STOP Early Journal Content on JSTOR, Free to Anyone in the World This article is one of nearly 500,000 scholarly works digitized and made freely available to everyone in the world by JSTOR. Known as the Early Journal Content, this set of works include research articles, news, letters, and other writings published in more than 200 of the oldest leading academic journals. The works date from the mid-seventeenth to the early twentieth centuries. We encourage people to read and share the Early Journal Content openly and to tell others that this resource exists. People may post this content online or redistribute in any way for non-commercial purposes. Read more about Early Journal Content at http://about.jstor.org/participate-jstor/individuals/early- journal-content . JSTOR is a digital library of academic journals, books, and primary source objects. JSTOR helps people discover, use, and build upon a wide range of content through a powerful research and teaching platform, and preserves this content for future generations. JSTOR is part of ITHAKA, a not-for-profit organization that also includes Ithaka S+R and Portico. For more information about JSTOR, please contact support@jstor.org. Classical Philology Vol. Ill 'January, igo8 No. i THE SEMANTICS OF MODAL CONSTRUCTIONS By Tbnney Frank II. NULLA CAUSAST QUIN DET We found in the preceding paper (Class. Phil. II, pp. 163 ff.) that some of the relative clauses of the usual consecutive form were prone to become expressions of "obligation or propriety." Such was the function of the group nil est quod det when quod is adverbial. The more explicit expressions, however, are wont to convey the notion of reason or motive more unmistakably. They do so by using connectives like cur, quin, quam ob rem, that teem with the idea of cause, and by use of the word causa or the like in the antecedent. In the present paper we propose to sketch the use of such idioms in Latin, discover to what extent they are connected with similar modal usages in closely related languages, and indicate their position with reference to the simpler idioms of Latin. I. The idea of obligation or propriety is not often strong in clauses with quod. It does not even permeate all of those in which quod 1 is an accusative of respect and almost equivalent to cur. Six of the Plautine instances have a negative main clause. They are as follows: nil est iam quod tu mihi suscenseas, Merc. 317; bono animo's .... nil est quod timeas, Amph. 1132; nil erit quod deorum ullum accusites, Most. 712; nil est quod metuas, 1 These lists exclude characterizing clauses of fact, however closely allied in form they may be. Of. Capt. 741 : in morte nil est quod metuam mali; Ps. 575, etc. [Classical Philoloot III, January, 1908] 1 2 Tenney Frank Ps. 1066 ; nil est de signo quo(riy vereare, Trin. 808 ; isti quod suscenseam ipsi nil est, As. 146. It is well to note that the verb is only once in the first person, that the governing clause pre- cedes in all but the last instance, and that the verb in every instance is one of mental activity. The occurrences after an affirmative are: bonamstf quod ha- beas gratiam mihi, Bud. 516; tibi quidem quod rideas magis est quam ut lamentere, Merc. 502 ; si animum vicisti .... est quod gaudeas, Trin. 310; si factumst . ... est quod mihi suscenseas, Trin. 1166. These are also in the second person and deal with verbs of mental activity. Two instances that show a slight difference are: quid est quod caveam, 1 Rud. 833, which repeats the thought of a virtual command in the context, and etsi nil scio quod gaudeam, Capt. 842, which slightly varies the phrase- ology of the ordinary negative type. These instances in Plautus represent pretty fairly the regular usages of Latin. As for Cicero, nihil est quod, when quod is adverbial, occurs some twenty times in the part of Cicero covered by Merguet, and invariably with a tone of obligation or pro- priety. However, the first person, usually a plural, which includes speaker and listener, is almost as frequent as the second. The third also occurs. Cf. Leg. Agr. ii. 88, timeremus; Sull. 9, ad- mirerej Sest. 104, dissentiat, etc. The verb is usually, though not always, one of mental activity. The formula quid est quod is very frequent in Cicero, but does not yield the meaning here discussed, except in a few rhetorical questions of the first person, like quid est quod plura dicamus? Clu. 59. This use of quid est quod chiefly with the first person and in the same phrases as employed by quid ( = cur) would indicate that the group had grown to be synonymous with quid and was perhaps pronounced 1 The formula quid est quod seldom introduces an expression of obligation. It usually takes a characterizing clause of fact. Of. quid est quod metuas? idem istuc quod tu, Pers. 239 ; also Poen. 884, 867 ; Bacck.92; Capt. 618. The retort istuc quod tu shows that quod is a pure relative pronoun. It is often used as an accusative of respect = cur ; but its subjunctive is still to be considered an instance of the charac- terizing type. Of. quid est quod non metuas? quia, Ps. 1087, also Bacch. 1156, pudeat; Rud. 629, 638, tumultues; Cur. 166, voces. In fact, the verb may still be found in the indicative in early Latin, as in other undeveloped characterizing clauses : quid est quod laetus es, Eun. 559. Cf. Andr. 447; cf. oportet, in est quod .... oportet, Hec. 273. Semantics of Modal Oonsteuctions 3 as one word. (The interrogative examples in the second and third persons and most of those in the first contain relative char- acterizing clauses of fact, or substantive clauses of motive like quid est quod accuses? Verr. ii. 49, "why do you accuse?" We are therefore not concerned with them.) A second distinct type consists of sentences introduced by nulla causast quin and the like. The instances in Plautus are as follows: (a) nulla causast quin me verberes, Cas. 1003; condones, Bud. 1070; feras, Bud. 1397. (b) hau causast .... quin facias, Most. 435 (cf. redeat, Ter. Hec. 588). (c) quid causaest quin te sauciem? Bud. 758 (proficiscar, Ter. Andr. 600). (d) numquae causast quin faciamus hodie? Aul. 261; des, Capt. 353. (e) numquid causaest quin uxorem eras domum ducam? Trin. 1188; condam, Ps. 533. (/) nullam causam dico quin siet, Capt. 626; insimules, Amph. Fr. 13 (ferat, Ter. Phor. 272). (gr) numquid causam dicis quin te hoc multem matrimonio? Amph. 852. (h) haud causificor quin earn habeam, Aul. 756. In these clauses the antecedent is always a negative or a negating question, as is to be expected of gmn-clauses. 1 In the questioning form the first person predominates, otherwise the second person. As for meaning, they produce a tone of pro- priety or reasonableness in a very narrow sense of those terms. They are not true synonyms of oportet or debet. The question, or exclamation rather, is usually not one of moral obligation in- volved by duty, as is usually the case of oportet and debet. It is rather a question of reasonableness as regards punishment for a crime, or permission to do a desired act. So in Cas. 1003, Am. Fr. 13, Most. 435, Bud. 1070, Capt. 353, 626, the sense is: "There is no reason why you should not (be allowed to) punish me for that." In Am. 852, Ps. 533, Bud. 758, Aul. 756, it is: "Why should I not (be allowed to) punish you for that?" And in Aul. 261, Trin. 1188, Bud. 1397: "Is there any reason why we should not (be permitted to) have this?" In Cicero this type is relatively rarer, giving way to expres- sions with cur. Some of the representative examples with quin 1 Of. Morris Aims and Methods in Syntax, p. 160, for an explanation that ia at least partially true. 4 Tenney Frank are: causae nihil esset quin secus indicaret, Cic. Quinct. 32; promiserit, ibid. 57; numquid est causae quin vicerim? Quinct. 84; also Q. Rose. 41; Leg. Agr. ii. 40; ibid. 74. A few instances with cur, quam ob rem, and qui are of the same nature: non iusta causast quo^r) curratur, ubi bibas? "Is there not fair cause for speed if there be a bottle in sight?" Poen. 533; nequest quor studeam has nuptias mutarier, St. 52; haud promeruit quam ob rem vitio vorteres, Am. 1142. This is a substantive clause. The rest are virtually characterizing clauses : nil q. o. r. id faciam (eos) meruisse arbitror, St. 82; quid .... (com}merui mali q. o. b. ita faceres? Aul. 736; quid merear q. o. r. mentiar? Most. 987; quid habetis qui mage immortales vos credam esse quam ego ? Poen. 276 ; quid de te merui qua me causa perderes? Men. 490. In this list 1 again the antecedent is negative or a question with a negative implication. The instances with quam ob rem are all connected with mereo or its compounds. These idioms change somewhat by the time of Cicero. The expressions with cur, particularly causa cur, grow in prominence. Some representative instances follow: causa cur mentiretur non erat. Quinct. 18; postulares, ibid. 37; facerent, Tull. 55; nulla causa est cur .... velis, Sex. Rose. 146; Clu. 64; quid est causae cur non pertimescat? Flacc. 5; cogerer, Phil. i. 11; quae causa est cur tollamus! Lael. 48; properaret, Milo 49. cur is also found without causa: nihil est cur alius alio melior existimetur, Font. 22; cf. Milo 49; Nat. deor. iii. 23; Tusc. i. 109. quam ob rem holds its own with such expressions, though it no longer shows its former fondness for mereor. Some examples are: quae fuit causa q. o. r. venenum dare vellet, Gael. 56; nihil erit q. o. r. requiratur, Font. 25; Verr. iii. 186; quid est q. o. r. quisquam putet, Clu. 132; cf. videatur, Sull. 89, also 81. i These particles are of course used freely in other dependent clauses both indica- tive and subjunctive. Thus cur frequently introduces an indirect question of fact (As. 730, scire cur ludatis, and Capt. 1007) ; or serves as a relative pronoun in charac- terizing clauses of fact; cf. Verr. i. 80; iv. 10, etc. quam ob rem may be a relative pronoun in a determinative clause (Pers. 780; Bud. 430), or in a characterizing clause (Most. 415, nequid potiatur, q. o. r. pigeat), or may introduce indirect questions (scio q. o. r. credas, As. 842; cf. Am. 19). The subjunctive after qui usually has a tone of potentiality, as is to be expected from its meaning ; cf . Most. 992 and Men. 304, and the preceding study. Semantics of Modal Constructions 5 These formulae usually occur in Cicero in the orator's attempt to refute an opponent by showing the unreasonableness of his contention. "What motive could Caelius have for wishing to poison that woman?" Cael. 56, is a typical instance. The opor- tet to which such expressions are nearest is one of logical necessity, not of moral obligation. II. In the present problem it will prove convenient to change to some extent the order of procedure employed in the former paper. I shall discuss the possible congeners as found in the cognate languages before attempting to find the grammatical affinities of our construction within its own language. Mr. Elmer (A. J. P. XV, pp. 313 ff. and Cornell Studies VI, pp. 213 ff. ) has hitherto provided the fullest 1 discussion of the sentences that "convey the should-idea," although he confines his observations in the main to simple sentences. On p. 225 of the latter work he quotes Horn. II. ii. 250, ra> ovk av fiaeriXrjas ava ctt6(i' e^atv ayopevois, as an example of this usage in Greek. The question arises: Is this usage so prevalent in Greek, Sanskrit, Germanic as to justify the assumption of an early and common Indo- European modal construction of this sort to which the Latin should be traced? Let us see. The closest and most prevalent parallel in Greek to a simple Latin construction of the kind is found in such questions as Soph. Ant. 1194, riydp avovp.e0'; ( = cur enim te deleniam verbis), where Wis used adverbially like quid ( = cur). I shall give all the examples that occur in Sophocles for illustration: rl 8' av fie koX KptvoK; Track. 314; ti St)t' av o-kottoIto t« ttjv Uvdo- fiavnv ear lav; 0. B. 964; ri 8' av 0o/3otr' avOpmiros; ibid. 977; tC Srjr' av aXXo tovt' eVevreXXot? en; Ant. 218; cf. Aes. Sept. 704, tiow er' av craivoifiev oXeOpiov fiopov; ( = quid igitur adule- mur?) There are five instances in Sophocles. They are not bonafide questions, but exclamations against the uselessness or 1 1 find the usage recognized earlier in Hale's Synopsis of the Modal Uses of the Finite Verb. He has long grouped it as a separate subdivision after the optative, influenced, I believe, by the nature of the Sanskrit and Greek optative of a similar kind. The Hale-Buck Grammar gives a pretty full list of dependent clauses of the same function, §§ 512, 513. It must be evident, however, that the theory underlying this discussion is very different from that which underlies the classification in the Hale-Buck Grammar. 6 Tenney Frank unreasonableness of some proposal actually made or condition expressed or implied in the preceding words. In order the better to analyze this usage, I would call attention to the other con- structions that occur with Tt. ti is usually a pure interrogative, meaning quid, "what." Its verb in such cases is naturally indi- cative if it is a question of fact. It may also take a subjunctive "question of deliberation." All of these we omit, and pass to the optative usages. There are three: a) ti is an interrogative pronoun = quid f ("what?"): cf. Ant. 646, ti tovS' av «Ws aXXo irkr]v avr<3 wovous v(rai; See also Ant. 39, 552; Track. 630; O. C. 71; Ai. 107, 581; while Phil. 895, and 1393 are nearly "deliberatives." b) Tt is an interrogative adverb = quif quomodo? ("how?"): cf. Ai. 537, re Siji-' av . . . . &e\otiu(re; "How then can I serve thee as the case now stands?" Jebb; also El. 1483, and a list of similar usages with «•&? in the preceding study. c) Tt is an interrogative adverb = cur ("why?"). These are the ones we are studying. I have listed them above. Mark that the mode of the verb is the same throughout (the optative with av}, but its function changes with the changes in meaning of the interrogative. In the first it is simply hypo- thetical, in the second "potential," and in the third a "question of propriety." For the interpreter of the sentences, the nature of Tt is determined by the nature of the situation. In all three cases we are dealing with situations and proposals that are more or less hypothetical. In the second case the proposal is obviously impossible of achievement. Hence we interpret the standing formula with some expression like "How .... could one?" In the third, the proposal is presented as unreasonable. Hence this time we interpret it with: "Why should one?" Is it not evident then that in such questions the notions of "potentiality" and "obligation" have their abode in the different introductory particles and in the rest of the context rather than in the mood of the verb? The mood is the same in all three questions. It is the mood of hypothetical expressions adapted 1 lit may well be, of course, that in this particular idiom the adaptation has become permanent, and that the idiom as it stands in classical Greek retains no trace of the earlier modal function. Semantics of Modal Constbuctions 7 to these particular circumstances. These "questions of obliga- tion" are never adequate synonyms of Bel. They are always exclamatory. In bona fide questions of obligation with ri ( = cur) Bel 1 is necessary. Apart from these questions with ri, one could readily discover various other modal expressions that under certain circumstances convey the idea of obligation or propriety. (For the matter of that, "no" may be equivalent to "yes" under the proper circum- stances: "Since maids in modesty say 'no' to that | which they would have the profferer construe 'aye.'" The fact was known before Shakespeare uttered this platitude, and yet the lexicographer does not consider "yes" a synonym of "no.") Such is the ex- ample quoted by Elmer (av .... ayopevoK, see above, p. 5), and the optatives of Od. xviii. 79, vvv (ih nqr' eii]<; ftovydle /Mjre yevoio have drifted so far from the tone of pure wishes as possibly to fall into such a group, if we tried to make one. However, to go about gathering up such mavericks from the four winds into a new herd and rebranding them would involve questionable ethics, to say the least. Modal expressions of "obligation and propriety" in Greek are, therefore, rare. They bear the appearance of being sporadic occur- rences performing a somewhat abnormal function beyond their regular spheres. At most we may say that a few, like the optative question with tC ( = cur), have been perpetuated by such "freaks" and have continued to perform to a limited extent the duties of Bel. As for Sanskrit there is a widespread and well-established usage of the optative that is not far from this in meaning. I refer to the prescriptive usage; cf. E.V. v. 18, 1: Pratar agnih puropriy6 vecdh staveta. Delbruck (Altind. Synt., pp. 333 ff.) gives a number of instances. This, however, differs funda- mentally from most of the examples that occur in Greek, for it is clearly an outgrowth of the optative. Just as a subjunctive ex- pressing will is prone to assume the burden of issuing commands, so an optative of desire may well have put on dignity and func- !0f. Soph. Phil. 11, 102, 1060; 0. R. 897, 1334; O. C. 1148, 1420; Track. 814; Ai. 393 ; and others. 8 Tenney Feank tioned as a prescriptive. One is almost tempted to read a chapter of ecclesiastical history into this semantic change. The humble Vedic priest perpetually praying employs a great number of op- tatives, the powerful Brahman has come to utter the same formulae in the tone of one whose desire is law. The Germanic dialects show no well-established usage of the optative performing such functions. The nearest approach to it may be found in some questions of the first person of which an illustration may be found in my paper on the "Optative in the Edda" {A. J. P. XXVII, p. 6), hvi of segjak p6rf "Why should I tell you?" (Skm. 4). Whatever notion of "obligation" may be found in such expressions certainly emanates from the expres- sion as a whole and does not lie in the mode of the verb. This survey of the most typical expresssions of obligation and propriety in the cognate languages can lead to but one conclusion. The idioms have nothing in common but function. They differ widely in form. Some have the form of questions of the first person in the optative or subjunctive; some, that of optative or subjunctive wishes and commands. Some (in Greek with the optative + civ) are akin to hypothetical expressions. Grammarians who study the semantics of words dare not recognize relationship on the basis of function 1 alone. In dealing with groups of words a rule certainly no less emphatic must be followed. We therefore dare not assume a common Indo-European usage of the optative or subjunctive for expressions of "obligation" or "propriety," and shall proceed to search for the creating forces of our idiom within the Latin language itself. III. A Koman must have employed a considerable portion of his time in discussing his own and his neighbor's obligations, if we may judge from the number of words and phrases he kept at hand for the expression of such ideas. Even the modal phrases of this nature are more numerous in Latin than in most languages ; for I think we must admit at once with Elmer that such expressions are frequent. I am not so confident, however, that Elmer is right in gathering all of them into one group and deriving the whole 'Of. F. A. Wood Color Names and Their Congeners, p. 7: "Synonymy is of itself on proof whatever of relationship and comparisons that are made on that basis merely are utterly worthless." Semantics op Modal Oonstbuotions 9 directly from the constructions of ''contingent futurity." Such questions and statements seem rather to constitute several distinct idioms that come from various sources. I even doubt whether they ever become so closely allied as to merge into one homogene- ous usage, each part of which exerted its influence upon the rest. However, the question of the origin of the independent phrases lies beyond the scope of this study. I shall here simply admit my recognition of them, and present them in so far as they will help to determine the nature of the more complex form which we are examining. Most of the independent usages in Plautus are in the form of questions, cur is one of the favorite interrogatives. Those in the affirmative are Capt. 739, cur me esse saluom postulem? As. 45-9, quaeram; miniter; suscenseam; Ps. 1295, adflicter; Poen. 152, mentiar; ibid. 354, curem; Pers. 620, mirer. The questions in the negative (always non) are Cos. 701, cur non ego id perpe- trem? M. G. 317, rogem; Most. 209, curem. quin performs the same function as cur non 1 M. G. 426, quin rogem? In this list there is not a genuine question for information. All are uttered in order to reject a proposal directly made by another, or in the tone of soliloquy. All are in the first person. quidni is even more limited in usage ; cf. Cure. 423, nostin ? quidni noverim? Am. 434, negem; Men. 912, sentiam; M. G. 554, fateare; ibid. 923, noverim; 1120, censeam; 1311, fleam; cf. Ps. 96, 652. St. 333, rogitem. The expression is simply an impatient "of course!" The adverbial quid, besides introducing deliberative questions, at times occurs in questions exclaiming at the uselessness or unreason- ableness of an act or proposal: Am. 41, quid memorem? Bacch. 1049, properem ; Most. 581, recursem, etc.; Trin. 1024, petam. quam ob rem is found to perform the same function in M. G. 319, quam ob rem jubeam? Pers. 531, enumerem; St. 208, praedicem. If such questions were closely akin to the subjunc- tive (optative) of "contingent futurity," it would seem strange that they should thus regularly occur in the first person and only in this very limited function. They are clearly as far from 1 Perhaps faciam, Capt. 856, and feram, Most. 614, are subjunctives. See Sonnen- schein's note on the latter. (2nd ed.) 10 Tenney Fbank resembling the prescriptive usage in Sanskrit. I shall accept the old interpretation 1 that finds in them a sort of nervous and irritable kin of the other questions that appear in the first person in various functions and are usually called by the various names "delibera- tive," "dubitative," "challenging," "questions of remonstrance." An additional idea of duty conveyed by the introductory adverb is all that is necessary to change such questions into "questions of obligation." The question of this type is relatively less frequent in Cicero, although it is somewhat more varied in form, for it often intro- duces the third person, and at times the second, cur is the most popular interrogative in such phrases in the first person with a negative ; cf. cur non extimescam, Acad. ii. 87 ; sequamur, Fin. v. 47 ; Pomp. 50 ; Cat. iv. 2 ; Piso 79 ; in the affirmative, censeam, Phil. v. 16; N. D. 47; Div. ii. 116; Leg. ii. 25. These are similar to the Plautine instances and seem nearly akin to the deliberatives, as we showed above. I suspect that the sentences in the third person may have had a different history. At least they are often used in a different type of sentence; cf. cur minimo declinent intervallo, maiore non? De fato 46, "Why should (would?) the atoms swerve the least possible distance, and not a greater one?" This is a "should" of logic, and belongs to the hypothetical forms in argument. Perhaps in the consciousness of the writer they are hypothetical. Elmer's theory would find sup- port in such cases. Other similar ones are: Zenoni licuit, cur non liceat Catoni? Fin. iii. 15 ; cur fortior sit, Fin. iv. 53 ; desi- deret, Tusc. v. 89 ; adimat, v. 112 ; cur non habeatj sit, non sit, Bep. iii. 17. However, there is also an evident extension into the third person of the function so common in the first person in instances like cur non sit praeda communis? Phil. ii. 72; cur non iuret? Dom. 80; confidat, Flacc. 5. It may well be too that after all three persons had become familiar with this function in dependent clauses like quid est causae cur non, the usage recoiled upon independent phrases. At least such a possibility is suggested by sentences like : quid est causae cur non is, qui Catilinam ex urbe 1 1 need not undertake a detailed discussion of Elmer's arguments. Bennett has convincingly overthrown the proofs upon which Elmer laid the greatest stress, Cornell Studies IX, pp. 1-30. Semantics of Modal Constructions 11 pepulit, pertimescat; .... cur sibi confidat is qui .... cura- vit ? Flacc. 5. The second cur-question confidently performs the necessary function after the fuller phrase quid est .... cur set the example. The second cur-clause is hardly to be considered dependent. The other questions of this type show little change in Cicero. Note the usage in the following: quidni neget? Verr. ii. 80 ; quidni laudet? Off. ii. 76 ; possim, Tusc. v. 12. The usage is rare. quid is found in abundance in Cicero in such questions, almost always with the first person. Most of its occurrences are in set phrases of the type: quid loquar plura? Pis. 79; quid ego Vetera repetam? Verr. iii. 182; quid memoremf disputem? etc. State- ments as well as questions betray this meaning. Elmer (p. 221) quotes the following sentence as an illustration: Pers. 123-25, Cynicum esse egentem oportet parasitum probe : pallium marsup- pium habeat. Morris clearly analyzes the force of this (A. J. P. XVIII, p. 279): "with an ideal or typical person — the subjunc- tive expresses only propriety, the direct form of will being excluded." The writer is simply using a familiar form, a sub- junctive of command, in a situation that later acumen recognizes as somewhat incompatible with that form. I am even inclined to think that such cases may not be merely accidental, but had gained their independent existence as a recognized idiom. If so, their history is something of a parallel to the Sanskrit "prescriptive," though they are humbler in tone and probably related to expres- sions of will rather than of wish. The occurrences in the past tense, like Bud. 842, caperes aut fustem aut lapidem, have also received a convincing analysis by Morris (ibid., p. 281) : Obligation is one of the many meanings of the subjunctive in the present, though it is somewhat infrequent, and it is the only one which has, so to speak, survived the transfer from present time to past. The other shades of will or desire cannot be used of a past feeling. Command, entreaty, advice, permission, determination are excluded; only obligation remains. Such are the commoner modal expressions of "propriety and obligation" in independent sentences. We must conclude, I think, 12 Tenney Frank that they are largely unrelated and highly specialized idioms which have emerged from several sources, i. e., expressions of com- mand, deliberation, desire, ideality, and that they have hardly coalesced into one type, though some are similar enough in form and function to influence each other. It must now be even more apparent that it would be quite unsafe to refer these idioms to a common usage in the parent speech. With so much accomplished we may now turn to the more complex expressions. Having already listed them and marked their habits, it only remains to attempt a search into their earlier history in the light of such facts as they and the closely related constructions afford. The first group appears in the form of ordinary characterizing clauses. There is a slight difference in function. Could the characterizing clause as it stands assume this additional function ? Compare the clause of fact, nil est quod metuam, "there is nothing that I fear," with the clause of "propriety," isti quod suscenseam nil est, "there is no reason why I should be angry at her." The second has the same structure as the first, except that quod is an "accusative of respect" and thereby becomes a synonym of cur, "why." Now, the answer 1 to a question of "why?" either motivates an actual state (or act), or explains the propriety of a proposed or conceivable act; "it answers either "why does (did, will do) ?" or "why should he do?" But the negative antecedent of our stereotyped idiom excludes the first. We do not ordinarily motivate the non-existing. The implication is therefore inevitable that the question has become one of propriety. I am sorry I cannot make the general statement more simple, for I seem to hear my solitary reader retorting that this is too intricate for folk-logic. But I do not pretend that the process was conscious. It is merely due to the limitation of the experi- 1 This is true for the ordinary situations that call for expression. Rhetorical or figurative uses would extend farther. I may add the obvious fact that even the first sentence involves a fallacy in form, in that it seems to characterize an antecedent the existence of which is denied by the negative; but that is a fallacious form which language has permanently adopted. Logic can do nothing with expressions like " I don't think he is here." What I mean to say is that language may be frugal, with- out being logical. She will use an unthinkable form if she can make it stand for a common experience, but she will not create fixed phrases to express unusual experi- ences simply because they may be conceivable. Semantics of Modal Constructions 13 ences we have to express. For instance, we have little need for a collocation like, "There is no reason why he does it," while we hear every day the expression, "There is no reason why he should do it." Our own ambiguous expression, "There is no reason for doing it," is usually one of propriety regarding acts proposed or conceived of, though we somewhat illogically extend it, for example, in censure, especially with a leading verb in the past: "There was no reason for doing it." That is a sort of "past propriety." The subjunctive characterizing clause in Latin was just as careless in distinguishing between the actual and the hypothetical as that English phrase is. It was therefore as easily adapted to express both ideas as this English phrase may, according to the demands of the context, be equivalent to "does" or to "should do." The sentences in the second person behave somewhat peculiarly, and yet naturally enough. In questions with quid est quod, even when quod is used adverbially ( = cur) , the tone continues to be that of a fact-characterizing clause (cf. p. 2, note) whereas after statements both affirmative and negative, when quod = cur, the feeling of obligation is strong. This points to a similar bit of psychology. We ask a man to explain his motives for an act (cf. quid est quod timeas, "why do you fear?") ; we are apt to tell him his obligations (est quod habeas gratiam, "you should be grateful"). I may add that the negative forms according to the above-given theory would be the source of the affirmative ones. If the theory is correct that such clauses assumed their peculiar function while serving as characterizing clauses, this will add one more illustration of a fact I have insisted upon before, that there is great danger in the popular method of referring to parataxis for a solution of. every dependent construction. No doubt simple questions with quid and cur influenced the complex form quid est quod. In fact, we have shown that quid and quid est quod are nearly equivalent in Cicero, though the shorter form is more popu- lar. However, many of the complex clauses of this form assumed the task of expressing obligation or propriety late, when in their form of characterizing clauses the resulting collocation suggested their fitness for that function. Besides, there would be no little difficulty in referring the nil est quod idiom to a paratactic stage ; 14 Tenney Frank for that would involve the assumption of an interrogative use of quod, whereas quod seems to have borrowed its subordinate con- structions from qui. Furthermore, as the idiom occurs in early Latin, it rarely appears in the first person, where we should expect it to appear if its simple element were an ordinary question of deliberation. Mr. Durham, in his excellent collection of substantive clauses (Cornell Studies XIII, p. 77), ventures the belief that "such expressions are a development of the relative clause of purpose." The purpose idea is certainly not far off, though as he points out, it could not directly explain such substantive clauses as quid est quod abeas, "why do you go away?" This seeming close con- nection is due, I believe, to the fact that the relative characteriz- ing clause is ultimately derived (at least in large measure) from expressions of will and tendency, 1 a doctrine now very unpopular, but I believe true. The actual evolution is then as follows: Clauses of purpose and tendency create characterizing clauses; some of these in turn evolve into expressions of obligation and propriety. In the second group, nulla causa est (quid causae est) quin, etc., we must again trace several idioms. The interrogative form usually appears in a verb in the first person: cf. quid causae est quin te sauciem? Rud. 758 (also Bud. 261, Trin. 1188; Ps. 533, Amph. 852) ; cf. also Cic. Quinct. 84, Q. Rose. 41, Leg. Agr. ii. 40, 70, etc. Kienitz De quin particulae usu, is probably right in explaining some of the quin-cl&uses as extensions of the deliber- ative usage. That is probably the source of these complex questions in the first person, though Plautus shows only one independent question with quin of the type that we should have to assume for their basis: M. G. 426, quin rogem? Morris 2 suggests that in such sentences the part that constitutes the main clause, nulla causa est, was originally spoken to emphasize and define the tone of the i During the last few years this theory has so steadily been rejected by our Ameri- can school grammars that the Gildersleeve-Lodge textbook which retains it has become quite heterodox in its conservatism. I have no space for a discussion here. Suffice it to say that a new and long study of the facts (the results of which may appear later) is the justification for the attitude here adopted. 2 Aims and Methods, p. 159. Semantics of Modal Constbtjctions 15 question that followed. Several of the Plautine instances were obviously not so simple. Some have come under the influence of, or arisen from, a legal phraseology [nullum causam dico quin siet, Capt. 626; Amph. Fr. 13; Amph. 852; also hand causiflcor quin earn habeam, Aul. 756), and are grammatically akin to the expres- sions of hindering, preventing, and the like, with quin. Finally, some bear the appearance of having passed through the history of other gwrn-characterizing clauses. In form there is little difference between nullust JEphesi quin sciat (" who does not know ?" ) , Bacch. 336, 1 and, hau causast .... quin facias ("why you should not do"), Most. 435. If these are originally of the same group, as I suggest with some diffidence, the subsequent history of the second, as to its new function, would be similar to that of the nihil est gwod-clauses, which I attempted to trace above. Be the original form of these various clauses what it may, the psychology that adapted them into expressions of propriety is much the same. As in the preceding group the keynote of the antecedent is motive, reason, or cause. It is even more explicit than in the group with quod, for causa is here expressed. The connective quin reiterates this note. The negative in the antece- dent precludes the possibility of its being a characterization of an actually existing motive or reason. The inevitable implication of the whole is that of propriety, obligation, reasonableness: an ethical one if the idea underlying causa is ethical, logical if that of causa is logical. The dependent cwr-expressions of obligation have similar his- tories. We have found cur some twelve times in Plautus intro- ducing an independent question in the first person and with the meaning here discussed. Cf. cur mentiar? Poen. 152. Now, nequest cur studeam, St. 52, may be connected almost directly with these. The widespread usage of Cicero is somewhat more complicated. We have seen how the independent question with cur not only extends into the second and third persons, but also is frequently used to express a logical necessity, an inference that ought naturally, 2 should, must, follow from the existing state of i Cf . Bacch. 1012, Cist. 18, Cure. 287, Men. 362, Pers. 365, Trin. 534, St. 208, Am. 1054. 2 The Hale-Buck Grammar was the first, I believe, to distinguish such usages. It employs the term "natural likelihood;" see § 515. 16 Tenney Fbank affairs. Cf. cur minimo declinent, Defato 46, "For what reason should they swerve only the least possible distance?" Zenoni licuit, cur non liceat Catoni? Fin. iii. 15, "Zeno had this privi- lege, by what reasoning do you contend that Cato should not have it?" This same function is as freely performed by the dependent cttr-clause, like nihil est cur alius alio melior existimetur, Font. 22, "There is no ground for concluding that ....," or "From the case one should (would naturally) not be esteemed better than the other." In the type represented by the last sentence the subjunctive may be hypothetical, 1 the added function coming from the nature of the subject dealt with (i. e., the reasonableness of the argument) . It may, on the other hand, be an extension in hypotaxis of the type so freely found in the first person in Plautus. Whatever the source, the complex and the simple forms naturally influenced each other, for they continue to be nearly synonymous, the simple form being somewhat more incisive. I have repeatedly mentioned the distinction between the logical and ethical functions of such clauses. They seem to bear the same relation toward each other as the two ordinary meanings of words like oportet, debet, Set, "ought," etc. In English we say, referring to an ethical sanction, "you ought to help this poor man;" referring to a logical sanction we say, "your thirst ought to be quenched, for I just gave you a cup of water." So in Latin, oportet in Amph. 78, uirtute ambire oportet, non fauitoribus, "an actor ought to win through excellence, not through favor," refers to what is morally proper, while, in Amph. 322, 2 olet homo quidam malo suo; haud longe abesse oportet, "he ought to be somewhere in the vicinity," it involves a question of reasoning. The second is probably a figurative usage of the first, having begun by attributing the language of moral obligation to that which does not exercise the faculty of moral control. So the second usage of the subjunctive of obligation may have grown up from a figurative use of the first, and may have been suggested and aided by the 1 This would probably be Elmer's explanation, except that he would use his own term "contingent futurity." 2 Other examples of this use of oportet in Plautus are Amph. 316, 318, 327, 783, 1031 ; Asin. 381, 465 ; Bacch. 552, 602, etc. Semantics of Modal Constructions 17 double use of oportet. If this is the true history of it, and if oportet is an exact parallel, the hypothetical subjunctive had less to do with the origin of it. As for quam ob rem, its subjunctive verbs in the early inde- pendent questions are in the first person. Cf. M. G. 319, quam ob rem iubeam? etc. Its early dependent form we have seen in such sentences as St. 82, nil quam ob rem 1 id faciam meruisse arbitror. Its constant appearance in Plautus with mereor (which usually takes an w/-clause) may be accidental and not the sole construction of that time. quam ob rem does not show its early partiality for mereor in Cicero. There it has become in dependent clauses a fair synonym of cur, appearing in the same kind of expressions, though with less frequency. A careful search throughout Latin will discover other dependent clauses expressing obligation, if the subject-matter of the antece- dent is favorable. This is true of such connectives as qua propter, qua re, qua de re, quam ob causam, or any other connectives that deal with motive or cause. But such occurrences were sporadic and never came to be adopted as fixed idioms. They would add little light to the study. In this discussion we have traced extensions of function only where a clear formal similarity appeared. Now the serious question arises as to whether these idioms, coming as they do, from various sources, in any way gave rise to a more general feeling that the subjunctive alone was adequate to convey the idea of obligation or propriety in Latin. May we ever at any period of the Latin language speak of a basic conception (Grundbegriff) of obligation or propriety as underlying some of the subjunctive usages, apart from these fixed idiomatic types ? This question involves another fundamental problem regarding the limitations of analogical pro- cesses in syntax, namely: can functions extend from group to 1 Though quam ob rem, independent, is oftener an interrogative than relative in Plautus, in dependent clauses of this type it is usually relative, introducing a consecu- tive clause. There is a peculiarly close connection between the ideas of "merit," "desert," "worth," and those of due and proper reward. Hence the collocation mereor .... quam ob rem .... habeam (=oporteat habere). Hence, too, in Eng- lishing ut- and qui-clauses after dignus, idoneus, mereo, and the like, one readily falls into some expression of propriety. Of. the Hale-Buck Grammar, % 514, 3. 18 Tenney Frank group except through similiar forms; and, if not, what are the limits of dissimilarity through which such extensions may not take place ? At the present stage of syntactical investigation this ques- tion cannot adequately be answered. Confessedly, investigation on this point is extremely difficult. In fact, it is the lack of knowledge on this problem that has rendered many a long-spun explanation in syntax mere ephemeral hypothesis, built up of indi- vidual opinion, only to be denied by more individual opinion. The assumption that one function may influence another regard- less of any direct formal analogy has been frequently ' made. But it must be admitted that the meager data which have been produced to substantiate the assumption do not form a sufficient basis to justify the lofty superstructure built upon them. Our present study extends over so small a range as to furnish conclusions for itself alone. If a great number of life-stories were sketched along the lines attempted here (with greater keenness of analysis, let us hope) , the necessary facts might be forthcoming. The conclusions from the present study can only be that we have found no usage in Latin that is not merely a brief extension from one form to another very similar form. The summary at the end should make this fact clear. Coming back to the previous question, then, we must answer that, since we have been able to trace the function here discussed from one fixed idiom to another throughout the usages found in Latin, and since we have not been forced to posit any general Grundbegriff to account for any such usages, we should not be justified in assuming that such basic concepts ever came into existence in Latin in connection with these idioms. In fact, the behavior of these idioms would present peculiar obstacles to such a possibility, for the phraseology in each idiom is so unique that the function of obligation would be felt as emanating from the combined elements of the phrase (e. g., car-non-faciam) and not from the mood of the verb alone. Unless the function was clearly associated with the mood, in individual instances, a general con- ception could hardly arise that the mood in general was capable of performing such function. il must confess that in writing Attraction of Mood in Early Latin I made such an assumption to an extent that I now feel cannot be supported. The intricacies of that problem will receive attention later. Semantics of Modal Constructions 19 Let us now summarize our observations on the story of modal expressions of obligation and propriety. I. These expressions differ so widely in form and function in the different languages of the Indo-European group that we are not justified in assuming a proethnic usage as the source of the different idioms. However, Greek and Sanskrit idioms furnish illuminating parallels for various Latin usages. II. Early Latin shows the following independent idioms per- forming this function, all containing a subjunctive verb: 1. Questions with cur, quidni, quid, etc., in the first person present (akin to a large group of questions of the first person), (a) A beginning of an extension into other than the first person, in quidni fateare, M. O. 554; largely extended by the time of Cicero, (b) Extensions into past tenses and all three persons, a highly specialized idiom that is never popular ; cf . caperes fustem, Bud. 842. 2. Statements in the third person; a figurative usage of the jussive directed toward "ideal or typical persons;" cf, marsuppium habeat, Pers. 125. III. In dependent clauses we find the following usages, whether direct extensions or new assumptions of the function. 1. A direct extension into dependent clauses from the indepen- dent questions, the interrogative acquiring subordinating function. a) By direct parataxis. Cf. quid mereor quam ob rem men- tiar, Most. 987. b) Through "defining" parataxis. Cf. nulla causast quin me verberes, quid causaest quin te sauciem ? Bud. 758. (Kemark: There is probably a reflex extension from such dependent clauses into new and freer independent questions and statements in all persons; cf. Flacc. 5). (1) Development of such into full characterizing clauses of all persons. Cf. nihil erit quam ob rem requiratur, Font. 25; nihil est cur alius melior existimetur, Font. 22. 2. Certain dependent clauses when already in the hypotactic stage assume this function. a) At times quite apart from any influence of the independent usage. Cf. the regular consecutive clause when dealing with 20 Tenney Frank motive or reason : nil est quod timeas, Am. 1132 ; est quod gaudeas, Trin. 310, etc. b) At times under the influence of the independent question. E. g., cf. quidplura dicamus? with quid est quod plura dicamus? IV. The above-mentioned extensions are the only ones for which our observations justify a certain affirmation. It is not impossible, however, that the Romans used the subjunctive even more extensively as a synonym of oportet, debet, etc., having gained from the above-mentioned idioms a conception that the simple subjunctive was capable of serving in that capacity. Finally these two studies have led to some conclusions of more or less general applicability. It may seem like a case of petitio principis to explain one modal construction by direct reference to another. Thus, while we have refused to recognize an independ- ent use of the subjunctive in expressions of "capacity" and "obligation," we have in the very same lines accepted the "hypothetical" and "volitive" uses as though they needed less explanation. Morris, for instance, from whose work I have received many helpful suggestions, does not expose himself to the charge of inconsistency by making such connections. He goes so far in avoiding the assumption of Grundbegriffe that he is prone to see a new and open problem in every construction. His attitude in this matter is well shown in the following lines from his paper in A. J. P. XVIII, p. 395: To explain di te perdant and ualeas by saying that they got the optative function from di te perduint and saluos sis, and that these acquired it at some remote time in some unexplained way, is to turn away from the simple forces, working under our eyes, to a vague hypothesis which, after all, explains nothing. The forces which can give an optative sense to salue or di te amabunt will explain all optative subjunctives. We have ourselves shown that with a sufficiency of "forces" an indicative or even an infinitive form might be employed to convey "potential" meaning. But the fact remains that it was the optative in Greek and the subjunctive in Latin that regularly permitted the growth of "potential" idioms. This optative and this subjunctive persist and must be explained quite as much as the function in question. Modal forms are quite as persistent as Semantics of Modal Constructions 21 word-forms. In the illustration Mr. Morris gives, a future indica- tive is found expressing wish in a certain fixed idiom, but amabunt standing alone would never express wish, while perduint even apart from its lexical meaning needs nothing but the most normal context to be a perfectly explicit optative, and even words need some context to be explicit. In dealing with modal constructions one cannot apply a single method to all alike, for they vary in age, stability, and wideness of range. Of course there is little benefit derived from referring all constructions to the limbo of what Morris calls "some remote time in some unexplained way." But some actually belong there ; they are no longer questions open for solution. There are certain widespread usages, e. g., of the volitive and hypothetical, that appear in several languages. They are quite as independent of context for their function as are ordinary words. No doubt many of their idiomatic usages need sifting by the pro- cess applied by Morris, 1 but it will prove dangerous to consider every one explicable by the data at hand in historical times. The principle I have followed then is something like this: When comparative syntax has revealed a common usage of a mood in free form and not only in fixed word-groups, I have assumed that the usage was established at a time and under circumstances into which it may now be futile to attempt to pry. Such usages it is reason- able to accept as earlier ; they must have served as an ever-present force in the creation of later ones. Others, such as those here treated, that are limited to stereotyped forms in individual lan- guages, may well be considered problems for solution. If then the most stable element in these is the modal form, that form is prob- ably the clue of nearest relationship. The minor details of this method have been illustrated in application and need no further discussion. Bbyn Mawr College J The methods of Morris may perhaps suggest how such usages originally arose, but the mischief may be greater than the benefit if he thus explains as primitive what is tertiary or worse. How, for instance, can the question of "adaptation versus agglutination" affect our solvable problems of syntax, when between the date of the early word-formations and of our first records a dozen Grundbegriffe may have come into focus and broken up again? (See Morris & Oertel, Harvard Studies XVII.)