As some readers may recall from previous posts on philosophical methodology (here, for example), I am an enthusiast of what can be described as ‘empirically informed philosophy’. I believe that philosophical theories not only can but should yield empirical predictions, or in any case that they should at the very least be compatible with our current best empirical theories about certain phenomena – what John Protevi likes to describe as ‘empirically responsible’. This does not mean that all philosophical questions can or should be reduced to empirical questions (I’m not a positivist!), but I plea for a lot more fluidity between philosophy and different fields of empirical inquiry. In fact, I dare say that it is not overly optimistic to hope for two-way collaborations, where philosophical theories can contribute to the formulation and interpretation of empirical results, and empirical research can provide a more robust empirical content to philosophical theories.
This week I was on the committee of a bachelor’s thesis that I think did a good job at doing exactly that. The student, Sanne Brederoo, already has a master’s degree in neuro-linguistics (if I’m not mistaken), so she is well placed to draw interesting connections between philosophical theories and empirical predictions. The topic of her thesis was the semantics of the term ‘now’; she started out working on neuro-linguistic research on ‘now’, but then felt that this research was lacking a more robust theoretical framing.
So she turned to what philosophers and logicians have said about ‘now’, starting with Prior and going through Kamp, Kaplan, and more recently Recanati and Predelli. The two main competing accounts of ‘now’ she addressed are the indexical account defended by Kamp and Kaplan, and Recanati’s contrastive account. According to the indexical account, ‘now’ will always refer to the present time, and in particular to the time of utterance. (This is something also contested by Predelli with examples of recorded messages such as ‘I am not here now’ at an answering machine, where ‘now’ seems to refer to the time of reception, not of utterance). On Recanati’s account, however, ‘now’ can in fact be used to refer to past (and presumably future) times, because the main function of ‘now’ would be contrastive: to contrast two different moments, one further away than the other. (I hope I am not misrepresenting these theories too much with such brief descriptions!)
To test Recanati’s thesis, Brederoo focused on the concept of tense violation: when in a sentence the tense of the verb conflicts with an adverb occurring in it, one speaks of tense violation, which presumably should be identified by speakers as an ‘abormality’. She discusses the following examples from a study by Dragoy et al. (the study was conducted in Dutch, but the English ‘now’ and the Dutch ‘nu’ are sufficiently similar so that the results can be said to generalize):
• The waiter who just grinds the pepper will not get any tips.
• The waiter who now ground the pepper will not get any tips.
In both cases, the adverb contrasts with the tense of the verb: in the first case, ‘just’ indicates a past action, while the verb is present tensed; in the second case, ‘now’ (presumably) indicates a present action, while the verb is past tensed. I now quote from the thesis:
As such, both types of sentences were considered to be ungrammatical by the researchers, as well as by participants in an offline grammaticality judgment. However, in an electro-encephalography (EEG) experiment, only sentences of the type of (17) induced a so-called P600-effect at the finite verb [grinds]. The event-related potential P600 is a well-known effect as used in EEG-experiments, and is typically assumed to reflect difficulties in processing of a morphosyntactic violation (Kutas & Hillyard, 1980). Finding a P600-effect time-locked to the present tensed verb when it violated a past time frame is indicative of participants’ encountering the mismatch as described above. No such effect was found for violation of a present time frame by a past tensed verb, which leads to the interpretation that no mismatch between [now] and [ground] has occurred for participants in this study. This can be easily explained if one takes into account that now can be used to refer to a time in the past.
There are many interesting points emerging from this passage. The most noticeable one is that, while the two sentences both count as tense violation, the combination ‘just’/present-tensed verb produced the P-600 effect, suggesting that participants found it ‘weird’, whereas the combination ‘now’/past-tensed verb did not. Brederoo took these results to strongly suggest that Recanati is right in claiming that ‘now’ can be used to refer to any point in time, having a contrastive rather than indexical function. Now, although I am quite unsure of the naturalness of many of the examples used to illustrate Recanati’s contrastive theory (and thus I am not convinced that ‘now’ is indeed regularly used to refer to the past), I am inclined to agree that these results do seem to point in the direction of Recanati’s contrastive account.
Another point that I found very interesting, but which remained rather underdeveloped in the thesis, is the discrepancy between the offline and online results regarding “The waiter who now ground the pepper will not get any tips”. This sentence was judged ungrammatical by participants in the offline study, and yet the same participants did not display the expected P-600 effect for this sentence (in contrast with the ‘just’/present-tensed verb combination, which was both considered ungrammatical and elicited the P-600 effect). When I questioned her on this, Sanne referred to the pragmatics of the offline experimental situation: participants want to ‘please’ the experimenter by giving the ‘right’ answer, and thus will more quickly say that a sentence is infelicitous. But I think something more interesting than this is going on, and something with potential philosophical implications. It seems very plausible that our conscious, explicit judgment of adequacy with respect to a given expression will not necessarily match our unconscious, implicit judgment with respect to the same expression, and this is what seems to be the case here. But then, what should be the object of a theory of language: the conscious or the unconscious judgments? With this experiment, it would seem that the conscious judgments go against Recanati’s contrastive account of ‘now’, while the unconscious judgments speak in its favor. I submit that this is philosophically relevant, to say the least.
In the remainder of the thesis, Brederoo reports on experiments she conducted herself in order to further probe Recanati’s contrastive account, in particular by formulating examples which include salient times of different kinds (durative and punctual). I cannot go into all the details of the thesis here, but in conclusion, I would like to say that this particular example illustrates precisely the possibility of a nice interplay between empirical research and theoretical/philosophical approaches. Many questions remain open (after all, it is a Bachelor’s thesis!), but I think the thesis managed to circumscribe an interesting way in which a philosophical theory can be empirically tested.