Let us look at the history of the name before we look at the mischief which it, like many other names, has caused by making people believe that whenever there is a name there must be something behind it. The name was invented by Greek philosophers who, in their first attempts at classifying and giving names to the various forms of language, did not know whether to class such forms as γράφειν, γράψειν, γράψαι, γεγραφέναι, γράφεσθαι, γράψεσθαι, γέγραφθαι, γράψασθαι, γραφθῆναι, γραφθήσεσθαι, as nouns or as verbs. They had established for their own satisfaction the broad distinction between nouns (ὀνόματα) and verbs (ῥήματα); they had assigned to each a definition, but, after having done so, they found that forms like γράφειν would not fit their definition either of noun or verb.[19] What could they do? Some (the Stoics) represented the forms in ειν, etc., as a subdivision of the verb, and introduced for them the name ῥῆμα ἀπαρέμφατον or γενικώτατον. Others recognized them as a separate part of speech, raising their number from eight to nine or ten. Others, again, classed them under the adverb (ἐπιῤῥημα), as one of the eight recognized parts of speech. The Stoics, taking their stand on Aristotle’s definition of ῥῆμα, could not but regard the infinitive as ῥῆμα, because it implied time, past, present, or future, which was with them recognized as the specific characteristic of the verb (Zeitwort). But they went further, and called forms such as γράφειν, etc., ῥῆμα, in the highest or most general sense, distinguishing other verbal forms, such as γράφει, etc., by the names of κατηγόρημα or σύμβαμα. Afterwards, in the progress of grammatical science, the definition of ῥῆμα became more explicit and complete. It was pointed out that a verb, besides its predicative meaning (ἔμφασις), is able to[20] express several additional meanings (παρακολουθήματα or παρεμφάσεις), viz.: not only time, as already pointed out by Aristotle, but also person and number. The two latter meanings, however, being absent in γράφειν, this was now called ῥῆμα ἀπαρέμφατον (without by-meanings), or γενικώτατον, and, for practical purposes, this ῥῆμα ἀπαρέμφατον soon became the prototype of conjugation.
So far there was only confusion, arising from a want of precision in classifying the different forms of the verb. But when the Greek terminology was transplanted to Rome, real mischief began. Instead of ῥῆμα γενικώτατον, we now find the erroneous, or, at all events, inaccurate, translation, modus infinitus, and infinitivus by itself. What was originally meant as an adjective belonging to ῥῆμα, became a substantive, the infinitive, and though the question arose again and again what this infinitive really was, whether a noun, or a verb, or an adverb; whether a mood or not a mood; the real existence of such a thing as an infinitive could no longer be doubted. One can hardly trust one’s eyes in reading the extraordinary discussions on the nature of the infinitive in grammatical works of successive centuries up to the nineteenth. Suffice it to say that Gottfried Hermann, the great reformer of classical grammars, treated the infinitive again as an adverb, and, therefore, as a part of speech belonging to the particles. We ourselves were brought up to believe in infinitives; and to doubt the existence of this grammatical entity would have been considered in our younger days a most dangerous heresy.
And yet, how much confused thought, and how much controversy might have been avoided, if this grammatical term of infinitive had never been invented.[21,][C] The fact is that what we call infinitives are nothing more or less than cases of verbal nouns, and not till they are treated as what they are shall we ever gain an insight into the nature and the historical development of these grammatical monsters.
Take the old Homeric infinitive in μεναι, and you find its explanation in the Sanskrit termination mane, i.e. manai, the native of the suffix man (not, as others suppose, the locative of a suffix mana), by which a large number of nouns are formed in Sanskrit. From gnâ, to know, we have (g)nâman, Latin (g)nomén, that by which a thing is known, its name; from gan, to be born, gán-man, birth. In Greek this suffix man is chiefly used for forming masculine nouns, such as γνώ-μων, γνώ-μονος, literally a knower; τλή-μων, a sufferer; or as μην in ποι-μήν, a shepherd, literally a feeder. In Latin, on the contrary, men occurs frequently at the end of abstract nouns in the neuter gender, such as teg-men, the covering, or tegu-men or tegi-men; solamen, consolation; voca-men, an appellation; certa-men, a contest; and many more, particularly in ancient Latin; while in classical Latin the fuller suffix mentum predominates. If then we read in Homer, κύνας ἔτευξε δῶμα φυλασσέμεναι, we may call φυλασσέμεναι an infinitive, if we like, and translate “he made dogs to protect the house;” but the form which we have before us, is simply a dative of an old abstract noun in μεν, and the original meaning was “for the protection of the house,” or “for protecting the house;” as if we said in Latin, tutamini domum.
The infinitives in μεν may be corruptions of those in μεναι, unless we take μεν as an archaic accusative, which, though without analogy in Greek, would correspond to Latin accusatives like tegmen, and express the general object of certain acts or movements. In Sanskrit, at least in the Veda, infinitives in mane occur, such as dấ-mane, to give, Greek δό-μεναι; vid-máne, to know, Greek ϝίδ-μεναι.[22]
The question next arises, if this is a satisfactory explanation of the infinitives in μεναι, how are we to explain the infinitives in εναι? We find in Homer, not only ἴμεναι, to go, but also ἰέναι; not only ἔμμεναι, to be, but also εἶναι, i.e., ἔσ-εναι. Bopp simply says that the m is lost, but he brings no evidence that in Greek an m can thus be lost without any provocation. The real explanation, here, as elsewhere, is supplied by the Beieinander (the collateral growth), not by the Nacheinander (the successive growth) of language. Besides the suffix man, the Aryan languages possessed two other suffixes, van and an, which were added to verbal bases just like man. By the side of dâman, the act of giving, we find in the Veda dâ-van, the act of giving, and a dative dâ-váne, with the accent on the suffix, meaning for the giving, i.e. to give. Now in Greek this v would necessarily disappear, though its former presence might be indicated by the digamma æolicum. Thus, instead of Sanskrit dâváne, we should have in Greek δοϝέναι, δοέναι, and contracted δοῦναι, the regular form of the infinitive of the aorist, a form in which the diphthong ου would remain inexplicable, except for the former presence of the lost syllable ϝε. In the same manner εἶναι stands for ἐσ-ϝέναι, ἐσ-έναι, ἐέναι, εἶναι. Hence ἰέναι, stands for ἰϝέναι, and even the accent remains on the suffix van, just as it did in Sanskrit.
As the infinitives in μεναι were traced back to the suffix man, and those in ϝεναι to a suffix van, the regular infinitives in εναι after consonants, and ναι after vowels, must be referred to the suffix an, dat. ane. Here, too, we find analogous forms in the Veda. From dhûrv, to hurt, we have dhû́rv-aṇe, for the purpose of hurting, in order to hurt; in Rv. IX. 61, 30, we find vibhv-áne, Rv. VI. 61, 13, in order to conquer, and by the same suffix the Greeks formed their infinitives of the perfect, λελοιπ-έναι, and the infinitives of the verbs in μι, τιθέ-ναι, διδο-ναι, ἱστα-ναι, etc.
In order to explain, after these antecedents, the origin of the infinitive in ειν, as τύπτειν, we must admit either the shortening of ναι to νι, which is difficult; or the existence of a locative in ι by the side of a dative in αι. That the locative can take the place of the dative we see clearly in the Sanskrit forms of the aorist, parsháṇi, to cross, nesháṇi, to lead, which, as far as their form, not their origin, is concerned, would well match Greek forms like λύσειν in the future. In either case, τύπτε-νι in Greek would have become τύπτειν, just as τύπτε-σι became τύπτεις. In the Doric dialect this throwing back of the final ι is omitted in the second person singular, where the Dorians may say ἀμέλγες for ἀμέλγεις; and in the same Doric dialect the infinitive, too, occurs in εν instead of ειν; e.g., ἀείδεν instead of ἀείδειν. (Buttman, “Greek Gr.,” § 103, 10, 11.)
In this manner the growth of grammatical forms can be made as clear as the sequence of any historical events in the history of the world, nay, I should say far clearer, far more intelligible; and I should think that even the first learning of these grammatical forms might be somewhat seasoned and rendered more really instructive by allowing the pupil, from time to time, a glimpse into the past history of the Greek and Latin languages. In English what we call the infinitive is clearly a dative; to speak shows by its very preposition what it was intended for. How easy, then, to explain to a beginner that if he translates, “able to speak,” by ἱκανὸς εἰπεῖν, the Greek infinitive is really the same as the English, and that εἰπεῖν stands for εἴπενι and this for εἴπεναι, which, to a certain extent, answers the same purpose as the Greek ἔπει, the dative of ἔπος, and therefore originally ἔπεσι.
And remark, these very datives and locatives of nouns formed by the suffix ος in Greek, as in Sanskrit, es in Latin, though they yield no infinitives in Greek, yield the most common form of the infinitive in Latin, and may be traced also in Sanskrit. As from genus we form a dative generi, and a locative genere, which stands for genese, so from gigno an abstract noun would be formed, gignus, and from it a dative, gigneri, and a locative, gignere. I do not say that the intermediate form gignus existed in the spoken Latin, I only maintain that such a form would be analogous to gen-us, op-us, fœd-us, and that in Sanskrit the process is exactly the same. We form in Sanskrit a substantive càkshas, sight, càkshus, eye; and we find the dative of càkshas, i.e. càkshase, used as what we should call an infinitive, in order to see. But we also find another so-called infinitive, jîvàse, in order to live, although there is no noun, jîvas, life; we find áyase, to go, although there is no noun áyas, going. This Sanskrit áyase explains the Latin i-re, as *i-vane explained the Greek ἰέναι. The intention of the old framers of language is throughout the same. They differ only in the means which they use, one might almost say, at random; and the differences between Sanskrit, Greek, and Latin are often due to the simple fact that out of many possible forms that might be used and had been used before the Aryan languages became traditional, settled, and national, one family or clan or nation fancied one, another another. While this one became fixed and classical, all others became useless, remained perhaps here and there in proverbial sayings or in sacred songs, but were given up at last completely, as strange, obsolete, and unintelligible.