On March 31st and April 1st, 2012, PHLING hosted Maryland's first-annual interdisciplinary research symposium on //events//, as part of our ongoing events project. The symposium brought together young researchers working on "events" in philosophy, linguistics and psychology. The symposium related discussion of events in these fields, bringing into conversation work in ontology, logic, semantics, and perception. Our two invited speakers were Achille Varzi of Columbia University, and Paul Pietroski of the University of Maryland. The symposium was designed with the goal of providing maximal amount of time for discussion, which, we are pleased to report, was well-utilized. Students and faculty across these disciplines clearly have a lot to say to and learn from each other.
PHLINC1 was sponsored by the Departments of Linguistics and Philosophy, the Graduate Student Government at Maryland, and UMD's NSF-IGERT. PHLING was formed as part of the Language Science initiative at Maryland.
|9-10||Breakfast & registration|
|10-10:45||Brendan Ritchie & Chris Vogel, Maryland. Introduction, and Speaking of (and seeing) events.|
|10:45-11:30||Brent Strickland & Brian Scholl, Yale. Visual perception involves 'event type' representations: The case of containment vs. occlusion.|
|11:45-12:30||Lilia Rissman, Kyle Rawlins, & Barbara Landau, Johns Hopkins. Event participants and verbal semantic structure.|
|12:30-1:15||Joshua Hartshorne, Timothy O'Donnell, Yasutada Sudo, Miki Uruwashi, & Jesse Snedeker, Harvard. Linking event structure to language: Linguistic universals and variation.|
|1:15-3||Lunch (on your own)|
|3-4:15||Invited talk: Achille Varzi, Columbia. Negative Events and Logical Form.|
|6||Dinner party @ Queensbury house|
|10-10:45||Robert Rovetto, University at Buffalo. Processing events: toward a specialization of temporally-extended entities.|
|10:45-11:30||Ashley Atkins, Princeton. Making progress.|
|11:45-12:30||Kaeli Ward, UCLA. Durativity in Chechen pluractionality.|
|12:30-1:15||Curt Anderson, Michigan State. Dimensions of comparison determine telicity in verbal comparatives.|
|1:15-3||Lunch (on your own)|
|3-4:15||Invited talk: Paul Pietroski, Maryland. Event variables and framing effects.|
Negative Events and Logical Form
Achille C. Varzi, Columbia University
On a broadly Davidsonian view, the distinction between objects and events is central to the way we talk. If we say that John is smart, what we say is true or false according to whether a certain object (a person) is smart, but if we say that John kissed Mary tenderly, what we say is true or false according to whether a certain event (a kiss) was tender.
This view is not uncontroversial but I am happy to go along with it. The question I am interested in concerns the scope of the view. Assuming events form a genuine ontological category, shall we take our domain of discourse to also include "negative" events—things that didn't happen—along with positive ones? After all, we often seem to be talking about such things, too. We speak of John's kiss with the same easiness with which we speak of the speech he was supposed to give, the nap he did not take, the party he failed to organize. Davidson himself was candid about this: "We often count among the things an agent does things he does not do". Yet this may offend our ontological scruples. If John did not actually do those things, why should we include them in an inventory of the actual world?
In a way, negative events are like non-existing objects. We often speak as though there were such things, but deep down we may want our words to be interpreted in such a way as to avoid serious ontological commitment. For example, we may say that reference to Sherlock Holmes or to the golden mountain is to be understood within the context of a fictional story, a pretense, a game of make-believe, and we may likewise say that reference to the things John didn't do is to be understood within the context of a suitable fictional scenario. It is true that Sherlock Holmes lived in Baker Street insofar as that is true according to Conan Doyle's narrative; it is true that the John's canceled speech was on freedom insofar as that is true according to certain counterfactual speculations about how the world might or should have been.
Unfortunately, negative events appear to be more resistant to Occam's razor than the analogy with non-existing objects might suggest. Not only do we often speak as though there were such things. We often speak in such a way as to suggest that reference to or quantification over negative events is to be taken strictly and literally. Typical examples include statements such as 'John often hasn't paid taxes' or 'Mary saw John not leave', whose Davidsonian logical form appears to require a straight commitment to actions that John failed to perform. Or think of what we say when we engage in causal explanations. We say that John's failure to water the flowers caused their death, that his omitting the cutlery from the wedding list made Mary angry, or that the main cause of the fire was lack of rain. How can we uphold to the view that causation is a relation between events (a view that has many advantages over its competitors) and yet deny that sometimes the terms of a causal transaction involve failures, omissions, events that did not happen?
The point of this talk is to outline a way of resisting such difficulties on behalf of the view according to which the only events to be seriously countenanced at the level of logical form are the positive ones—those that actually feature in the history of the world. This involves articulating two distinct but related ideas. First, there is the idea that in some cases a "negative" event (a non-leaving) just is an ordinary, positive event (a staying); it is a positive event under a negative description. Second, there is the idea that not all causal explanations are reports of causation. When we mention John's failure to water the flowers as an explanans of why they died (for example), we do not say what caused the death of the flowers. We do not mention any of the relevant causes. We just remark that one sort of event that was supposed to occur, and whose occurrence would have helped the flowers to survive, did not in fact occur.
Event variables and framing effects
Paul Pietroski, University of Maryland
To represent is to represent in a particular way. Representing in a particular way makes one susceptible to framing effects, as discussed by Kahneman and Tversky (among others). For example, our natural systems of judgment are influenced by whether fuel efficiency is described in terms of miles per gallon or gallons per mile. A modern Cartesian might say, "I cognize, ergo I frame." In some domains, these effects are dramatic and systemic, raising questions about whether our "intuitions" are stable enough to be described as judgments about mind-independent states of affairs. I'll discuss an example, and then turn to a question within semantics that has generated considerable discussion: how are "event" variables related to events?
Competent speakers can be induced to judge, in a controlled context, whether or not (1)
(1) Alvin chased Theodore.
can be used to report a certain situation that might be described in many ways. Likewise, speakers who understand (2-5) can evaluate these sentences as partial reports of what happened.
(2) Alvin chased Theodore joyfully and athletically, but not skillfully.
(3) Theodore chased Alvin joylessly and unathletically, but skillfully.
(4) Harry played the song dramatically in two minutes.
(5) Harry played his tuba dramatically for two minutes.
Such judgments confirm Davidsonian analyses, according to which (1-5) are understood as existential claims of the form '∃[...e...]'. But as many authors have noted, the judgments also tell against construing the 'e'-variable as ranging over language-independent events. Despite heroic attempts to preserve this construal, and maintain that sentences like (1-5) have recursively specifiable truth conditions, I think there is an objection to truth conditional semantics that runs deep. If sentences like (1-5) have meanings that determine (functions from contexts to) truth conditions, then speakers are prone to severe framing effects that make their judgments unreliable as a source of evidence for or against proposed theories of meaning/truth for naturally acquirable human languages. But if meanings are more psychologistic and internalistic, then the data are less surprising, and judgments are more reliable as a source of evidence in semantics.
Introduction, and: Speaking of (and seeing) events.
Brendan Ritchie & Chris Vogel, University of Maryland
Humans are capable of representing the world as a series of causally related events. The character of these representations seems to be hierarchical (Zacks et al 2001, Zacks 2004) and often driven even by low-level visual stimuli (Scholl & Tremoulet 2000). The neo-Davidsonian tradition in natural language semantics treats the meanings of transitive expressions (like Mary kicked Bill) as containing thematic relations between events and their sub-event constituents (Parsons 1990; Pietroski 1998, 2003; Schien 2010). In particular, some simple transitive clauses are analyzed as making reference to an initial event and terminative event linked by some relation R (often interpreted as direct causation). Such logical forms suggest satisfaction by non-linguistic representations of events that are hierarchically structured and causally linked. Introducing the first annual PHLINC, we present our ongoing research into the relationship between the posited satisfiers of neo-Davidsonian logical forms and the human capacity to represent the world eventively.
Visual perception involves 'event type' representations: The case of containment vs. occlusion
Brent Strickland & Brian Scholl, Yale University
Recent infant cognition research suggests that our core knowledge involves "event type" representations: during perception, the mind automatically categorizes events into broad types (e.g. occlusion and containment), which then guide attention to different properties (e.g. with width processed at a younger age than height in containment events, but not occlusion events). We tested whether this aspect of infant cognition also structures visual processing in adults. In a series of experiments, adults viewed dynamic 2D displays that each included several repeating events wherein rectangles moved vertically between the top of the screen and into/behind containers. Occasionally, the moving rectangles changed either their height or width while out of sight, and observers pressed a key when they detected such changes. Change detection performance mirrored the developmental results: detection was much better for width changes than for height changes in containment events, but no such difference was found for occlusion events. In follow-up experiments where the moving objects moved horizontally into or behind the static containers, the prioritized dimension for containment events flipped: height was detected more readily than width for containment but not for occlusion. Notably, this overall pattern of results held even for observers who failed to notice the occlusion/containment difference. These findings demonstrate a novel way in which core cognition from infancy continues to automatically guide adult visual performance, and they suggest for the first time that event-type representations are part of the underlying currency of visual cognition.
Event participants and verbal semantic structure
Lilia Rissman, Kyle Rawlins, Barbara Landau, Johns Hopkins University
Semantic theories differ as to how many arguments a verb encodes: Davidson (1967) proposed that verbs encode an eventuality argument as well as agent and patient-type arguments, while in neo-Davidsonian approaches, the eventuality argument is the verb's only argument. Regardless of one's position on this issue, the intuition remains that an event of "giving" fundamentally involves three participants, whereas an event of "wanting" fundamentally involves two. The nature of this "participant" level of verbal representation is poorly understood: defining in terms of necessary and sufficient conditions what it means for a verb to "fundamentally involve N participants" is essentially impossible. In this study, we investigated verbal encoding of instrumental participants, as in John sliced the cake with a knife. For verbs such as slice, hit, dig and write, it is not obvious whether these verbs encode an instrument as one of their "essential participants." In an experimental task, we probed English-speaking adults' intuitions about the semantic argument structure of verbs. In many respects, subjects' intuitions were consistent with traditional theories of argument structure. In the case of instruments, however, subjects' responses were quite variable across-verb and across-subject, suggesting that verbs encode event participants as part of a non-categorical representational space. We discuss how these data may be reconciled with neo-Davidsonian approaches to event semantics.
Linking event structure to language: Linguistic universals and variation
Joshua Hartshorne, Timothy O'Donnell, Yasutada Sudo, Miki Uruwashi, and Jesse Snedeker, Harvard University
To communicate, speakers must place the different participants of an event (e.g., causal agent, affected entity) in predictable syntactic positions (e.g., subject, object) so that listeners will know who did what to whom. This mapping is often argued to be governed by linking rules that are consistent within each language and universal across languages. This hypothesis is challenged by verbs of psychological state: The experiencer of the state can appear as either the subject (Mary fears/hates/loves John) or the direct object (Mary frightens/angers/delights John). The present studies explore whether this variability may actually result from differences in these verbs' meanings. Specifically, we find that both English- and Japanese-speakers use the typical duration of a psychological state to guide novel verb learning, mapping the experiencer of a long-lived state onto subject-position and the experiencer of a short-lived state onto object-position. Thus, even these verbs are subject to potentially universal mapping constraints.
Processing Events: Toward a Specialization of Temporally-Extended Entities
Robert Rovetto, University at Buffalo
The background motivation, herein, is (a) to understand the difference, if any, between processes and events, and (b) whether a top-level specialization of these temporally-extended entities is viable. In discussing some proposed differences between the ontological categories of event and process, I put forth an understanding of event that brings to bear its fiat aspect. I also describe processes as having a patterned, structured or ordered nature. In using the notion of pattern I put forth some specializations of temporally-extended entities for use in upper-level ontologies.
Ashley Atkins, Princeton University
A progressive sentence like 'Mary is baking a cake' raises surprising and interesting puzzles for semantic accounts of the progressive. It seems to have features that set it apart from a sentence like 'Mary is icing a cake.' The latter intuitively entails that there is some thing that Mary is icing as well as that there is a cake and Mary is icing it. It is not obvious that 'Mary is baking a cake' shares this entailment pattern. This raises the question "What existential commitments are carried by these sentences?"
I will consider four semantic accounts of the English progressive with a view to comparing and evaluating their answers to this question. The first two are extensional accounts of the progressive and the last two intensional accounts. As each of these accounts is situated within an event semantics framework, the main question I will address may be put in the following way: Given that these sentences describe events in progress (in a sense to be explained), do they also characterize objects as being in progress?
Against these representative accounts, I will argue that 'Mary is baking a cake' neither entails that there is a cake and Mary is baking it nor that Mary is baking some thing, whereas 'Mary is icing a cake' does exhibit this entailment pattern. A semantic analysis of the progressive should capture these facts. I will sketch a semantic analysis that does.
Durativity in Chechen Pluractionality
Kaeli Ward, UCLA
In this paper, I provide a semantic analysis of event plurality, called pluractionality, in Chechen (East Caucasian, Chechnya). In particular, I demonstrate how durativity (1), a relatively rare cross-linguistic interpretation of pluractionality (Yu, 2003), falls out naturally as a pluractional interpretation. I demonstrate this by allowing events to be measured by μ, a function standardly used to measure the quantity of individuals (Cresswell (1976), Rett (2008), and others).
(1) cyna chow xiizhira
3SG.POSS wound hurt.PLR.WP
'His wound ached (for a long time).' Chechen (Yu 2003)
To do this, I draw on the analysis of Kaqchikel pluractionality in Henderson 2010 as well as the analyses in Nakanishi 2007 and Burnett 2012 of how μ can be used to measure events. In so doing, I demonstrate strong parallels between the individual and event domains, both with respect to plurals and the mass/count distinction, providing further evidence for the ontological existence of events. In addition to providing an analysis of Chechen (all of the data for which is provided by Yu (2003) and Wood (2007)), I will update the analysis of Kaqchikel to capture the fact that whereas both Kaqchikel and Chechen have a frequentative pluractional interpretation, (2) and (3) respectively, Kaqchikel cannot have a durative interpretation for its pluractianal.
'He touched me repeatedly.' Kaqchikel (Henderson 2010)
(3) as qiigashna twop-qissira
1SG crow.PL.DAT gun-throw.PLR.WP
'I shot crows many times.' Chechen (Yu 2003)
Dimensions of comparison determine telicity in verbal comparatives
Curt Anderson, Michigan State University
Well known is that direct objects often determine the aspectual properties of an event, but less studied is how other constructions are used to calculate aspect. Previously unnoticed is that some but not all verbal comparatives can trigger telic interpretations of events, varying both with the dimension of comparison and with the verb. Varying the verb but keeping the dimension of comparison constant allows for different temporal adverbials to be licensed in (1). A similar fact holds when the verb is kept constant but the dimension of comparison varies, as in (2).
(1) a. Mike ran further than Mandy in an hour / *for an hour.
b. Mike hit the ball further than Mandy for an hour/ *in an hour.
(2) a. Mandy ran further than Mike in an hour / *for an hour.
b. Mandy ran harder than Mike for an hour / *in an hour.
I explain this pattern by appealing to Schwarzschild (2002)'s notion of monotonicity, where a dimension of measurement is monotonic to the part structure of a measured object if the measurement tracks that object's part structure. Comparisons that license in-adverbials have dimensions that are monotonic with respect to the part structure of the event denoted by the verb, while comparisons that license for-adverbials have dimensions that are nonmonotonic.
This analysis poses two questions. First, why does monotonicity trigger telic interpretations? Telic interpretation has often been argued to depend on quantization, the lack of proper parts (Krifka 1992). Given that monotonicity also relies on a lack of proper parts with the same measure, this suggests a close relationship between quantization and monotonicity. Second, what determines the part structure of an event? I demonstrate that in-adverbials can coerce monotonic interpretations with suitable context, showing that part structures cannot be lexically predetermined, but must be imposed with help from world knowledge and linguistic context.
Travel directions to College Park, to the university, and to Marie Mount Hall are here.
Some discount bus lines stop at the Greenbelt Metro Station, northeast of College Park. This is a good place to get off, as it is just one stop away from the College Park Metro station, on the Green Line of the D.C. Metro.
Note that College Park is just outside of Washington D.C., four miles up from its northeastern border on Route 1. It is a stop on the Green Line of the D.C. Metro, a twenty minute ride from (for example) the L'Enfant Plaza station. So you may want to combine your trip here with a visit to the capital.
On Saturday and Sunday, parking is unrestricted in most (but not all) lots on campus (signs make it clear when this is not the case).
There are several generic hotels in College Park itself. The closest to Marie Mount Hall is the Quality Inn on Route 1, just south of the campus. It is easy to walk from the Quality Inn to Marie Mount Hall. There are other hotels north of the campus on the Route 1 - but walking from these is rather unpleasant.