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. THE SEMANTICS OP MODAL CONSTRUCTIONS By Tennf.y Frank I. NON HABET QUOD DET The recent activity in syntactical study of languages, ancient and modern, has been unusually suggestive as to methods, though perhaps not so fruitful in actual results. The emphasis laid upon psychological forces and the application of the methods of com- parative grammar, the adoption of hints from the sciences of morphology and semantics, the continued patient devotion of a few to the historical grouping of large ma sses of material, have al contributed much toward creating a hope that chaos contains the possibility of order even in this field. However, the enthusiastic devotion to a single theory that is wont to assert itself in discov- erers has often resulted in the overemphasis of partial methods, and has perhaps needlessly postponed a desired synthesis of meth- ods that might bring more abiding results. Furthermore, in modal syntax, with which we are here concerned, the intricacies of the problems presented by the more complex sentences have intimidated many students into confining their researches within the simple sentence. Many of those who have essayed the harder problems of hypotaxis have been satisfied to explain the enigmas of the complex group by a too ingenuous reference to the behavior of the simple ones, not realizing that the compound of two or more simples even in language is often a wholly new entity, not a mere sum of the original forces. It was with this general situ- ation in view that the present studies 1 were undertaken. I believe that what is most needed at present is a number of life-history studies of phrases, clauses, and idiomatic constructions, just as the larger problems of biological evolution need a great number of life-histories. Such must not only be descriptive in nature, but, to be of the greatest service, must also connect the individual with its cognates in the general scheme of evolution. 1 A second paper will follow, devoted to expressions like nulla causa est quin det. [Classical Philology II, April, 1907] 163 164 Tenney Frank The complete story of a highly specialized idiom like this involves three tasks. (1) The idiom must be described as it behaves in the language studied. (2) Its grammatical relationships within its own language must be examined for light upon its immediate provenance. Here care must be observed to search for such forces in its context as may have created, fixed, and preserved the idiom within the language. (3) Cognate constructions in the closely related languages must always be scrutinized lest we explain as of Latin origin a construction that is due to circumstances not existing in Latin. I hope that the paper will not only furnish data in regard to the idioms here studied, but will give some definite proof as to the limits of applicability of the methods now generally used, as well as offer some new suggestions for future work. I. Expressions like (won) habet quod (let, with potential force, are of frequent occurrence throughout Latin. In Plautus they appear nearly a hundred times ; in Cicero the number exceeds this ; and they continue to be familiar throughout Latin literature with little change in form or function. The relative clauses contained in them are sometimes grouped in the grammars with so-called consecutive or characterizing clauses. They are like these in that they qualify an indefinite or negative antecedent of the same kind. They differ from them, however, in that they are invari- ably in the subjunctive mood, whereas ordinary characterizing clauses may at times be found in the indicative, particularly in early Latin and in the poets. Secondly, the character that they express as a consequence of the nature of the antecedent is not expressed as an existing fact, but rather as possible, nihil est quod det in this idiom means, "he has nothing that he can give," "he has not the means." The idiom is usually a mere statement of existence or non-existence of some means not further defined, but the nature of which is made evident in the relative clause that follows, (non) est, (non) habet, {non) copiast, (non) paratumst, are the antecedents in a large majority of the occurrences; often, however, a semi-defining noun stands as antecedent, as in nil est dotis quod. In the most characteristic phrases the negative ante- cedent is somewhat more frequent than the affirmative. Semantics of Modal Constructions 165 The most striking peculiarity is one that I do not think has been pointed out. The relative clause in the group nil est quod det (except in a limited class of instances) changes its function materially with the change of case in quod: 1. When quod is the subject of det, the clause expresses a fact: "there is nothing that gives, etc." 2. When quod is used adverbially, the clause expresses an "obligation": "There is no reason why he should give." 3. When quod is the direct object of det, the clause is "poten- tial in force": "There is nothing he can give." We are here concerned with the third type. The second type will be taken up in the next study. Some data will make the matter clearer. Of the relative 1 clauses in Plautus that express capacity or "potentiality" in modal idioms, almost all use the relative as the direct object of the sub- ordinate clause. In this function quod occurs more than forty times; quern, quos, etc., ten times. Finally the dative and abla- tive objects appear a few times. This is not saying that the relative never serves as direct object in pure characterizing clauses of fact. It may do so when the action of its verb is not purely momentary and physical in the present time. This exception holds true of most verbs whose action covers more than the imme- diate present; cf. nil est qain dixerim (Bacch. 1012), "There is nothing that I have not told him;" of general truths; cf. quisnam fluviust quern non recipiat mare {Cure. 86) ; also verbs denoting mental activity and therefore coming under the same rule; cf. nullast quam aeque amem (Am. 509), "There's no girl I love so much," ". . . . quern metuam" (As. Ill); ". . . . quod cupiam'''' (Cure. 172), etc. All of these express actuality, not potentiality, Now contrast these with nil habet quod det, .... faciat. .... videam, .... edim, etc. The latter do not express actu- ality. They convey an additional idea, namely, that of posse. This distinction seems so intricate that it will probably be suspected by the grammarians who fear "metaphysical syntax," but it is not a distinction of Latin alone. It holds true in general for other languages also; for it denotes a limitation of thought, 1 1 omit for the time being clauses with unde, ubi, qui-instrumental, and the like. 166 Tenney Feank and therefore naturally of expression. In English we may say, "There is no one here that I love, .... hate; .... that I wish to see, .... that I have ever seen, .... that I (ever) associate with." "There is nothing here that I wish to eat, .... can eat," etc. But we do not — never have reason to — say in a literal sense: "There is nothing here that I eat" (in the sense of "am eating"), ". . . . that I see, strike, touch." Nor do other languages use such phrases. The following expressions would hardly be intelligible, except perhaps in some figurative sense: Es ist keiner hier, den ich sehe; II rCy a personne id que je voie; Det finnes ingen her, som jag ser, etc. Now, the remarkable fact about the behavior of Latin in the use of this idiom is that, while in other languages we neither express nor possess any idiomatic collocation that might express such a notion, Latin has such a collocation which, taken at face- value, as we understand such values in the Latin we possess, would express that piece of aSvvara el/cora, which, however, expresses something quite different, i. e., capacity. The phrase nil habet quod det, with quod as an accusative, could therefore never have been in its present form a pure characterizing clause of fact, as is the case with the nominative quod. It may have been an extension of such, adapted to a new function. It may, as I shall suggest, be connected with a relative clause of purpose, or it may have come more directly from independent uses of the subjunctive after quod had acquired the power of subordination. As for the current method of classifying such clauses, it is usual to include them under the consecutive relative clauses without explanation. Mr. Hale 1 believes that they come from an independent "potential." Mr. Elmer 2 thinks that they must either be grouped with volitives or with what he terms the sub- junctive of contingent futurity. Such attributions imply certain life-histories, even if these are not given in full. The truth may be that several usages are here included in one group. Whatever the past history may have been, I believe that the expressions described above made up a fairly homogeneous group by the time i Cf . Cum-Constructions, pp. 106, 107 ; Trans. A. P. A. 1900, pp. 160 ff. ; Hale-Buck Grammar, sec. 516. 2 Cornell Studies VI, p. 196. Semantics of Modal Constructions 167 of Plautus, and acting together as a definite Gebrauchstypus, and that we are doing no violence to the correct conception of language in studying them as one consistent group. The idiom is fairly stable throughout Latinity. Some later divergencies are noticeable, however. Cicero, while showing no aversion to the idiom, is fond of substituting for it a more explicit expression of capacity; i. e., a characterizing clause whose verb is some subjunctive form of possum or queo, as for example, habere se putant quod .... queant dicer e (Lael. 71; cf. Off. 2. 39; Div. Caec. 10; Caecin. 36, etc.). This habit is no doubt in accord with the equally redundant habit of ending an implied quotation of a reason with a verb declarandi; e. g., egit quod aequum esse diceret. The fact is obvious that the subjunctive was assuming an ever-increasing burden of work, and was becoming so overloaded that an exact writer must unconsciously have used such means for the sake of clearness. Another novelty is the appearance of nil habet quid det, if in fact its non-appearance in Plautus is not to be attributed to a Plautine or colloquial peculiarity. It seems to be a relatively late outgrowth of a semantic change in habeo, which verb in this new idiom becomes almost a synonym of scio. Plautus has nil habeo certi quid loquar (Mil. 407), and Terence uses the phrase once in a dependent but "extended" deliberative of the first person: Andr. 498, teneo quid erret et quid agam habeo. By the time of Cicero the new idiom is frequent. Considering that this use of habeo is highly specialized and late, we cannot bring forth habeo quid as an explanation of habeo quod. It is more probable that the influence worked in the opposite direction. It is also probable that the Greek o&x ex<° on played a considerable part in the creation of (non) habeo quid. I do not mean to imply that these two expressions were ever perfectly synonymous, but they certainly overlap in function at times. There is finally an interesting example of "contamination" in the late Latin idiom non habuit quod (quid) dicer e, which is obviously due to a fusion of non habuit quod (quid) diceret and nihil habuit dicer e. Cf . Je n'ai que faire de lui. 1.68 Tenney Frank I have considered it worth while to give a full list of the Plautine examples to show the early usage of the idiom. For later authors abundant examples may be found in the lexicons. We have not space for them here. I have grouped them accord- ing to the nature of the antecedents and connectives, and have separated the affirmatives from the negatives. Such as depend upon other subjunctives and infinitives of indirect discourse are inclosed in brackets. 1. est quod, -unde, -qui, etc.: est domi quod edimus (Poen. 537); erit extemplo mihi quod dem tibi (Capt. 122); doleat (Cist. 67): miserescat (Ep. 526); des (Men. 492); dicas (Merc. 642); gaudeant (St. 394); (ut quod abrodat sit [Am. 724]; det [As. 83]; edis [Poen. 867]). sunt divitiae? unde excoquat sebum (Capt. 281); occipiam (M. G. 676). est qui ventrem vestiam (St. 376); (qui conteram [As. 419]; possis [Trin. 653]). est ubi ambules (Capt. 12); (ut sit ubi sedeat [M. G. 82]; essem [True. 140]; quern caederes [Am. 377]). a) Negatives of the same class : non est quod dem (Capt. 121; As. 242; Trin. 761); (nega esse quod dem [St. 256]; darem [Pens. 119]; det [Cist. 737]). nil est quod respondeam (Trin. 188); dem (Aul. 238); perdam (Men. 665). nil est qui tergum tegam (Most. 992); dimminuam (Men. 304); funga- tur (Trin. 354); vivamus (Trin. 561); quo recipiam (Capt. 103); quicum litigent (Poen. 587); qua possit (M. G. 329). non ubi sedeas locus est (Capt. 12); (ubi possit [As. 767]). unde dent (True. 146); quos emerem (Pers. 262); quibuscum haberes (Bacch. 564); nee tibi qui vivas domist (Capt. 581); qui ulciscerer (Men. 636). quod dem nusquam . ... est (As. 631); queat (Trin. 801). nullus est tibi quern roges? (Ps. 294); qui reddatur (Bacch. 609) (= quern reddat); qui negem (Capt. 937); despondeam (Merc. 614); procurent (St. 200). 2. Other verbs of existence ; habeo, copiast, paratumst, etc. : habet quod det (Poen. 833); dent (True. 76); ut habeat quod det (As. 103); des (As. 188); renumeret (Bacch. 46); feram (Trm. 728); gwod amesparaiwwstf (Bacch. 219); edit (Poen. 9); gwod ames . . . . praestost (Ep. 653); agat (Af. G. 205); copiast . ... qui solvas (-Rwd. 558); feabeo Semantics of Modal Constructions 169 unde dem (As. 234; Trin. 158); expetam (Cist. 671); qui utamur (Trin. 335, 336); (qui possent [Ep. 384]). a) Negatives of the same class: quod edit wow habet (Capt. 463; Trin. 564; True. 213 [True. 242]). neque id wwde efficiat habet (Poen. 185); occipias (Ps. 399); wW collocem (i£p. 531); (wow habere quoi commendarem [Merc. 246]; qw> intromittar [ilfew. 669]; o/«i nosceret [Pwd. 393]; qui aleret [TVw. 14]). 3. Similar expressions conveying a potential idea may follow verbs of action like dare, quaerere, etc. A volitive idea is always evident in these expressions, especially when the leading verb is an imperative (or subjunctive of will). Sometimes the volitive idea is the predominant one. Their relationship with the pre- ceding classes will be discussed later. dat quod edit (Trin. 359); raro dabat quod biberem (Cist. 19); qui occideret (Trin. 129). da illi quod bibat (Most. 344); dem (M. G. 692); monstra quod bibam (Men. 742). da ... . ubi condormiscam (Bud. 571); ubi ames (Poen. 603; Cure. 311; St. 752; Aul. 407). quod edit quaerit (Capt. 461); quaero hospitium ubi curer (Poen. 693). dum illi quod edit .... praebeas (Men. 90). ratu's nactum hominem quern defraudares (Hud. 1387 ; Bacch. 506, 864). vix reliquit deo quod loqueretur (Cist. 150; cf; As. 233; Aul. 14; Cas. 251; Cure. 388). ewe lanam ttwde tibi pallium conficiatur (M. G. 687; cf. also Aul. 672; Bacch. 233; Merc. 53; M. G. 768; Poew. 178, 668; Trin. 558). Some negatives of the same nature are: nee quod edim quicquam datur (Poen. 1284); quern rogitem (Bud. 226; Trin. 135, 155). There is also a peculiar instance with Mud quod: "blanditur dum Mud quod rapiat videt [Men. 193). It is usually believed that Me has not yet assumed the characterizing function in Plautus. Perhaps this is an early instance. The clause is cer- tainly not one of "volition." The meaning of videt precludes that interpretation. II. We have seen that Latin literature exhibits our idiom as well established and almost unchanging. Yet it bears the appear- 170 Tenney Frank ance of being highly specialized. Latin possesses verbs and collocations of words with which to express capacity and possi- bility. Why should our elliptical expression with its verb invariably in the subjunctive have invaded this field? A scrutiny of the material given above, and a comparison of this with similar but more primitive constructions, may give some hints as to its immediate provenance. Its most stable element is the mood of the dependent clause. On the basis of this the comparison should be made. 1. There is a close resemblance between this and relative clauses of purpose. Purpose expressions readily fall into expres- sions of character and capacity. When one says: "I shall hire a man to do this," his words leave the implication that he will try to get a man who can do it. Language constantly reveals phrases that betray this habit of our thought processes. English has frankly adopted an auxiliary of capacity for many of its pur- pose expressions ("that .... may," "in order that .... may''''). Even the English infinitive, which is largely a dative of purpose, often expresses not purpose but availability: "Water, water everywhere, nor any drop to drink.'''' Greek did not hesi- tate to express a purpose clause in the optative with av or icev which is the mechanism most frequent with the various "potential" idioms. Cf. Od. viii. 21: KCU /XIV [JWLKpOTtpOV KCU TTCUTCTOVa. 6fJK€V lOOT&U ws Ktv QcuyKtcrcri <£t'A.os iravTecrcri yivovro. See also v. 166. What is even more surprising, the mechanism is found in a relative clause after a primary tense in II. i. 64, dAA' aye. Siy nva fidvTiv iptio/JLCV , . . . os «' ciiroi. Now, returning to Latin, note in the following series the gradual submission of the note of will to that of capacity: a) ut commostrernus locum ubi potes, " to show you a place where you may drink" (Poen. 603). quo furatum venias vestigas loca (Rud. 111). b) quaerit quod edit (Capt. 461). da illi quod bibat (Most. 344). c) paratumst quod edit (Poen. 9). nee quod edit quicquam datur (Poen. 1284). Semantics of Modal Constructions 171 d) est domi quod edimus (Poen. 537). nee tibi qui vivas domist (Capt. 581). The same change may be observed after the leading verb relinquere in the following: a) relinque aliquantum orationis eras quod mecum litiges (Cos. 251). b) satin vix reliquit deo quod loqueretur loci (Cist. 150). c) agri reliquit non magnum modum quo .... viveret (Aul. 14). d) non omnino iam perii; est relicuom quo peream magis (As. 233). e) qui me aleret nil video esse reliqui (Trin. 14). If the first sentence had been introduced by est relicuom (practically equal to est or habet) instead of by the imperative, which gives a tone of intention to the whole, we should have found a strong potential tone present. In other words, in propor- tion as intention shades off from the leading verb, and as the leading verb changes into the tense and voice that make it a mere expression of the existence or non-existence of an effective agent, the relative clause shifts from the tone of a purpose into that of a "potential" expression. Observe again that with a leading verb expressing volition by means of an imperative or subjunctive, the dependent verb even in this form has a predominantly volitive shading; e. g., Am. 724, u ut quod abrodat sit/' 1 As. 105, "ut habeat Alius amicae quod det ;" and several others in the list. In fact, in such situations Plautus often finds it necessary to rescue the potential idea from being thus overwhelmed by the volitive of the main verb. This he does by means of the verb posse, which is not otherwise necessary. Cf. Trin. 653, ut tibi sit qui te corrigere possis. In a statement of fact he would probably have written est qui corrigas. Other good instances are -4s. 767 ; Epid. 384; M ere. 53 ; Trin. 155. This again betrays the very close connection between the two idioms. I would add an illustration of a slightly different type. Com- pare "aliquid quod poscas paras,' 1 '' As. 168, with "quod ames paratumst," Bacch. 219. In the former an active verb, energetic with intention, fills the dependent clause with purpose. In the latter the verb is passive; the end has been attained, the means secured. Now, means implies capacity, the expression of which becomes predominant in the dependent clause. Other expressions 172 Tenney Frank like the second one are: Bud. 558, "tibi copiast . . . . qui rem solvas;" Epid. 653, "tibi quod ames domi praestost opera mea." (Note how the words opera mea keep a reminder of effort, i. e.> of a purpose expression, still lingering in the sentence.) To an intelligent mind, even the words copia, praesto, and the like are superfluous, "est qui rem solvas" becomes fully sufficient to reproduce the idea required. My conclusion is therefore that the habet quod det idiom may well be an extension of the ordinary relative purpose clause, 1 which is more primitive and probably proethnic; or at least that legitimate and natural extensions of purpose expressions readily become "potential." The mechanism of this extension is in the main quite simple. Verbs of effort and intention in the active voice, and particularly in the imperative mood or in the subjunc- tive of will, are naturally followed by clauses that express primarily will and intention, secondarily a capacity to attain the desired or intended object. Now, when the leading verb becomes a passive expression of the attainment or existence of the means sought for — that is, when the tone of effort or intention wanes — the idea of capacity becomes predominant. If we are ready to admit this extension of the purpose clauses under certain conditions, there is still abundant room for doubt whether such extensions should still be treated individually as sporadic and remote cases of purpose clauses, or whether they con- stitute a new and well-defined group with form and function distinct from the purpose clause, acting as a new independent idiom. What follows may help to decide the question. Meanwhile I shall assume the truth of the second supposition. 1 The extension of function that the infinitive of purpose passed through in most languages furnishes an illuminating parallel for the story I have here sketched as well as for the next study. Note the behavior of the adverbial infinitive in English. It began as a dative of purpose (to writanne) governed by the preposition " to." This use still prevails of course. However, in various surroundings the purpose idea weakened and other forces prevailed. It came to suggest destination ("ready to work"), result or consequence ("he was so weak as to yield"), character ("he is a man to thrive in the world"), capacity ("easy to do"), availability ("not a crumb to eat"), propriety ( ' ' worthy to be read"), obligation ( ' ' there are three strokes to make ;" at least, the golfer will agree with me that this denotes obligation), necessity ("nothing is to be done"), and so on. The road is as long from da bibere to tempus est abire in Latin, and the extensive use of the infinitival constructions in Sanskrit would furnish an abundance of additional illustrations. Semantics of Modal Consteuctions 173 2. The sketch here given attributes the change of function to forces found within the fully developed complex sentence. It does not explain anything by reference to usages in parataxis. There are signs, however, that some independent usages exerted an influence in the process. A large proportion of the verbs of these relative clauses appear in the first person, and many of them follow expressions like nil est, non habeo: "I am at a loss." In both of these respects they resemble dependent "deliberative" expressions. Sentences like nil est qui tergum tegam (Most. 992), "I have no means of protection for my back;" and neque unde auxilium expetam habeo (Cist. 671), are not far from the para- tactic forms, in which the qui- and tmde-clauses would be taken as "questions of deliberation." Mark the use of qui? It is an instrumental ablative, and therefore, even where introducing simply questions of the first person, like qui ego istuc credam tibi (Merc. 627), it steeps them in a tone of (in) capability. So strong is this feeling that qui is found continually associating with various forms of posse? The same may be said to some extent of quo, qua, unde, and ubi. When "deliberative" questions with qui (= quo modo) appear in hypotaxis, they do not necessarily abandon the tone they had before. A large proportion 2 of the examples of our idiom use qui (="how") as a connective. In this discussion of "deliberatives" I have assumed that in- dependent questions like qui ego istuc credam tibi? should not be regarded as independent "potentials," but as deliberatives. Let me illustrate. Here are three questions expressed in the first person: (a) quid faciamf (b) qui faciam? (c) cur faciam? [(a) "do what?" (b) "do it, how?" (c) "why do it?" i. e., "what shall I do?" "how can I do it?" "why should I do it?"] The first would normally be answered by an imperative; the second, by a statement regarding the ways and means, i. e., by some form of posse; the third, by a statement regarding the ethics of the situation, i. e., by some form of oportere. Now, by a well- respected, but, as I think, a dangerous, rule, the mood of a question 'Ci. Epid. 414; M. G. 1277; Most. 643; Poen. 986, 1156; Ps. 930; Rud. 1110; Trin. 87, 1145 ; and many others. 2 As. 419; Capt. 581, 937; Epid. 384; Men. 304, 636; M. G. 768; Most. 992; Rud. 338, 393; St. 376; Trin. 14, 129, 135, 335, 336, 561, 354, 653. 174 Tenney Frank is classed according to the nature of the answer. By this rule, the first question becomes a "volitive," the second a "potential," the third a "prescriptive." Is this not reading too much into the mood of faciam? To me the verb seems in each case to stand in the subjunctive of "deliberation." In (b) the qui, "how," raises an additional question of means, and in (c) the cur, "why," raises an additional question of propriety. It is the qui and the cur that call for potest and oportet. faciam is innocent. One more point needs emphasis. Even in such combinations as neque unde auxilium expetam habeo it must not be thought that the entire feeling of incapacity comes from the questioning part (unde expetam f). The question, to be sure, is originally asked in a tone of helplessness, and therefore implies lack of means; but the new element in the complex sentence, non habeo, plainly asserts what the question only implied. The whole ex- pression is now one of incapacity, and it is unfair to say that the mood of the verb or the original question is the part that conveys the meaning. Even had these not performed the function, the new combination 1 could hardly have escaped doing so. As for the mood of the verb, in such cases it is to be explained as con- tinuing into hypotaxis the mood of the original question ; and that, as we have just indicated, is the same as that of other questions in the first person, like quid faciam? 3. There are some independent constructions in Latin, like quis putetf vix crederes, that express various ideas of capacity. Now, obviously our idiom would find its simplest explanation if, when we had resolved these into their paratactic elements, the dependent clause proved to be an ordinary independent usage of the subjunc- tive. I shall postpone the larger problem of a possible descent 1 It is impossible, of course, to draw the line where such dependent clauses cease to be reminiscent of the interrogative stage and begin to associate with other relative clauses. At such a time the two classes here discussed (the descendents of relative purpose clauses, and those of questions) must have fused. The quod expression in non habeo quod must have come from the former source, for it acquires its subordinating function from the qui-ciauses, and yet the early writers were confusing relatives and interrogatives to an extent that made fusion easy. Plautus may write : scit quod velit (Cist. 706), and Cato: quid volet faciat (R. R. 148). The persisting forms quidlubet and quidvis are witnesses, too. In Poen. 627, vium nescio qua veniat ad mare, a relative pronoun referring to viam serves to introduce a quoted " deliberation." The testimony is abundant. Semantics of Modal Constructions 175 from a proethnic "potential" to the third part of this paper, and here briefly indicate why I did not simply connect (non) habet quod det with such independent expressions of capacity as exist in Latin. Such certainly exist. I find it impossible with Elmer ' to annihilate them by translation. On the other hand, I am not fully convinced by Hale's 2 presentation that the verb in such ex- pressions stands in the "potential" subjunctive. I mean that the expressions as a whole convey this peculiar force, not the mood of the verb alone. Mr. Hale recognizes some eleven 3 distinct usages of this construction in Latin, and connects them with similar expressions in Greek as representatives of an Indo-European usage of the optative. This is not the place to discuss these in full. I can only indicate by one illustration the reasons for my skepticism. The method I shall apply to the Greek examples will hold in general for the Latin as well. Let us, then, examine the potential expression with vix and facile. These adverbs obviously have no place except in expressions of capacity. We speak of "difficulty " and "ease" only with reference to the capacity to act. So true is this that in Cicero* about one-half of the occurrences of these adverbs modify the verb posse, and the rest contain an implication of the idea that belongs to posse, whatever the verb or the mood of that verb may be. English says almost invariably: "he could hardly do it." Now, be the mood of the verb what it may, if vix or facile be present, the verb, since the verb is the part that expresses the action of the sentence, will seem imbued with the potential idea. If the verb is, e. g., a characterizing subjunctive, as in Caes. B. O. 1. 6. 1, "erant omnino itinera duo quibus itineribus domo exire possent ; unum per Sequanos, angustum et difficile, vix qua singuli carri ducerentur," it seems natural to call it a "poten- tial" subjunctive. By the same token, one would find a " potential" clause of fear in "metuo ne vix sibi constet^ "they can hardly be consistent" (Acad. 132); a potential result clause in "tempes- tates coortae sunt maximae ut non modo proficisci non posset sed i Cornell Studies VI. 2 Trans. A. P. A. 1900. s Ibid., pp. 155-60. The statement in the Hale-Buck Grammar, sec. 516, is more conservative. * Note, too, the regular presence of posse in purpose clauses with quo facilius whereas it seldom is found with quo maturius, quo libentius, etc. 176 Tenney Frank vix in oppido consisteret," "they could hardly hold the port" (Verr. i. 46); a "potential" imperfect indicative in "vix risum tenebant," "they could hardly restrain their laughter" ( Vat. 20) ; and so on through the moods and tenses. Is it not true that in such cases the moods and tenses have their raison (V&tre regard- less of the "potential" tone, and that this is only an adumbration of vix and the general tone of the sentence? The potential idea, then, seems to belong to the whole expression, 1 and has in no case become associated with the mood of the verb and that alone. This is clear from the fact that the verb when standing alone in the subjunctive is never interpreted in the potential sense. A vix, forsitan, facile, or a peculiar antecedent sentence must be near by. We therefore have no independent "potential" usage of the subjunctive to form the nucleus of our complex phrase. At best we should have to assume as nucleus some stereotyped expression the combined parts of which might perform the function. However, the processes of subordination do not employ such methods, and obviously that is not the constitu- tion of this idiom. On the other hand, we must admit that we cannot entirely disprove the older hypothesis. It may be that at some time the mood alone without any accessories could do service as a potential expression, and that the usage finally withered away into a few phrases as they appear in historical times. We can only say that the process sketched in the foregoing seems more in conformity with the behavior of modal usages as we know them. III. We now come to our third task; i. e., to find what light a comparative study may throw upon the provenance and history of this construction. 1 1 have not space to analyze every Latin idiom in order to prove the truth of this. The analysis of the Greek idioms will suffice for the Latin ones, since they are of the same nature. The general principle underlying the statement made above has been ex- pressed by Morris Aims and Methods, pp. 96 ff ., in a better manner than I find myself able to do. As for quis putet, I would suggest that it was given its function and kept alive by force of quis est qui putet. The shorter phrase may have begun as a hypothetical expres- sion, but in Classical Latin it is usually neither hypothetical nor purely potential in tone. It is rather a brief equivalent of quis est qui putet, which expresses actuality as all normal consecutive clauses do. Semantics of Modal Constructions 177 1. There are in Greek several sentences like e/Mol s av tTre.iT' 'OSvotJos eyia deioio \adoifx.rjv; "How could I possibly forget the princely Odysseus?" b) Od. iii. 114, ti's k«v ixetva wdvra ye fi,v$r}S)Ta ouaxreias . . . . | iirei %apTrrj- Sov' a/xa $ctvov kcu iralpov KaAAwres; " How could you save a lesser man, | when you deserted Sarpedon?" Also Od. viii. 352; ix. 351. (2) Od. xviii. 31. irSs 8' av vvaripio dvSpl pA)(oio; "How could you fight one younger than yourself? " Also II. xvii. 325; xix. 81. (3) Od. i. 65. 7Tuis av In-eii-' 'OSvotJos XaOolp^v; "How could I ever for- get him? " Also II. x. 243; ix. 437. The sentences teem with suggestions of impossibility, but the "potential" mechanism does not lie in the mood of the verb. It lies in 7rco?, in the nature of the condition, in the order of the words, in the sentence stress, in particles like hreira, 'yap, ye, that lie scattered about to prod the reader's memory of these things. The very architecture of the phrases does its part. Mark how the tone of improbability rises with each syllable in 7rw? av eyd> ae Seoi/ji,i; Od. viii. 352. What, then, is the function of the opta- tive in such questions ? Since impossible situations are naturally expressed by the mood of an "ideal" condition, the optative is the mood that most frequently occiirs with such questions. The verb simply stands in the mood of hypothetical expressions. In fact, similar expressions with the indicative and subjunctive, though these are naturally rarer for the reason I have just given, may readily convey a potential idea, if the rest of the mechanism is present. Note the effect of tt&s with a dubitative subjunctive in Od. xvi. 70, trw yap Brj rbv %elvov ev vTro&egofiat, oIkw; (Palmer translates: "How can I take a stranger home?") and with an indicative in Od. xix. 325, trw yap i/usv av, %elve, Satjcreai, ei . . . . 180 Tenney Frank (II. i. 150), "How could you think me better than other women — if you should sit unkempt at table?" Such cases are familiar enough throughout Greek, yet no one has offered to supply an intricate classification of "potential future indicatives," "potential dubitative subjunctives," and the like. 1 Sanskrit shows even fewer examples than Greek of independent expressions that might be styled "pure potentials." However, kathd (katham), the cognate and functional equivalent of 7rpovr]pa rts Xeyot, Kal yvvauv p€v 7ravro\/xovs epcoras ara«7t (tvvvo/aovs fipoTu>v; "Who could tell of the overbold purpose of man, or the all-daring infatuation of wretched woman?" The verb might have been in an infinitive form. There are some things one knows to be impossible "once he has come to Forty Years." c) The above given questions are in an affirmative form with a negative implication. At times a less rhetorical form is em- ployed; i. e., a simple negative statement; but the conditions are still of the same nature, and the verb is put in the mood of unreality to suit the situation. The occurrences in the Odyssey are iv. 78, "No mortal man would (might) vie with Zeus;" iii. 112; ix. 242; xii. 77, 88, 107; xiii. 87; xx. 392. Such state- ments are found throughout Greek. As for Sanskrit, I believe the Rigveda will fail to produce any examples, and in the remain- ing early writings they will be found very rarely if at all. d) These negative statements and questions implying a nega- tion constitute the greater mass of sentences that might be called upon to support a theory of a "pure potential" optative in Greek. There are also a few occurrences in the affirmative in which the expression as a whole produces a similar effect. Several of these are of the vix facile type which has already been discussed. What was then said will also hold for these. In Od. vi. 300, pela 8' apiyvmr' 1 ecrri, Kal av 7rat? i)yr\aano vijTrios, "It can easily be recognized, even a small boy might guide you to it," both clauses convey a tone of capacity ; the optative f)yrjiKoiTO, xii. 84; ov Be' Kev a/xfiair), xii. 77; and the sentence often quoted as an example of the potential, icai Kev BioicrTevcreia<; , xii. 102, "he might reach it with an arrow." That is, the effort implied in the Bid, ek, avd colors the whole expression into one of capacity, while the compound verb practically serves as an apodosis to a condition of which the verb in simple form would be the predi- cate, thus el 6ira 6a\do-o-rjS (cf. Od. v. 142; xvii. 146; ix. 126). II. V. 192, tmroi 8' ov irap&un /cat appara, tu>v k' imfSairjv; cf. xiv. 299; V. 483, ov rL pjoi ivOdSt rotov otov k' r/i epoicv 'A^atoi iq Ktv dyouv. I find only one example of this type in Homer with an anteced- ent in the affirmative. The sentence is quite loosely constructed : II. X. 171, etcrtv ft,fv /xot iratSes d/ni^noves, eto"t 8c Xaot Kal iroXees, tS>v k€v tls cirot^o/tevos KaXeo-eitv. A longer list of examples from Greek tragedy may be found in Mr. Hale's discussion of " Deliberates " (Tr. A. P. A. 1893, pp. 192, 193), in which paper reference is made to these Homeric examples. The usage is found throughout Greek literature and shows such a fixed and withal peculiar form that it may fairly be treated as idiomatic. An explicit antecedent of the relative is more regularly given than in the corresponding Latin idiom, where simply {non) est or (non) habet is quite sufficient. It is also quite regularly a negative, whereas in Latin the affirmative is as frequent as the negative. Now, if Greek possessed a "pure potential" optative in inde- pendent usage in affirmative and negative, and if this were the base of our complex sentence, it would be strange if the dependent form should thus limit itself to the negative. The fact is, as we have seen, that Greek has no such independent usage of the optative mood. In the idioms discussed above (even granting that any of them deserves the name of idiom) there is no type that could serve as a simple base for this complex usage. Much less did they form a compound of similar idioms from which a "potential" Modusbegriff could have emanated to create the iSee "The Uses of the Optative in the Eddas," A. J. P. XXVII, p. 11. 184 Tenney Frank inspiration of a complex potential idiom. This idiom must there- fore tell its own history. I should interpret these sentences as they were formerly inter- preted, before the new names were invented. The verbs are in the mood of the "irrealis." The implication of (im)possibility and (in) capacity comes from the denial of possible or capable means or agents in the antecedent phrase ov Trapeao-i, ov /xoi eariv. The psychology 1 is the same as of the simpler types already dis- cussed. In fact, the mood of the dependent clause matters little, except that indicatives are usually too much impregnated with a precise and matter-of-fact tone to permit of such contamination; and yet this same tone betrays itself frequently even with the indicative, if the same mechanism is present. See the following, together with Jebb's (unprejudiced) translation: ov yap tis op/iios ioriv ovS' oirot irXiuyv eeejiMroA^crei xepSos y £evo>