This method of classification was no doubt suggested in the first instance by the characteristics of the Slavonic verb system. In this system a clear distinction is drawn in nearly all verbs between those which express a process (durative verbs) and those which express a completed action (perfective verbs). When perfective and durat1ve verbs are formed from the same root, the perfective are distinguished from the durative forms (a) by having a preposition prefixed, or (b) by having a different stem formation. Thus in the Old Bulgarian (Old Ecclesiastical Slavonic) to» strike (hit) and to strike dead are expressed by the same verb, but in the latter meaning a preposition is found which does not appear in the former. btti (infinitive), “to strike”; n-biti, “to strike dead.” To strzke is durative, to strike dead is perfective. As an example of difference of stem formation expressing this difference of meaning, we may quote sésti, “to sit down” (perfective), sédtéz. “to s1t ” (durative). Verbs with a suffix in -n- have often a perfective meaning: cf. the Sanskrit and Greek verbs quoted above. The perfective verbs correspond in meaning to the Greek aorist, and are to be carefully distinguished from perfect forms. The same distinction of meaning is often achieved in other languages also by means of prepositions, e g. in Latin (Seneca, Epp xcui. 1o), Qu1d antem ad rem pertinet, qnamdiu vites, quad evitare non possis? “ What does it matter how long you go on avoiding [durative] what you cannot escape [perfective].” From this example, however, it is clear that, though the means employed to make the distinction are different, there is no difference in meaning between such perfective verbs and those classified by Delbruck as terminative. Here, as in many other parts of this study, the ideas are new, and grammatical terminology has not yet sufficiently crystallized, and still leaves something to be desired both in clearness and in precision. As regards the moods, the difficulty has been to find any criterion whereby the functions of one mood should be differentiated from those of the others. It has long been recognized that the difference between indicative and subjunctive is one of meaning and not one of formation; that, ag, in Sanskrit bharati (3rd sing. pres. indic), “ bears, ” is morphologically identical with hanati, “may slay ” (3rd sing. pres. subj), and that the latter is described as a subjunctive only because of the meaning, and because there exists a dis syllabic form, hanti, which makes the indicative “ slays.” Similarly in Greek it is impossible to distinguish morphologically between iraiiaw, “I shall check” (fut. indic.) and irai/aw, “let me check” (1st aor. subj). Moreover, in the earliest forms of the languages which preserve the moods best (Greek and Sanskrit), the connexion syntactically between the indicative and the subjunctive forms is closest Not only does the future express futurity, but also the determination of the subject to carry out the action expressed, which, in Delbruck's discussion of the moods, is precisely the point chosen as characteristic of the subjunctive. On the other hand, the present optative differs from the present (and future) indicative and present subjunctive in having a special mood suliix, and in having secondary while they have primary personal endings. Nevertheless its meaning overlaps that of the other forms, and some excellent authorities, like Professor W. W. Goodwin, see in future indicative, subjunctive and optative only different degrees of remoteness in the future, the remoteness being least in the future and greatest in the optative. Delbruck, however, abides, with slight modifications, by the distinction which he propounded in 1871 that the subjunctive expresses Will and the optative Wish. Here again the problem has not been solved, and it is doubtful how far any definite solution is likely to be arrived at, since there are so many gaps in our knowledge of mood forms. These gaps, owing to the break-up of the system at so early a period, it is hardly probable we shall ever be able to fill It is possible, however, to do a great deal more than has yet been done even in the most familiar languages In Latin, for instance, even now, the facts for the uses of the moods within the two centuries of the classical period are very imperfectly known, and it is no exaggeration to say that more has been done in the last hundred years for Sanskrit than has been done in two thousand years of continuous study for Latin or Greek. A still later addition to the domain of Philology-the study of meaning-presents fewer difficulties, but until recent years has been equally neglected. The study is so recent that the literature of the subject is still extremely small. The only attempts to deal with it on a large scale are M. Bréal's Essai de Sémantiqne (1897), now translated into English under the title of Semantics (1900), with a valuable introduction and appendix by Dr Postgate, and M. de la Grasserie's Essai d'une Sémantzque integrate (1908), a work which deserves mention for its attempt to make a thorough classification and a corresponding terminology for semantic phenomena, but the value of which is much diminished by hasty compilation and imperfect knowledge of many of the languages quoted. From the practical point of view many of the phenomena have been classified in works on rhetoric under the headings of Metaphor, Synecdoche and Metonymy. The psychological principle behind this superficial classification is that of association of ideas. Here, as elsewhere. changes proceed not by accident, but according to definite principles. Here, as elsewhere in language, in history, and the other moral sciences, the particular principle in operation can be ascertained only by beginning with the result and working back to the cause. In the development of meaning much more than in phonetics is this necessarily the case. In phonetics all speakers of the same dialect start with approximately the same sound. But the same combination of sounds which we call a word does not recall the same idea to all persons who use that word. The idea that the phrase railway station calls up in the mind of a Londoner is very different from that which occurs to the mind of a child acquainted only with a wayside station serving the wants of a country village of a few hundred inhabitants. The word herring suggests one idea or train of ideas to the fisherman who catches the fish, another to the merchant who purchases it from the fisherman, a third to the domestic who cooks it, and so on. To members of the same family the same word may often have widely different associations, and, if so, the metaphors for which the word will be employed will differ in each case.
For the history of meaning it is necessary to have regard to all the forms of association of ideas which psychology recognizes. These are contiguity in place or in time, resemblance and contrast. Contrast, however, as I S. Mill and Bain have shown, is not a simple form of association, but is evolved partly from contiguity, partly from resemblance. An artificial hollow generally implies also an artificial height made of the materials excavated from the hollow. Hence in most languages some words occur with the two contrasted meanings. Thus in English we find dyke in use both for a ditch and for a mound fronted by a ditch, the word ditch being, in fact, but a dialectal form of dyke. In Scotland, on the other hand, where earthen mounds and stone walls form more frequent boundaries between fields than in England, the word dyke is now practically limited to elevated boundaries, while ditch is limited to excavated boundaries. Thus the proverb, “February fill dyke, ” which in England implies that the February rains will fill the ditches, is often understood in Scotland to mean that in February the snow will be level with the tops of the stone or turf walls. Similarly in Latin Tacitus can say fossas prornere, which can only apply to levelling raised mounds; while in Greek Xenophon also talks of the ditch (trench) thrown up (rddapos al/aBeB);;.¢év17). It is only natural, therefore, that other words with several meanings should be used similarly' moat, originally a mound of earth or peat, has come to mean a big ditch, while, conversely, soldiers in trenches are not so much in ditches, as the word ought to signify, as behind breastworks. Sometimes, when two actions opposed to one another are contiguous, a word seems to change to the exact opposite of its original meaning. Thus the English verb wean, which meant originally to accustom (to cooked food), has been transferred to the necessary preliminary, to dis accustom to the breast.
Resemblances may be (i) genuine, and (a) of external appearance, or (b) of other characteristics; or (ii.) fanciful or analogical. From resemblance in the external appearance of the object, the word gem, which in Latin (gemma) usually means a bud, has
