The Contextors Tense Conjugator is a tool for changing the tense of English sentences. In what follows we discuss some of the linguistic assumptions we have made in constructing its first version, some of its shortcomings, and elements of its underlying technology.
In preparing the first version of the Tense Conjugator we have adopted the traditional grammar analysis of tenses in English, namely that there are three: past, present, and future. To every sentence in one of the three tenses, the Tense Conjugator assigns two sentences as output, one in each of the remaining tenses, as demonstrated by the present tense sentence in (1) and the past tense sentence in (2).
(1) He knows many politicians.
(2) He didn’t earn much.
The traditional grammar analysis differs from the one that has been espoused by linguists for quite some time now, namely that there are only two tenses in English: past and present (see Huddleston 2002 for a detailed discussion).
Thus, while the sentence in (3) locates the event of her visiting him in the future, its grammatical tense is taken to be the present.
(3) She will visit him.
According to the more recent analysis, will is used to discuss the actualization of hypothetical states of affairs: it says that the hypothetical state of affairs under discussion is certain to actualize in the future. This meaning contribution is also made by would. In (4), the combination of would and visit him also gives the meaning that her visiting him is certain to actualize in the future, only now the future is not relative to the time of uttering the sentence but relative to the time of Kim’s realization:
(4) Kim realized that she would visit him.
The reason why we do not generate, for example, she would visit him as the past tense form of she will visit him is that when a would sentence is not embedded under a past tense verb (like in (4)), the reading having to do with the actualization of a hypothetical state of affairs is not the most natural one, and therefore the meaning relation with will is not apparent.
We follow Huddleston (2002) in distinguishing between two tense systems in English: primary tense in finite sentences is marked by the form of the first auxiliary in the sentence, or, if there is no auxiliary, by the form of the main lexical verb; secondary tense is marked by the presence or absence of auxiliary have followed by a verb in past participle form. The primary tense of finite sentences is either present or past; the secondary tense of a sentence is either perfect or non-perfect. We also follow Huddleston (2002) in assuming that the perfect designates temporal anteriority with respect to a time introduced by some element in the sentence. Consider the sentences in (5).
In (5a) the perfect designates that her visiting him occurred before the time designated by has, namely the time of the utterance; in (5b) it designates that her visiting him occurred prior to a past situation, namely the time designated by the adjunct by the time I got there; in (5c) the perfect designates that her visit stands in the relation of temporal anteriority to a future situation, namely our getting there. In preparing the first version of the Tense Conjugator we have decided to retain the secondary tense of the input sentences and conjugate them only in the three primary tenses we posit: past, present and future; an example is given in (6).
(6) She will have been solving the problem.
A third assumption we make following Huddleston (2002) is that the aspect of English sentences can be either progressive or non-progressive, progressive aspect being marked by the presence of be followed by a verb in gerund participle form. Progressive aspect designates that the duration of the situation described by the verb includes the time the sentence refers to, which may be given explicitly by an adjunct in the sentence. In (7), for example, the duration of her reading the book includes the time of our entering the room.
(7) When we entered the room she was reading a book.
(6) above is an example of a conjugation of a sentence in the progressive; (8) and (9) below are examples of conjugations of non-progressive sentences.
(8) They loved their dog.
(9) Kim ate the cake.
Note that the present tense conjugation of (9), Kim eats the cake, is odd: it means that Kim habitually eats the cake. This is the case more generally with sentences designating accomplishments, events that have a completion point (in the case of eating a cake, the event is completed when there is no cake left). In this respect, conjugating Kim ate the cake in the present as Kim eats the cake is inaccurate: the past tense sentence describes a non-habitual activity, whereas the present tense sentence describes a habitual one. But in the current version of the Tense Conjugator we have decided to change only the primary tense of the input sentence, and just like the secondary tense of the input is kept constant, so is its aspect.
We currently do not provide tense conjugations for modal sentences. The reason is that there isn’t a uniform way to conjugate a modal sentence. Consider the sentences in (10)-(11).
It seems that sentences (10a-d) can be used to convey an ability of Kim’s to climb this mountain in the present, future and past; (10e), on the other hand, seems to be less appropriate in conveying this content, but is perfectly fine when, for example, epistemic modality (it is possible that Kim has climbed the mountain) is involved, and in this respect contrasts sharply with be able. So there is more than one way to conjugate a modal sentence in the different tenses, and what the proper way is depends on the kind of modality involved. The sentences in (11) emphasize this point. The first two sentences can only convey Kim’s lack of choice with respect to climbing this mountain, in the present (11a) or in the future (11b). (11c), on the other hand, seems only to convey that according to what is known, it must be the case that Kim has climbed this mountain. Finally, (11d) seems to convey either Kim’s lack of choice or what must be the case according to what is known.
The Tense Conjugator is based on the Contextors parser, which assigns to sentences of English constituent structure trees where the nodes are annotated with syntactic category and grammatical function labels. We will discuss the sentence in (14) and its tense conjugations in order to demonstrate the interaction between the Tense Conjugator and the parser.
(14) They will most likely be reading the book.
In order to identify the primary tense, secondary tense and aspect of the input sentence the conjugator must know whether, for example, the be in the sentence is the progressive auxiliary, a copula or the passive auxiliary. This information is provided by the parser, which chooses the lexical item that matches the syntactic environment of the occurrence of be in the sentence.
In order to use the information provided by the parser (which includes the identification of be in the sentence as the progressive auxiliary), the Tense Conjugator inspects the sequence of verbs will be reading and retrieve the properties of each one of the verbs. In addition to the information whether a given form is an auxiliary (and if so, what kind of auxiliary), it also retrieves its tense, person, number and polarity (positive or negative). The latter is involved, for example, in switching doesn’t in (2) above to won’t and didn’t.
In addition to retrieving nodes and their properties, the Tense Conjugator must be able to change the morpho-syntactic properties of words so that they agree with whichever other node in the output sentence they should agree with. In changing (14) from future tense to either present or past tense, the progressive auxiliary be, which in the input sentence is in the infinitive, must agree with the subject they.
One additional thing that the Tense Conjugator must be able to do is to change the position of a node when conjugating a sentence. In (14), the progressive auxiliary be follows the adverb phrase most likely, but in the output sentences it must be moved to a position preceding this adverb phrase.
In conclusion, in this article we demonstrated that in changing the tense of English sentences, the Tense Conjugator relies on constituent structure as well as the ability to retrieve and modify the morpho-syntactic properties of verbs and noun phrases. We have also discussed some of the (less than ideal) choices we had to make regarding the analysis of tense and aspect in English.