Recent Posts

The Bertrand Russell Show

Feminist Philosophers

fragments of consciousness

Gender, Race and Philosophy: The Blog

Knowability

Leiter Reports: A Philosophy Blog

Long Words Bother Me

semantics etc. highlights

Thoughts Arguments and Rants

Nostalgia

Nostalgia

Sunday, October 08, 2006

Tense and Relative Clause Sentences II

Sentences like:

(1) John met a man who was a cyclist (King 2003)

pose a problem for most standard theories of tense. (1) has three readings, which can be paraphrased as follows:

(1a) John met a man who was a cyclist before John's meeting him
(1b) John met a man who was a cyclist at the time of the meeting
(1c) John met a man who was a cyclist after John's meeting him

Every theory of tense predicts that (1a) is a reading of (1). For every theory of tense takes past under past to mean anteriority with respect to some past reference time. The problem is how to account for (1b) and (1c).

Toshiyuki Ogihara has made a number of interesting suggestions about how to account for the tenses in relative clause sentences (most of his papers are available on his website).

First, Ogihara suggests that English has an SOT rule (sequence-of-tense rule), which allows for optional deletion of embedded past-tense morphemes. This is what is going on in:

(2) John said he would buy a fish that was still alive (Ogihara 1989)

The past tense morpheme in the relative clause is deleted. So, (2) requires for its truth that John said that he would buy a fish that was still alive at the time of the buying event.

Interestingly, a similar suggestion was made by the Danish grammarian Otto Jespersen (1860-1943) (in A Modern English Grammar on Historical Principles). Jespersen suggests that unlike certain other languages (e.g. Japanese) English has a rule that allows past-tense morphemes to be deleted when they occur under another past-tense morpheme.

Second, Ogihara suggests that the (1c) reading where the time of the cycling is later than the time of the meeting can be gotten via (optional) quantifier raising. "A man who was a cyclist" raises to a wide-scope position. On this reading, the matrix tense becomes the "embedded" tense. So, on this reading, the time of the cycling is naturally interpreted as being later than the time of the meeting.

Unfortunately, this simple story cannot be the whole story about tenses in relative clause sentences. For consider:

(3) I will marry a man who went to Harvard (from Partee)

(3) has a reading that does not require for its truth that I marry a man who went to Harvard prior to the speech time. It only requires that I marry a man who went to Harvard prior to the time of the marriage. Since 'a man' is not the surface-grammatical subject of the sentence, there is clearly quantifier raising. But if "a man who went to Harvard" raises to a wide scope position, then (3) does require for its truth that a man I will marry went to Harvard prior to the speech time (as in "a man x who went to Harvard is such that I will marry x").

Another problem with the wide scope reading is that (3) could be true if the speaker marries a man who is not born yet (suppose the speaker is 3 years old). So, we do not want the existential quantifier to have wide scope. Given Priorean tense logic the correct reading of (3) is:

(3a) F(Ex(man x & I marry x & P(x goes to Harvard)))

In English: it will be the case that, there is someone who is a man and who marries me and who went to Harvard.

What to do? Here is my working hypothesis ("working hypothesis" because I am still struggling with the syntax).

The matrix tense can either take wide scope or narrow scope with respect to the existential quantifier. If the matrix tense takes wide scope, then the relative clause tense is interpreted as past (future) relative to the matrix tense (and there is (optional) deletion of the embedded past tense). If the matrix tense takes narrow scope, then the relative clause tense is intepreted as past (future) relative to the time of speech. Moreover, if the matrix tense takes narrow scope, then it can take wide or narrow scope with respect to the predicate restricting the existential quantifier.

This working hypothesis gives us rather nice results in the case of:

(4) John was talking to a student who will run for president

My hypothesis predicts that (4) has the following three readings:

P(Ex(student x & John talks to x & F(runs x)))
Ex(P(student x & John talks to x) & F(runs x))
Ex(student x & P(John talks to x) & F(runs x))

In English:
It was the case that, there is someone who is a student and who is talked to by John and who will run for president.
There is someone who was a student and who was talked to by John and who will run for president.
There is someone who is a student and who was talked to by John and who will run for president.

The first reading seems unavailable. But notice that it is unavailable only if we allow that "will" can introduce a time that is earlier than the time of speech. It has been suggested in the linguistic literature that since English features two future modals, namely "would" (as in "It was the case that John would run for president) and "will" (as in "It was the case that John will run for president"), "will" always introduces a time that is future relative to the time of speech (the exception being special narrator contexts).

Also, my account does not straightforwardly predict that (1c) is a reading of (1). What it does predict is the following reading:

Ex(P(man x & John meets x) & P(cyclist x))

This reading is compatible with "John met a man who was a cyclist after John's meeting him" but it does not entail it. So, we should expect the "cyclist-after-meeting" reading to be triggered only by the presence of temporal adverbials and/or other cues in the surrounding discourse context. And that is entirely consistent with the data. For, as Ogihara reports, speakers need a little bit of help for the "cyclist-after-meeting" reading to be salient.

No comments: