My dissertation examines the effects of tense on evaluation times of nouns and other non-verbal predicates, which manifest in observable patterns. For instance, in (1), the noun ‘prisoners’ can either refer to current (adult) prisoners who were babies in the 1960s or infants incarcerated in the 1960s. Thus, while either a “present” or “past” interpretation of ‘prisoners’ is acceptable, there is a third logical possibility which is notably absent, namely one in which people who will be incarcerated in the future were babies in the 1960s.
(1) In the 1960s, the prisoners were babies. (Frazier 2019)
The previous work on nominal evaluation times concludes that there is no all-compassing systematic relationship between tense and the evaluation times of nouns (e.g. Enç 1981, Keshet 2008, Musan 1995, 1999, Tonhauser 2002, 2006, 2020). By disentangling several misleading empirical confounds (namely, the effects of lexical aspect and novelty), I demonstrate a definitive link between sentential tense and predicate evaluation times: the time arguments of all predicates are locally bound by the nearest c-commanding time abstraction. My dissertation offers a systematic formalism for predicting the evaluation times of nouns and other non-verbal predicates, comprised of three additions to formal theory:
(i) Nominal Lexical Aspect: I demonstrate that stage nouns naturally separate into two classes which exhibit different temporal behaviors (O’Leary 2017a, O’Leary & Brasoveanu 2018), akin to the well-known lexical aspect in verbal predicates. For instance, English speakers typically judge that, despite their similarities, (2) is an acceptable sentence, while (3) is a contradiction. Both are present tense sentences containing noun-predicate pairs which cannot hold of the same individual simultaneously, yet ‘fugitive’ allows a “past” interpretation where the nominal property time precedes the matrix tense and verbal event time, while ‘bachelor’ does not; instead, ‘bachelor’ requires that the nominal property time and the matrix tense time overlap, leading to an incompatible state of affairs in (3).
(2) A fugitive is in jail. (modified from Enç 1981)
(3) # A bachelor is married. (modified from Keshet 2008)
Under my classification, a noun like ‘bachelor’ must have an evaluation time that holds for an interval overlapping its time argument (discussed in (ii)). Thus in (3), the state of bachelor-hood must overlap the state of being married. Meanwhile, a noun like ‘fugitive’ holds for an interval overlapping or preceding its time argument, and therefore the state of being a fugitive may hold prior to being in jail. Recognizing this previously unnoticed distinction resolves a number of problematic data points that previously impeded the identification of systematic relationships between nominal evaluation times and tense.
(ii) Syntactic Locality Constraint: I propose that the time argument of any predicate is locally bound by the nearest time abstraction (O’Leary 2021). Since verbal predicates, adverbs, and predicates acting as the main predicate of an utterance cannot undergo movement, and therefore must be interpreted in situ, they will be locally bound by the tense of their clause, as noted in Percus (2000). I propose that this locality constraint applies to the time arguments of all predicates. This fact has been difficult to observe because noun phrases and their modifiers can be affected by raising and reconstruction effects, making the location of the predicate in the LF less predictable. In a single clause utterance, there are two potential input times for any given noun or NP modifier – the utterance time and the time defined by the tense. For instance, in the sentence “Many professors were in kindergarten in the 80s,” the noun “professors” can describe the profession of the subjects at the current time or at the time of being in kindergarten (however unlikely). When raising is blocked for any reason, as with the post-copular noun in an existential “there” construction (Heim 1987, Poole 2017), the predicate’s time argument may only be bound by the tense—the nearest time abstraction. Therefore, in “There were professors in kindergarten in the 80s,” there is only one reading available, namely the one in which the subjects are simultaneously professors and in kindergarten.
(iii) Discourse Anaphora: Lastly, (i) and (ii) only apply to predicates novel to a discourse, as any subsequent use of the predicate is anaphoric, rather than descriptive. In a novel use of a predicate, the time argument is used to pick out the correct group of entities to which the speaker is referring, but any non-novel use of a non-verbal predicate (for instance, with a definite determiner) is merely anaphoric to a previously established set of entities, and therefore does not need to consider a time argument in order to denote the correct referents. This explains why (5), but not (4), is acceptable out of the blue.
(4) # A bachelor is married.
(5) My best friend in college was a bachelor. I saw him again recently. The bachelor is married.
In my dissertation, I build on Tonhauser’s (2006) dynamic semantic account of nominal evaluation times in definite determiner phrases, extending the analysis to other non-verbal predicates and arguing that the effects are tied to novelty rather than definiteness.
With the nominal lexical aspect classes in (i), the locality constraint in (ii), and the anaphoric nature of non-novel predicates proposed in (iii), the set of acceptable evaluation times for any given predicate becomes predictable. This establishes a newfound coherence between the temporal behavior of many different types of predicates. Additionally, these theoretical additions provide novel tools for diagnosing tense effects, lexical aspect, discourse anaphora, and the location of predicative material within the LF (which can help to elucidate matters such as covert raising and reconstruction effects).