H. P. Grice teaches Latin at Rossall

Consider Varro. In “De lingua Romana,” he was the first to exploit the two axes, aspectual and temporal, which appear in the Grecian texts. His presentation of the verbal system is based on an aspectual binary opposition, “infectum”/“perfectum” (cf. 10.48), built on the semantic opposition between the unachieved (“infectum”) and the achieved (“perfectum”). Moreover, this presentation is based on the properly temporal tripartite classification “praeteritum/praesens/futurum” (in that order, that is, along an axis oriented from the past toward the future), in which each term is a participle of the tense it is supposed to represent (“praeteritum,” past participle of “praeterire,” “pass before, flow away”; “praesens,” present participle of “praeesse,” “to be at the head of,” whence “to be there personally, to attend”; “futurum,” future participle of esse, “to be”). Varro suggests (but does not say explicitly) that the same tripartite temporal classification can be found in the “infectum” and the “perfectum,” which would give us a series with six terms. The “infectum” would have a present, a preterite, and a future, corresponding in our nomenclature to the present, imperfect, and future. The “perfectum” would have once again a present (see XII.B for the problems raised by such an analysis from the interpretive point of view), a preterite, and a future, corresponding to our perfect, pluperfect, and future anterior. The forms that appear along these two axes entertain relationships of engenderment. The point of origin is the present. “Lego,” “I read,” engenders both the two other temporal forms of the “infectum” and the form of the “perfectum,” “Legi,” “I have read,” which necessarily governs in turn the two other forms of the “perfectum.” This relationship of engenderment substitutes one order for another. In a first phase, the temporal axis is presented as a realistic progression (from the past to the future), and the aspectual opposition is constructed according to bi-polar opposition in which the unachieved seems to be the marked pole in relation to the achieved. On the other hand, engenderment makes the present the source of the past and the future, and it makes the unachieved the source of the achieved, the unmarked pole. This substitution in the form of an inversion is clearly of morphological origin: in the Roman verbal system the form called “present” is unmarked morphologically, which makes the others appear to proceed from it by the simple adjunction of temporal morphemes, just as the form of the “infectum” is generally (and in any case in living formations at the historical period) unmarked with respect to the form of the perfectum (on the way this perfectum is formed, see XII.B). The form that we would call the present, which is the least marked morphologically, is thus presented as engendering all the others, in accord with a principle (explicit in Varro) of the development of linguistic forms by branching out from a root form. At the end of antiquity, Priscian (sixth century), in book VIII of his “Institutiones grammaticae,” preserves the principle of this engenderment: his whole presentation of verb tenses is based on the model of cognatio (family relationship), but with notable differences. Like the whole Roman academic tradition preceding him, Priscian retains (“Grammatici Romani,” 2.414.9–418.21) only one temporal series, with five tenses: three fundamental tenses, the present, past, and future, and a division of the past into three tenses, imperfectum, perfectum, and plusquam perfectum (imperfect, perfect, and pluperfect). This does not mean that Priscian is unaware of or rejects the opposition between the unachieved and the achieved, but under the influence of Grecian classifications, he uses it to connect tenses with one another, not to account for the fundamental morphological opposition of the Roman verb. The mode of engenderment Priscian describes thus proceeds in accord with principle of continuity: the present includes partly the past and partly the future; the past contained by the present corresponds, if it remains unachieved, to the imperfect, but becomes a perfect if it is achieved, and a pluperfect if this achievement is distant. This effect of continuity allows Priscian to preserve the image of engenderment on the basis of the present, and to indicate a realist legitimation: as soon as the present includes a past element and a future element, it contains, as it were, the embryo of the past and the future. In addition to this family relationship that we could call paradigmatic, Roman texts attest to the existence of a syntagmatic type of family relationship, the cunjunctio temporum, or “marriage of times.” This is a study of the organization of tenses with respect to one another when there are two verbal forms in an utterance. It was in the work of Diomedes, a fourth-century grammarian/compiler (who was undoubtedly drawing on an earlier author), that this study appeared in the most developed form (Grammatici Romani, 1.388.11– 395.10). This text, and the parallel passages we find in Charisius, another grammarian/ compiler of the same period, cannot be creations ex nihilo. But it remains that the grammatical tradition abandoned this kind of effort. In fact, properly grammatical analysis was recentered on the isolated minimal utterance that constitutes for grammarians the privileged analytical framework. Then the only original studies on temporal relationships, which Diomedes echoes, were these isolated and largely fragmentary “fossil” texts. Consequently, we see what the four models arrive at: emptying the aspectual problematics of what nonetheless constitutes it, namely, the process, in its accomplishment. A paradoxical effect, when we have said everything except what we were supposed to say. On the empirical level, this leads to theories of aspect abandoning, in just as paradoxical a way, a question that is nonetheless situated at the heart of the problem of aspect, namely, the opposition between process and property. None of the four models truly succeeds in rendering the distinction. Processes and properties can both cover more or less limited intervals, even if there are probably more temporally limited properties than unlimited processes. Properties are by definition qualitatively homogeneous, but processes may be. Properties may even begin, situating themselves in a gradual sequence. Although a point of view is necessary to posit a property, the use of that point of view may involve any kind of process, including procedures, particularly when the latter are the object of a description. The grammatical tradition encounters this question as soon as, seeking to characterize the category of verbs, it repeats that verbs designate actions, and then has to add that they also sometimes designate states and properties. This shows that the difficulty is central: what these aspectual models and traditional conceptions of the verb lack is a theory of what the referent of a verb can be. Thus they lack precisely what Aristotle set about to elaborate, a theory of what he called “movement,” of what in this movement exceeds both time and being: we lack a physics, in which what happens is distinguished from what is true. Thus to deal with the question of aspect we need something that linguists are not, a priori, competent to provide, and which is not their subject: philosophy. Aspect is an subject that is irremediably philosophical. What happens has to be distinguished from what is true. Because languages distinguish the two, and because this distinction is one of the central dimensions of aspect, we must recognize that processes have a qualitative dimension that makes it possible to distinguish them qualitatively from one another, probably a temporal extension, but also a qualitative extension: verbs refer to something that takes place, or does not take place, that is in time, but is not only time. This thing also has qualitative properties. But that is because it does not have only qualitative properties, but is also accomplishment, because the question of aspect exists and develops.