journal of Philology 15

(1874) 189-90.

7 Gildersleeve, "Studies in Pindaric Syntax," in American Journal of Philology 3

(1882) 435.



mined [ei] p,q is in this group] and the undetermined [e]a<n p,q is in

this group]. The point in "determined" [ei] p,q] is that the premise

or condition is assumed to be true. . . . The indicative is used for

this type. . . The other type is the undetermined condition. Natu-

rally the indicative is not allowed here. The element of uncer-

tainty calls for the subj. or the optative. . . .8 In broad outline

these four classes of conditions may be termed Reality [ei] p,q],

Unreality, Probability [e]a<n p,q] and Possibility. . . . This brings

us to the other theory. . . expounded by Goodwin. . . . Goodwin

confuses the "fact" with the "statement" of the fact. He describes

his first condition thus: "When the protasis simply states a present

or past particular supposition, implying nothing as to the fulfill-

ment of the condition, it takes a present or past tense of the indic-

ative with ei]." The words to which I object. . . are "implying

nothing as to the fulfillment of the condition." This condition [ei]

p,q] pointedly implies the fulfillment of the condition. . . . This is

the crux of the whole matter9 (Robertson's emphasis).


Robertson moderates his stance slightly to account for the many

examples in which ei] p,q clearly does not imply truth of the proposi-

tion in the protasis. Such an instance is Matt 12:21, where Jesus says,

"If [ei]] I cast out demons by Beelzebul . . ." Concerning this Robert-

son says,

(5) This class of condition [ei] p,q] assumes the condition to be a

reality and the conclusion follows logically and naturally from

that assumption. . . This condition therefore, taken at face value,

assumes the condition to be true. The context or other light must

determine the actual situation. This is a good example (cf. also

Gal 5:11) to begin with, since the assumption is untrue in fact,

though assumed to be true by Jesus for sake of argument.10


What Robertson is saying here is that Matt 12:21 should be translated,

"Assuming for the moment that I do cast out demons by Beelze-

bul. . ." instead of with the causative, "Since I cast out demons by

Beelzebul . . ." In this statement Robertson makes it clear that he is

not asserting that the propositions in the protasis are in fact true.

However, Robertson's claims are vague and untestable. He calls

the condition of the type ei] p,q "determined," in contrast to "undeter-

mined." He calls it a condition of "reality," in contrast to "possibility."

He says that this condition assumes the premise to be true, in another

that it pointedly implies the fulfillment of the condition and finally that


8 Robertson, Greek Grammar (Nashville: Broadman, 1934) 1004.

9 Robertson, Greek Grammar, 1005-6.

10 Robertson, Greek Grammar, 1007-8.



it assumes the condition to be a reality. Apparently misunderstanding

Robertson, some pedagogical grammars, which claim Robertson as

their authority, have gone so far as to identify conditionals of the form

ei] p,q with causal constructions.


D. The Claim of Summer's Pedagogical Grammar

Only one of the pedagogical grammars is quoted here as an

example of what some of Robertson's followers claim. Others may be

examined by the interested reader.11 Ray Summers, in his pedagogical

grammar says,

(6) The first class condition [ei] p,q] affirms the reality of the condi-

tion. . . "ei] maqetai> tou? kuri<ou e@smen swqh<setai" . . . This con-

struction is best translated, "Since we are disciples of the Lord,

we shall be saved.”12


E. Boyer's Rebuttal

Boyer attributes much of the confusion in this argument to Rob-

ertson's unclear terminology. Furthermore, he notes that Robertson is

inconsistent in the application of his theory to conditionals in his com-

mentary Word Pictures. In Word Pictures sometimes Robertson notes

that a protasis is assumed true, but in many cases where it is obviously

false, he fails to mention that a first class conditional is used in the


Boyer sought to bring some focus to this debate by examining all

of the conditionals in the New Testament. He used gramcord to search

the New Testament for all the examples of each kind of condition.14

He then sorted first class conditionals into three groups: (1) instances

where the condition was obviously true, (2) instances where the condi-

tion was obviously false, (3) instances where the condition was unde-

termined. According to his classification, 115 of the condition in the

NT are obviously true and 36 are obviously false.15 He considers these


11 Some other grammars which assert claims like Summers' are: F. Blass, A. De-

brunner and R. Funk, A Greek Grammar of the New Testament and other Early Christian

Literature (Chicago: University Press, 1961); H. E. Dana and J. R. Mantey, A Manual

Grammar of the Greek New Testament (Toronto: Macmillan, 1957); Huber L. Drum-

wright, An Introduction to New Testament Greek (Nashville: Broadman, 1980).

12 Ray Summers, Essentials of New Testament Greek (Nashville: Broadman, 1950)


13 Boyer,"First Class Conditionals," GTJ 2.1 (1981) 79-80.

14 Boyer's work is reported in four articles in Grace Theological Journal. In addi-

tion to the one cited above there are: "Second Class Conditions in New Testament

Greek," 3.1 (1982) 81-88; "Third (and Fourth) Class Conditionals," 3.2 (1982) 163-75;

"Other Conditional Elements in New Testament Greek," 4.2 (1983) 173-88.

15 Boyer, "First Class Conditionals," GTJ 2.1 (1981) 76.



36 conditions in the obviously false category to be counterexamples to

those who would translate the ei] p,q with "since."

Boyer's work is exhaustive and convincing. However, there is still

an element of uncertainty in Boyer's analysis because the methodology

by which he separated the conditions into categories of "obviously

true" and "obviously false" is apparently his own intuition. There are

many examples in his obviously false category concerning which it is

not so obvious that they are false. For example:

(7a) If [ei]] you are the Christ, tell us. Luke 22:67

(7b) If [ei]] to others I am not an apostle, yet I am to you. 1 Cor 9:2


In sentence (7a), Jesus was in fact the Christ, though the speakers of

this sentence may not have believed He was. In (7b) there were in fact

others who believed Paul was not an apostle, which makes the protasis

in fact true, even though Paul was in fact an apostle and believed him-

self to be one.




Significant progress has been made in linguistic description in the

past two decades in the area of implications. The work of H. P. Grice16

is foundational in this area. Many unambiguous tests for identifying

and proving the existence of implicatures 17 have been developed. One

of these tests will aid us in this endeavor.18

Grice made a useful distinction between two kinds of implicature:

conventional implicature and conversational implicature. A conven-

tional implicature is one which is associated with the meaning of the

words and the grammar of a sentence, which cannot be canceled by the

context. For example, factive verbs19 have the conventional implicature


16 See n. 5 above.

17 Grice defined the term "implicature" saying, "I wish to introduce as terms of art,

the verb implicate and the related nouns implicature (cf. implying) and implicatum (cf.

what is implied). The point of this maneuver is to avoid having, on each occasion, to

choose between this or that member of the family of verbs for which implicature is to do

general duty" (Grice [1975] 43, 44). Generally speaking, one may think of an implica-

ture as an implication. But Grice introduced this unique term, because terms like "impli-

cation," "presupposition," and "assumption" have been used for a variety of different

and poorly defined uses.

18 Some helpful introductory texts on Gricean implicature are: Stephen C.

Levinson, Pragmatics (Cambridge: University Press, 1983) 97-166; John Lyons, Seman-

tics (Cambridge: University Press, 1977) 592-606; John McCawley, Everything that

Linguists Have Always Wanted to Know About Logic (Chicago: University Press, 1981)


19 Factive verbs are verbs which presuppose the truth of their complements. This

class of verbs was first identified by Paul and Carol Kiparsky in their article "Fact" in

Progress in Linguistics, ed. M. Bierwisch and K. Heidolf (The Hague: Mouton, 1970)



that the proposition in their complement is true. Evaluative verbs20

have a conversational implicature that the proposition in their comple-

ment is true. Consider the following sentences with the factive verb

"regret" and the evaluative verb "criticize."


(8a) I regretted that John told a lie.

(8b) I criticized John for telling a lie.


The complement's proposition in both cases is the same: "John told a

lie." But what about the implicatures? Does a person who utters (8a) or

(8b) implicate that John told a lie? It may seem that both sentences do,

but on closer inspection we find that they are different with respect to


A common test for implicature is to place the utterance in a con-

text which attempts to cancel the implicature. If a sentence with a con-

ventional implicature is placed in a context which attempts to cancel

the implicature, a pragmatically ill-formed sentence results. If a sen-

tence with a conversational implicature is placed in a context which

attempts to cancel the implicature, the implicature is canceled and the

sentence remains well formed. For example the sentences in (8) are put

in such contexts in (9) below.


(9a) #I regretted that John told a lie, but I shouldn't have regretted it

because it was Joe who lied.

(9b) I criticized John for telling a lie, but I shouldn't have criticized

him because it was Joe who lied.


I use a pound symbol (#) to the left of a sentence to indicate that the sen-

tence is pragmatically ill-formed. Since (9a) is ill-formed, this proves

that the sentence (8a) has a conventional implicature that John told a lie.

In sentence (9b) the implicature that John told a lie is canceled by the


143-73. Some examples of factive verbs in English which take object clause comple-

ments introduced by that are: regret, resent, deplore, be odd, be glad. Some examples of

factive verbs in Greek which take object clause complements introduced by o!ti are: qau-

ma<zw, lanqa<nw, xai<rw, lupe<omai, metame<lomai. See L. W. Ledgerwood, "Syntactic Insu-

lation of Factive Clauses," in The Journal of the Linguistic Association of the Southwest

5.2 (1982) 105, 112.

20 Evaluative verbs are verbs like criticize, accuse, praise, congratulate. Filmore first

identified this class of verbs in C. Filmore, “An Exercise in Semantic Description," in

Studies in Linguistic Semantics, ed. C. J. Filmore and D. T. Langendoen (New York: Holt,

1972) 273-89. Karttunen and Peters showed that the implicature associated with them was

not conventional but conversational. Lauri Karttunen and Stanley Peters, "Conventional

Implicature," in Syntax and Semantics 9, Presupposition (New York: Academic, 1979).



context without resulting in a pragmatically ill-formed sentence. There-

fore the implicature in (8b) was a conversational implicature.21

English causal sentences have a conventional implicature that the

proposition in their protasis is true but English conditionals do not.

Sentences (10) below illustrate this. Sentence (10a) implicates conven-

tionally that the moon is full, but sentence (10b) does not.


(10a) Since the moon is full, it is opposite the sun.

(10b) If the moon is full, it is opposite the sun.


To speakers of English this seems intuitively obvious. However, this

claim may be moved beyond the realm of intuition by placing both

sentences in a context that attempts to cancel the implicature as shown

in sentences (11) below.


(11 a) #Since the moon is full, it is opposite the sun; but the moon is

not full today.

(11b) If the moon is full, it is opposite the sun; but the moon is not

full today.


This suggests a way to formulate a test of Summers' claim that ei] p,q is

best translated with English "since p,q." Summers' claim entails ei] p,q


21 By using Gricean terminology in this paper I do not mean to imply that Grice has

said the last word on implicature. There have been challenges to Grice's methodology.

Most recently several books and papers have appeared proposing relevance theory

as superior to the Gricean framework. Relevance theory and discussions of the problems

with Grice's theory are contained in: Dianne Blakemore, "The Organization of Dis-

course," in Linguistics, The Cambridge Survey Vol. 4, ed. Frederick J. Newmeyer (Cam-

bridge: University Press, 1988); Dianne Blakemore, Semantic Constraints on Relevance

(Oxford: Blackwells, 1987); Ruth Kempson, "Grammar and Conversational Principles,"

in Linguistics, The Cambridge Survey Vol. 1, ed. Frederick J. Newmeyer (Cambridge:

University Press, 1988); D. Sperber and D. Wilson, Relevance, Communication and

Cognition (Oxford, Blackwells, 1986).

Two comments are offered in defense of applying Gricean terminology in this pa-

per. First, most of the challenges to Grice's work have come in the area of what he called

conversational implicatures (for example, Jerrold M. Sadock, "On Testing for Conversa-

tional Implicature," in Syntax and Semantics 9, Pragmatics, ed. P. Cole [New York: Ac-

ademic, 1977]). The notion of conversational implicature is not used in this paper;

conventional implicatures are. (For more on conventional implicature see the following

papers by Lauri Karttunen and Stanley Peters: "Requiem for Presupposition," in Papers

from the Third Annual Meeting of the Berkeley Linguistic Society, 360-71; "Conven-

tional Implicature," in Syntax and Semantics 11, Presupposition (New York: Academic,

1979); "Presuppositions of Compound Sentences," in Linguistic Inquiry, vol. 4 (1973)

169-93. Secondly, the goal of this paper is to show that by making use of a methodology

like that of Grice, one can formulate clear and testable hypotheses which facilitate com-

munication and advance research in applied areas such as this. These arguments could be

reformulated in terms of relevance theory without changing the result.



having a conventional implicature that the proposition p is true. Sum-

mers' claim can be formulated in a hypothesis based on this entail-



(12) Summers' hypothesis: Sentences of the form ei] p,q have the

conventional implicature that p is true.


Formulating his hypothesis in this manner yields one that is very test-

able. If indeed ei] p,q does have a conventional implicature that the

proposition p is true, then it will not occur in contexts which cancel


In an investigation of Koine Greek, it is not possible to record

speech of native speakers nor to quiz them concerning their intuitions

about their language. So, a disciplined methodology is needed for test-

ing hypotheses from texts. David Lightfoot says in his Principles of

Diachronic Syntax,22 "One can never demonstrate the truth of a the-

ory, only its falsity. Thus progress in scientific endeavors can be

viewed as the successive elimination of theories shown by empirical

investigation to be false." I take this somewhat Popperian view of sci-

entific progress to be axiomatic. Thus the historical grammarian's goal

is to formulate hypotheses that are well enough defined that they can

be proven false. No hypotheses will ever be proven true in an inductive

endeavor such as this; they will only be supported by arguments from

silence. The confidence that may be placed in a hypothesis will be a

function of how "silent" the text is; that is, of how many possibilities

were examined in which the hypothesis could have been proven false

and was not.

Large volumes of Greek texts must be searched to find whether ei]

p,q occurs in contexts which cancel the implicature. If ei] p,q is not

found in such contexts, then this will be an argument from silence that

it contains a conventional implicature. This is a weak argument. But if

ei] p,q is ever found in a context in which the implicature is canceled,

then it will be proven that the ei] p,q does not have a conventional

implicature that p is true.

A systematic way of searching large amounts of text to look for

examples like this is to imagine discourse forms which always cancel

the proposition in the protasis. Sometimes this process can be made

regular enough that a computer may be used to do some of the search-

ing for such occurrences. For example, two conditionals linked by an

adversative or disjunctive with the second protasis negated is such a



22 David Lightfoot, Principles of Diachronic Syntax (Cambridge: University Press,

1974) 74f.



(13) if P then q but if not p then r


Another construction which cancels the proposition in the protasis is a

modus tolens argument which has the form:


(14) if p then q, but not q, therefore not p


                                    V. TESTING THE HYPOTHESIS

The first two books of Arrian's Discourses of Epictetus,23 the

Cynic Epistles24 and the New Testament, all dating from around the

first century A.D., have been searched for examples in which a condi-

tional of the form ei] p,q occurs in a context in which the proposition p

is negated. Such examples are abundant. Following are some of them.25


A. Examples of the Form ei] p,q but ei] not p, r


(15a) ei] ga>r mh> ei]si>n qeoi<, pw?j e]sti te<loj e!pesqai qeoi?j; ei] d ] ei]si>n

me<n, mhdeno>j d ] e]pimelou<menoi, kai> ou@twj pw?j u[gie>j e@stai;

For if [ei]] there are not gods, how is it an end to serve gods?

But if [ei]] there are and they don't care, how will this be sound?

Epictetus 1.12.4

(15b) Ei] me>n ou#n a]dikw? kai> a@cion qana<tou pe<praxa< ti, ou] paraitou?-

mai to> a]poqanei?n, ei] de> ou]de<n e]stin . . .

If [ei]] I am a wrongdoer, and have committed anything worthy

of death, I do not refuse to die; but if [ei]] none of those things

are true . . .

(Acts 25: 11)

Note that in both of these cases, translation with "since" is not possible

because the conventional implicature that "since" generates is canceled.


(16a) #Since there are not gods. . . , but since there are . . .

(16b) #Since I am a wrongdoer. . . , but since none of these things are

true. . .


23 Epictetus in Epictetus, the Discourses as Reported by Arian, T. E. Page et a1.,

eds. (Cambridge: Harvard, 1967). Also the machine readable text of Epictetus' Dis-

courses encoded in the Thesaurus Linguae Graeca database at the University of Califor-

nia at Irvine was used.

24 Abraham J. Malherbe, The Cynic Epistles (Missoula, MT: Scholars, 1977).

25 0ther examples not listed here are: Epictetus 1.12.4, 1.29.7, II.1.17, II.2.24,

II.4.4, II.5.25, II.10.13, II.15.6; Ma1herbe, The Cynic Epistles, Crates 30, p. 80, 1. 6; 35,

p. 88, 1. 19; Diogenes 5, p. 96, .1. 1; 24, p. 116, 1. 10. In the NT see Matt 12:27-28,

26:39-40; Luke 11:19-20; John 10:37; 18:23; 1 Cor 9:17; James 2:2-9.



B. An Example of a Modus Tolens Argument


(17) Ei] de> a]na<stasij vekrw?n ou]k e@stin, ou]de> Xristo>j e]ge<ger-

tai. . . Nuni> de> Xristo>j e]gh<gertai e]k nekrw?n. . .

But [ei]] if there is no resurrection of the dead, not even Christ has

been raised. . . . But now Christ has been raised from the dead. . . .

1 Cor 15:13, 20

Note that the argument makes no sense if ei] is translated with "since"

because Paul intends for the Corinthians to deduce that there is a resur-

rection of the dead.


(18) #Since there is no resurrection of the dead, not even Christ has

been raised. . . But now Christ has been raised from the dead.


Examples such as these disprove the Summers hypothesis as formu-

lated above. That is, they prove that conditionals of the form ei] p,q do

not have the conventional implicature that the proposition p is true.

Therefore the English causal "since p,q" is not a good translation for ei]

p,q across the board.


C. Examples of ei] p,q in which p Is True

Nevertheless; sometimes there are cases in which conditionals of

the form ei] p,q can be translated with English "since." Following are

two such examples.26


(19a) ei] e]me> e]di<wcan, kai> u[ma?j diw<cousin.

            If [ei]] they persecuted me, they will persecute you also.

                                                                                                            John 15:20

(19b) Ei] de> kalo>j h#n Pla<twn kai> i]sxuro<j, e@dei ka]me> kaqh<menon e]k-

            ponei?n, i!na kalo>j ge<nwmai h} i!na i]sxro<j, w[j tou?to a]nagkai?on

            pro>j filosofi<an, e]pei< tij filo<sofoj a!ma kai> kalo>j h#n kai> filo>-


            Now if [Ei] Plato was handsome and strong, is it necessary for me

            to sit down and strive to become handsome or strong on the

            assumption that this is necessary for philosophy, since [e]pei<] some

            philosopher was at the same time both handsome and strong?

                                                                                                Epictetus 1.8.13


            26 For other examples in which the proposition in the protasis is true and translation

with "since" is possible, see Malherbe, Cynic Epistles, Crates 30, p. 80 1. 8 and Sopho-

cles Fr. 877N (sentence 28 in this paper); Rom 3:29, 30; 11:21.



Translations with "since p, q" are appropriate for these examples as

shown in sentences (20) below.


(20a) Since they persecuted me, they will persecute you also.

(20b) Since Plato was handsome and strong. . .


To the people who originally heard these utterances, and to those who

are acquainted with Jesus' life and Plato's physique, it is generally

known that Jesus was in fact persecuted and that Plato was in fact hand-

some and strong. That is, it is known from other sources that the prop-

osition in the protasis is true. For this reason, translation with "since

p, q" is acceptable, because the implicature generated by "since" does

not conflict with the known facts of the case. In all the cases in the cor-

pus under investigation where "since p,q" may be used to translate ei]

p, q, it is clear from the context that p is true. The truth of p comes from

the context, not from a supposed implicature associated with ei] p, q.

            But the fact that ei] p, q sometimes can and sometimes cannot be

translated with "since p,q" indicates that there is something else going

on in these conditionals other than conventional implicature and for

this reason it is not appropriate to recommend a translation of ei] p, q as

"since p, q."

            Why does ei] p, q have this on again-off again implicature? Why

don't such implicatures occur with e]a<n p, q? These are not the subject of

this paper. Answers to these questions have been proposed elsewhere.27

What this paper claims to offer is unambiguous proof that the first class

conditional does not conventionally implicate the truth of its protasis.

            The following quotes from ancient Greek grammarians show that

they agree with this conclusion.




            Passages from four ancient Greek grammarians are presented

below. The grammarians are:28


            Dionysius Thrax                    (1st century B.C.)

            Apollonius Dyscolus            (2nd century A.D.)

            Stephanos                               (Byzantine period)

            Heliodorus                             (Byzantine period)


            27 Unpublished proposal presented by L. W. Ledgerwood at the 1989 meeting of

the Linguistic Association of the Southwest in San Antonio, TX, and the 1990 AAR/SBL

meeting in New Orleans, LA.

            28 The text used is found in G. Uhlig, Grammatici Graeci I I/II, Dionysii Thracis

and Grammatici Graeci, II II/III, Apollonii Dyscoli (Hildesheim: Georg Olms, 1878-

1910, reprinted 1965). The English translations are original.



Dionysius is the father of western grammatical tradition; however, his

work is quite short. Stephanos and Heliodorus wrote commentaries on

Dionysius' grammar which flesh out his arguments with example sen-

tences. Apollonius wrote the most voluminous and original grammar

of the four. We will examine Dionysius and his commentators first,

then Apollonius.


A. Dionysius Thrax

Dionysius classed conditional and causal particles (ei] "if," e]pei<

"since," e]a<n "if") along with conjunctions (kai< "and," h@ "or," de<

"but," etc.). He has only one short passage on conjunctions. The por-

tion of this dealing with conditionals and causals is listed below.

If Dionysius' account seems unclear, his commentators adequately

explain his meaning.


(21)  Conditional particles are those which do not assert existence,

but they signify consequence. They are: ei], ei@per, ei]dh<, ei]dh<per.

Causal connective particles are those which assert order

along with existence. They are e]pei<, e]pei<per, e]peidh<, e]peidh<per.

Expletive conjunctions are those which are used on

account of meter or adornment. They are: dh<, r[a<, nu<, pou?, toi<,

qh<n, a@r, dh?ta, pe<r, pw?, mh<n, a@n, nu?n, ou#n, ke<n, ge<  (20.3.4,8).


Note that Dionysius does not discuss the conditional particle e]a<n. e]a<n

is constructed from ei] plus the modal particle a@n. He mentions the

modal particle a@n under Expletive Conjunctions.29


B. Dionysius' Commentators, Stephanos and Heliodorus


(22) The conditional particles differ from the causal connective par-

ticles as follows: the conditional particles only connect proposi-

tions, they do not affirm the reality. For example, if I say, "If

[ei]] the sun is over the land," it is not clear whether the sun is

over the land. But the causal connective particle, in addition

signifying consequence and connecting to another proposition


29 Dionysius has lumped a lot of different types of particles into his "Expletive

Conjunctions." His statement about them indicates that he considers that they add little

or no meaning to a text. Rather, they are added simply to make meter (i.e., in poetry)

come out right and to add adornment. It seems that he really did not know what to do

with these. Apollonius discusses a theory which said that expletive conjunctions merely

"fill up the empty holes in a text" and takes strong objection to this theory. He says that

each of the expletive conjunctions adds some special meaning such as "transition in

logic" for dh<, "moderation" for ge<, etc. (III.127-29). Unfortunately, he does not tell us

what the special meaning of a@n or e]a<n is.



also affirms the reality, for example, "since [e]pei<] the sun is

over the land, it is day" (Stephanos, in Uhlig 1965 I/III,

p. 284.30).

(23) Of the conjunctions, some assert existence, others assert order

and others both. Coordinating conjunctions [i.e., kai< "and"]

assert existence. For example, if I say, "God and day and justice

exist," everything is affirmed.30 The conditional particles dis-

close order. For example, if I say, "If I am walking I am mov-

ing," the sentence holds consequence, but it is not also affirmed;

for I can say this while I am sitting. But if I turn it around, the

truth is destroyed. For example, "Whenever [o[tan] I am mov-

ing, I am walking" is not true, for it is possible for me, while

sitting, to move something. The causal connective particles

have both the reality of the coordinating conjunctions and the

order of the conditional particles; for "Since [e]pei<] I am walk-

ing, I am moving" is both affirmed and has order. In the same

way, it being turned around is no longer true (Stephanos, in

Uhlig 1965 !/III, p. 286.5).

(24) The difference between the coordinating conjunction and the

conditional particle is this: the coordinating conjunctions have

the force of reality but they are unordered with respect to the

flow of speech. For example, "I am walking and I am thinking,"

and the reverse, "I am thinking and I am walking.”31 But the

conditional particles do not affirm the force of reality; rather

they affirm the consequence of the expression and they preserve

the order. For example, "If [ei]] I shall walk, I shall be moving."

But I may not say, "If [ei]] I shall be moving, I shall be walk-

ing," for it is false (Heliodorus in Uhlig 1965 I/III 105.10).


(25) The conditional conjunction stands in place of e]a<n, in "If [ei]]

there is light, it is day." . . . It also, stands in place of the causal

connective particle e]pei< in, "Since [ei]] you have done terrible

things, you must suffer terrible things."

One must see that the causal connective particles have this

much more than the conditional particles, they not only have


30 By "Everything is affirmed," Stephanos means that a person who utters the

phrase, "God and day and justice exist," is asserting that God exists, it is presently day

and justice is presently occurring. On the contrary, a person saying, "If I am walking, I

am moving," does not assert that he is presently walking or moving.

31 Heliodorus is saying that with the conjunction kai< ("and") it does not matter

what order the propositions come in. Thus, "I am walking and I am thinking" means the

same as "I am thinking and I am walking." However, in the case of the conjunction ei],

changing the order changes the meaning.



consequence and order, but also they indicate the existence of

reality. For I may say, "Since [e]pei<] it is day, there is light,"

. . . and there is not uncertainty as with the conditional particle

(Heliodorus in Uhlig 1965 lIIII, pp. 439.4-11).


Dionysius and his commentators address specifically the questions of

implicata of Greek conditionals. They here are interested in two prop-

erties of the so-called conjunctions. These are: (1) existence and (2)

what they refer to as consequence and order. The following definitions

of these terms are proposed for these passages.


Existence:      Uttering the phrase implies that the propositions joined

by the conjunction are true in reality.

Consequence: There is a logical or causal relationship between the

phrases joined by the conjunction.

Order:             The linear order of the propositions in speech flow is

significant. The order cannot be reversed.


The Greek grammarians quoted above tell us that their so-called con-

junctions have the following properties:


Conjunction                                       Properties

Coordinating Conj. (kai<, and)          existence

Conditional Conj. (ei], if)                  consequence and order

Causal Conj. (e]pei<, since)                existence, consequence and order


The examples they give leave no doubt as to their conclusion. Stepha-

nos gives the sentences:


(26a) If [ei]] the sun is over the land, it is day.

(26b) Since [e]pei<] the sun is over the land, it is day.


He says that (26a) does not imply that the sun is over the land while

(26b) does.

Of particular interest is Heliodorus statement in quote (25) above.

He says that ei] may be used in place of e]a<n and gives an example

repeated as (27) below and that ei] may be used in place of e]pei< and

gives an example repeated in (28) below.


(27) If [ei]] there is light, there is day.

(28) Since [ei]] you have done terrible things, you must suffer terrible

things (Soph Fr 877 N).

Sentence (27) is a statement of general truth. It does not assert that it is

necessarily day or not, it just asserts the entailment that whenever it is



light, it is day. It seems that Heliodorus considers it more natural to

make such a generalized statement in Greek with e]a<n p,q (what Good-

win called the present general condition: e]a<n and the present subjunc-

tive in the protasis and a present indicative in the apodosis). But he

gives sentence (27) as an example of a case in which ei] p,q means the

same as the present general condition e]a<n p,q. Sentence (28) is an

example of ei] p,q being used in a context in which it is clear that p is

true. In this example, he says that ei] p, q means about the same as e]pei<

p, q.

Yet, he cannot mean that ei] and e]pei< are equivalent in meaning,

for he says clearly in other passages that e]pei< p,q implies that the prop-

osition p is true in reality while ei] p,q does not. He just observes, as

has been observed above (pp. 110-11), that ei] can sometimes be used

where the causal could also be used.


C. Apollonius Dyscolus (from Syntax, Book III)32

In the following passage Apollonius is discussing the origin of the

names of the moods. Previous to this passage, he has dealt with the

indicative and optative and shown that these names ("Indicative" and

"Optative") come from the meaning of the mood. But in the case of the

subjunctive, the term subjunctive does not refer to a quality of its

meaning, but to its syntax. That is, it occurs primarily in clauses that

are subjected (i.e., subordinated) to another clause and it got its name

from this property. Specifically here he is refuting the theory that the

subjunctive should be called the dubative.

This naming theory is relevant to the discussion at hand in that

Apollonius asserts that conditionals with ei] and e]a<n have about the

same degree of doubt. Furthermore, he is the only grammarian to say

anything substantive about the conditional e]a<n p,q.


(29) Next it is necessary to speak about the subjunctive mood which

some call dubative because of its meaning, just as also the pre-

viously mentioned moods have received their names. For it is

clear that "If [e]a<n] I ever write" and the like express a doubt

concerning a future matter.

But perhaps someone will object that these [i.e., the

moods] are not the source of the sense of doubt, but the accom-

panying conjunction is the source of doubt. Now, if it is reason-


32 Two very helpful works on Apollonius have recently appeared. They are: David

L. Blank, Ancient Philosophy and Grammar, The Syntax of Apollonius Dyscolus (Chico,

CA: Scholars, 1982); and a translation of Apollonius' extant books on syntax in F. W.

Householder, The Syntax of Apollonius Dyscolus (Amsterdam: John Benjamins, 1981).

Another helpful work discussing Apollonius' model of AN is R. Camerer, "Die Behand-

lung der Parikel AN in den Schristen des Apollonius Dyskolos," in Hermes 93 (1965)




able to name verb forms after the meaning of their conjunctions,

then nothing prevents us from changing the names of the other

moods also when they receive this meaning from their conjunc-

tions. . . . Roughly speakIng, "If [ei]] you are talking you are

moving" falls under the same doubt as "If [e]a<n] you walk you

will move," but "If [ei]] you are walking" is not called dubative



Apollonius' point is that an indicative introduced by ei] is just as duba-

tive as a subjunctive introduced by e]a<n. Therefore the source of the

dubative meaning is not the mood (subjunctive or indicative) but the

conjunction (e]a<n or ei]) is the source. This is important for evaluating

Robertson's model of Greek conditionals, because Robertson bases his

classification of conditionals primarily on the distinctions between the

moods accompanying the conjunction.

In the following passages, Apollonius gives us an interesting

statement concerning the tenses which are grammatical with e]a<n p,q.


(30) The above-mentioned mood [the subjunctive] with the conjunc-

tion e]a<n and its equivalents33 is accompanied by the future or

present tense. For example, "If [e]a<n] I study Dion will come,"

and "If [e]a<n] I ever read, Tryphon comes." For a past tense is

ungrammatical (3.131).

(31) It is necessary also to examine the syntax of the conjunctions, to

determine why they refuse the endings of the past tense. For the

syntax of "If [e]a<n] I was saying" is not acceptable, or "If [e]a<n]

I have trusted”34 and the like. . . It is evident that the cause of

such ungrammaticality is the conflict of the past tense with the

meaning of the conjunction. For they present a doubt about com-

ing matters and also about those matters to be completed. . . .


33 One would like very much to know what Apollonius meant by "Its equivalents"

(i]sodunamou<ntwn). He probably means the terms e]a<n, e]a<nper ("if indeed") and the like,

since Dionysius classes ei] with ei]per, etc. However, would Apollonius include o!tan

("whenever") in this class? Both e]a<n and o!tan are constructed by adding a@n to another

particle. e]a<n comes from ei] + a@n; o!tan comes from o!te + a@n. Both e]a<n and o!tan take the

subjunctive. o!tan is frequently interchangeable with e]a<n. (For example, note that Steph-

anos uses o!tan for e]a<n [quote (23) above].) In spite of these similarities, there are ex-

amples of o!tan with the indicative, used to express an iterative sense, which cannot be

written off as grammatical quirks. See for example: Polybius IV .32.5, Ignatius Eph 8:1,

Exod 17:11 (LXX), Num 11:9 (LXX), 1 Sam 17:34 (LXX), Ps 119:7 (LXX), Mark 3:11,

11:19. Apollonius does not tell us what he thinks about such uses of  o!tan.

34 "If I was saying" (e]a<n e@legon) is e]a<n plus an imperfect indicative verb. "If I

have trusted" (e]a>n pe<poiqa) is e]a<n plus a perfect indicative verb. One would have to use

the conjunction ei] instead of e]a<n to make these sentences grammatical in Greek. For e]a<n

to be used grammatically, it must be used with a subjunctive, which is atemporal.



Because how can that which has happened be brought together

with that which is coming? (3.137-138).


In the quote (30), Apollonius is saying that in e]a<n p,q, the proposition

q cannot be in the past tense of the indicative. In the quote (31), he is

saying that the proposition p may not be in the past tense of the indic-

ative. This second statement seems a bit odd, because e]a<n is not sup-

posed to have any form of the indicative in the protasis proposition p,

no matter what tense.35

The import of this passage for this investigation is as follows.

Apollonius said earlier that ei] p,q and e]a<n p,q have about the same

degree of doubt, but in this passage he seems to consider e]a<n p,q more

dubative in some way than ei] p,q, though he does not explicitly say so.

For he says that there is a conflict between the meaning of the past

tense and the meaning of the conjunction e]a<n. But he and we both

know that the conditional ei] can be constructed with past tense indica-

tives in either the protasis or apodosis. So, either e]a<n seems more

dubative to him in some way than ei], or he had not thought out thor-

oughly the consequences of his statement.



It has been proven, and the ancient Greek grammarians agree, that

a conditional of the form ei] p,q does not have a conventional implica-

ture that the proposition p is true.

Conditionals of the form ei] p,q should not be translated across the

board with the English causal "since p,q." Such a translation is appro-

priate in some cases, but is not in the majority. In the few cases that ei]

p,q can be translated with "since p,q," the English "if p,q" will also be

appropriate because, in these cases the context carries the implication

that the proposition p is true. The use of English "since p,q" in these

cases only adds redundancy.

Robertson's assertions are unclear. The way that he is interpreted

by some today yields an erroneous analysis of conditionals. Robertson

claims to be in the tradition of Gildersleeve; however, he went farther

than Gildersleeve went. Gildersleeve never said that ei] p,q implies the


35 It is noted here that in Apollonius' day, significant diachronic changes in the syn-

tax of conditionals were occurring. The conditional ei] was dying out and the conditional

e]a<n was taking over. Not long after Apollonius' day, e]a<n came to be used with the indic-

ative (see A. N. Jannaris, An Historical Greek Grammar [Hildesheim: George DIms,

1968] §§ 1772 and 1987). There are some examples of e]a<n used with the indicative in

the NT (1 Thess 3:8, 1 John 5:15, Luke 19:40, Acts 8:31). These may be a reflection of

this change. However, these grammarians were writing about the classical forms of their

language, the language as they felt it should be. At any rate, diachronic factors are ne-

glected in this paper for simplicity.



proposition p is true; some read Robertson as saying that it does. The

ancient Greek grammarians disagree with Robertson and those in his

tradition, but they do not disagree with either Goodwin's or Gilder-

sleeve's claims. Goodwin and Gildersleeve were writing more about

aspectual and temporal interpretations than about implications con-

cerning truth.

Bible students should not be taught that ei] p,q means "since p,q."

Exegetes should be honest in their hermeneutics and should refrain

from stating or implying in an exegesis of a passage that the Greek

conditional ei] p,q itself implies that p is true. Nor should an exegete

state that ei] p,q does not imply doubt like English "if p,q" can and that

it would be better translated with "since p,q." In those cases where one

wishes to make a point that the proposition p is not being called into

question, it should be demonstrated that the context implies that the

proposition p is true or that the participants in the communication

knew that p was true in fact.





This material is cited with gracious permission from:

            Grace Theological Seminary

            200 Seminary Dr.

            Winona Lake,  IN   46590

Please report any errors to Ted Hildebrandt at: