Review of "Wolfgang Klein (1994) Time in Language, Routledge, London and New York".

Eric Hoekstra

Glot International 1.4, 14.

1. What the book is about
The book is an extremely convincing presentation of K's ideas about tense, aspect and the connection between these two.
The Reichenbachian analysis operates with three time parameters. 1. the time of utterance (S); 2. the time of whatever is expressed by the verb (E); 3. the time (point) of reference (R). Tenses are then defined in terms of elaborations of the precede relation and the overlap relation between these three time parameters.
K. presents empirical arguments against this type of analysis. An example is given below (p.22). The past tense can be defined on the Reichenbachian view basically as follows: the time of the event precedes the time of utterance (E < S), as in (1) I ate an apple (yesterday). K. points out a counterexample: (2) (They found John in the bathtub.) He was dead. Clearly, the speaker does not want to assert here that the time of John's being dead must precede the moment of utterance, since John will obviously continue to remain dead at and after the moment of utterance.
Another problem with the Reichenbachian view is that, conceptually, it has nothing to say about the relationship between tense and aspect, whereas it is well-known that several empirical observations testify to the existence of such a connection. A third problem with the Reichenbachian view is that no clear content is normally given to the third time parameter, the time of reference.
K's proposal, like the Reichenbachian analysis, distinguishes the time of the event (which K. calls time of situation) and the time of utterance. It differs from it with respect to the content it gives to the third parameter, and with respect to the relations which are claimed to obtain between the three parameters.
The third parameter is referred to as topic time. The topic time is the time for which the claim is made. Reconsider the example in (2). The total amount of time during which John is dead is absolutely irrelevant for the speaker's assertion. The only thing relevant is that at some point in the past, the topic time chosen by the speaker, John was dead. This examples illustrates K's analysis of tense. The time of the situation (or event) is not encoded by tense, contrary to what is normally assumed. Tense encodes a relation between the topic time and the time of utterance.
As K. notes (p.5) "it does not matter for tense whether the event, state or process is before, at, or after the time of utterance". This is made clear by the examples in (3): There was a book on the table. It was in Russian. The first sentence expresses a temporary property (stage level), the second a permanent property (individual level). Yet both are in the past tense.
Aspect, in contrast, is what encodes the relation between topic time and the time (or time properties) of the situation. The situation is described by the non-finite part of the sentence. Thus, aspect and tense are indirectly related since both rely on the notion of topic time. The perfect in English is treated as an example. Consider (4), from p.104: John has slept without a blanket. The perfect aspect (Tsit < TT) entails that John's sleeping without a blanket is before the time for which the assertion is made, while the present tense entails that the time for which the assertion is made includes the time of utterance. Aspectual distinctions are then worked out in further detail.

2. Contents and style
K. presents his own framework in a lucid Introduction (1-13). Problems with the traditional analysis of tense and aspect are pointed out in chapter 2 (14-35) Tense, aspect, and Aktionsart: the conventional picture. Topic time (3: 36-58) gives this notion further intuitive content, while pointing out problematic areas like the interaction with negation, mood (imperative), definiteness, and so on. Time Structure (4:59-71) presents some generally accepted properties of time (e.g. linear ordering), and discusses deixis (e.g. now) and anaphoric reference (e.g. at three o'clock). Inherent temporal features of the lexical content (5:72-98) features K’s lexical aspectual semantics. Aspect (6:99-119) deals with the proper definition of aspectual tenses like the perfect. Tense (7:120-141) speaks for itself, temporal and other adverbs are discussed in the following chapter Temporal adverbials and their meaning (8:142-158), and the relation between the two is worked out in The function of positional adverbials in the utterance (9:159-183) (which has nothing to do with locativity) and in Adverbials of duration and of frequency (10:184-214). Difficult areas of research are pointed out in the final chapter with the list title Non-declarative clauses, subordinate clauses, noun phrases (11:215-224). All notes are endnotes (225-234), and kept to an absolute minimum, in keeping with the general nature of the book. There is a good index, in which subjects and persons have been conveniently joined (a move which deserves wider popularity).
The book is an extremely well-written introduction to matters pertaining to tense and aspect, conveniently mixed with K’s own views. One cannot but agree with the blurb in which Professor Arnim Von Stechow is quoted (“I will certainly use it for my courses”).
K. has explicitly chosen to present what is known about tense and aspect from his own perspective. K. writes that he “feels uneasy to have passed over so much interesting work”. I don’t think that this is a problem since choices have to be made or else the book becomes too large. It is a problem, however, that occasionally it is not clear what piece of analysis belongs to K. and what piece belongs to others. For example, what is well-known and what is new, apart from terminology, about the division of situation descriptions into 0-state, 1-state and 2-state? Otherwise this would have been the ideal textbook. Nevertheless, it is an extremely instructive book. Exaggerating somewhat, I had not thought that a book on tense and aspect, a notoriously difficult field, could be so crystal clear.

3. Of generative interest
Of generative interest is the discussion on p.202-4 under the heading (10.3.2) English is verb-final. Consider the paradigm in (5): (a) Twice, one of our boy scouts caught a duck; (b) One of our boy scouts twice caught a duck; (c) One of our boy scouts caught a duck twice. In (5a-b), twice has scope over whatever is to the right of it. If we generalise that rule, then (5c) ought to be ungrammatical because twice does not have scope over anything. In order to maintain the rule, K. postulates a verbal trace to the right of twice, accounting for it that in (5c) twice has scope over the verb only. The verbal trace occupies the position in which infinitives are spelled out in German.
Also of generative interest is K’s view of modals in English (p.180). On standard assumptions these are directly inserted into INFL. The standard view, to my mind, immediately raises the question why Dutch and German can have modal infinitives at all, a question which should be faced (but normally is not) by postulating a minimalist theory of parametrisation of insertion possibilities. K. adopts the view that modals are inserted in V, also in English, since he semantically factors out tensedness (associated with INFL) from the non-tensed meaning of the verb (the description of the situation, associated with V and its projections). Interestingly, Marcel den Dikken and myself have presented a syntactic argument supporting the idea that English modals are inserted into V in our talk Parasitic Participles, presented at the 10th Workshop on Comparative Germanic Syntax (Brussels, 18-19/01/1995). Thus semantic and syntactic lines of research converge.

4. Concluding Remarks on Generalised Quantifier theory
K’s book is very well suited for the purpose of becoming acquainted with research into tense, aspect and related matters. Personally, I regretted that K. did not devote more attention to the presentation of insights from the theory of Generalised Quantifiers. I will give two examples.
First, K. mentions in the final chapter in passing the fact that nouns can also affect aspectual interpretation, but does not go into the questions this raises for a proper theory of aspect. The implication of this fact is that a theory of aspect must be sufficiently abstract so as to be able to point out properties which verbs and nouns have in common. K.’s theory will not be able to do so, but Henk Verkuyl’s (Linguistics & Philosophy 12, 39-94 (1989)) Generalised Quantifier theory has been able to express some of these properties.
Second, GQ-theory can account for the correlation between patterns of inference and patterns of negative polarity, which also reveals itself in the domain of tense. Simplifying somewhat, we can say that only downward entailing expressions license negative polarity items. This correlation can be illustrated with the temporal complementiser voordat “before”:

(6) a. Voordat Jan gedroomd had (was het drie uur) ===> , <=/=
before Jan dreamed had (was it three o’clock)
“Before Jan had dreamed, it was three o’clock.”
b. Voordat Jan onrustig gedroomd had (was het drie uur)
before Jan restlessly dreamed had (was it three o’clock)
“Before Jan had dreamed restlessly, it was three o’clock.”

The inference from (6a) to (6b) is valid, but not the other way around. The complementiser voordat “before” is downward entailing: truth is preserved under subsets (from dreaming to dreaming restlessly). Items having this property license negative polarity, as shown below:

(7) Voordat er ook maar iemand gedroomd had was het drie uur.
before there anybody dreamed had was it three o’clock
“Before anybody had dreamed, it was three o’clock.”

Conversely, the temporal expression nadat “after” does not exhibit the pattern of inference shown in (6), and, correspondingly, it cannot trigger the negative polarity item ook maar iemand “anybody”. This is an example of how the formal semantics of temporal operators can be gleaned from relatively simply observable phenomena like patterns of inference and negative polarity.
Thus, what Time in Language lacks, perhaps, are the first steps towards formalisation of the relevant insights, and such steps towards a theory are necessary if the field of research into the nature of aspect and tense is to gain the coherence which derives from having a shared set of assumptions and a shared technical vocabulary. For the rest, Time and Language is a perfect book for getting acquainted with the main facts and issues pertaining to this field of research.