Similar books and articles
Filters are simply ambiguous between plugs and holes. These rules are fairly familiar from the narrative-understanding literature, although given a unique formulation. Standardly, activities are taken as having exactly the same temporal effects as states. Ter Meulen treats them instead in a way that is somewhat intermediate between achievements and states. That is, there is overlap, but there is also some movement in time. I agree with this general type of approach to activities, and have myself argued for a similar, though not identical, treatment in Almeida , DATs are trees of event types and are intended to have a number of functions with respect to temporal processing.
Much of the book is taken up with demonstrations of their utility. For example, because they are trees, DATs can greatly simplify reasoning about the temporal relations among events of the story. Each node of a DAT is either a hole or a plug, and represents at least one event. The parent-child relationship between nodes represents temporal inclusion of the child event in the parent event, latex; mind MATHKAP mindr Each DAT has a unique rightmost node, called the source, which represents the event of issuing the information of the text, and each DAT has a single root node, which dominates all other nodes in the DAT, including the source node.
This root node represents the entire described episode. The rules informally presented in the introductory chapter for deriving the temporal structure of narratives are more precisely described in this chapter in terms of updating operations on DATs. The rules for updating a DAT depend on whether this node is a hole or a plug: if the current node is a hole, then the new node becomes its child; if the current node is a plug, then the new node becomes its right sister.
States, which also include perfect and progressive events, are treated separately. A state never introduces a new node; instead, it is appended as a sticker to the current node if the current node is a plug, or appended to the next node constructed if the current node is a hole. As an example, the figure below Figure 3.
She sat down, attempted to decipher the message, and looked at her watch. She sighed. The root of the DAT is at the top, with its source to the right. The filled-in nodes are plugs, while the open nodes are holes. The set of nodes on a path through a DAT from the root to a leaf is called a chronoscope. Chronoscopes play a major role in how DATs are used, especially with respect to temporal reasoning. Because the parent-child relationship represents temporal containment, all of the events of the nodes of a chronoscope are concurrent, at least for the duration of the leaf event.
Thus, chronoscopes are somewhat reminiscent of situations in the situation calculus. The next two chapters examine the applicability of DATs to some additional problems in narrative processing. Although this book is clearly written, it is not really for beginners in tempo- ral representation or narrative understanding, first of all because of the unusual representational approach taken, but also because the author concentrates almost entirely on presenting her specific proposals.
The book has an unusually thorough bibliography, but the author makes relatively little use of it. Little in the way of background information is given, and few of the ideas presented are compared with earlier work or with alternative proposals from the literature. For example, ter Meulen represents events expressed in perfect-form sentences as perfect states holding in the narrative present, but the alternative of representing them as events occurring in the narrative past is not discussed.
The idea that some abstract structures, such as DATs and their constituent chronoscopes, may be important in the interpretation of narratives is one well worth investigating. However, it still appears to me that DATs by themselves cannot constitute an adequate representation of temporal information. For one thing, the intricate set of possible temporal relationships between events will be hard to capture in a simple tree structure with only before and during as its possible relationships. In terms of expressivity, DATs appear to be intermediate between the situation-calculus style of representation and a full-fledged temporal logic.
Although I feel that the author tries to get too much out of her proposed structures, they may turn out to be important components of a more complete approach to temporal representation and reasoning. Almeida, Michael J. Duchan, Gail A. Bruder, and Lynne E. Hewitt eds.
Representing Time in Natural Language
It is opposed to those philosophers, such as a number of Wittgensteinians see, e. Hacker , or Tyler Burge , or Donald Davidson , for example , whose rejection of a Carte- sian view of the mind is largely independent of the question of materialism.
Typical instruments used in situations are more available when situation descriptions make use of the imperfective rather than the perfective aspect [ 13 ]. Imperfective descriptions also enhanced accessibility of intention-related concepts and led to greater intentional attributions than perfective descriptions [ 14 ]. Even the overall activation level of a situation over time is greater for situations described in the imperfective rather than the perfective aspect [ 6 ]. Becker, Ferretti and Madden [ 15 ], found that reintroductions of target events later in a story were easier to integrate when the target event had been initially described in the imperfective event left ongoing rather than the perfective aspect event described as finished.
Describing past actions in the imperfective rather than the perfective aspect has been shown to promote memory for action-relevant knowledge and reenactment of these actions in a future context [ 16 ]. In line with such findings, Coll-Florit and Gennari [ 17 ] found that imperfective verbs took more time to read and were related to a larger range of semantic associations than perfective verbs, suggesting that events described in the imperfective aspect involve more extensive semantic processing.
Recent studies have also shown that the influence of grammatical aspect on cognitive processing goes beyond methodologies targeting overt language use. For instance, Athanasopoulos and Bylund, [ 18 ] investigated the effects of aspect on non-verbal categorization, demonstrating how not only verbalizations but also non-verbal memories for video clips are more endpoint focused in Swedish than in English participants, because the Swedish language does not allow for a progressive ongoing verbalization. In addition, Flecken and colleagues [ 19 ] demonstrated an effect of grammatical verb aspect on the P3 wave in event-related brain potentials.
German-speaking participants showed larger P3 amplitudes for pictures that matched the endpoint rather than the trajectory of preceding motion videos. English-speaking participants did not show this difference because English is an aspectual language that focuses trajectory and endpoint equally, whereas German verbalization focuses the endpoint rather than the trajectory.
This highlights how a nonlinguistic task such as motion perception is sensitive to attentional biases instantiated by the grammar of a language. While significant progress has been made in our understanding of how verb aspect affects the representations of described information, several issues remain to be explored.
In particular, the focus of the event in perfect ive sentences is unclear. The imperfective aspect seems to boost availability of features of the event as well as the event as a whole, but does this mean that events described in the perfect ive aspect are somehow less available in our situation model? One alternative to this idea is that the perfect aspect focuses the representation on a specific facet of the event that has been under accessed in previous research.
This idea will be further explored below, after the temporal nature of language representations is discussed. A growing body of recent research supports the idea that comprehenders understand language by activating simulations that incorporate perceptual information [ 20 , 21 ]. For instance, Zwaan and colleagues have demonstrated that comprehenders are faster to verify pictures and sentences when the perceptual constraints of the sentence match the visually displayed features, such as shape [ 22 , 23 ] and orientation of objects [ 24 ], as well as direction of movement [ 25 ].
According to the simulation view of language comprehension, the words in a phrase or sentence activate lexical level simulations general word meanings that are combined to yield situation-specific simulations of the larger phrase or sentence. These situation simulations are partial reactivations of traces of our experience, and therefore, they necessarily imply a spatio-temporal perspective on the described situation.
Thus, when reading about walking in the rain or writing a letter, comprehenders simulate the experience of the described situation, including temporal and spatial features of the objects and instruments in the situation. The grammar of a sentence is an important way to constrain the simulation, and therefore grammar should have measurable effects on how the described situation is simulated. When the perspectives and constraints of a presented image overlap with those of the perceptual simulation activated by a sentence, then this overlap should produce processing facilitation.
Comprehenders do in fact seem to use verb aspect as a cue to regulate the activation of ongoing simulations of situations over time. For the past imperfective sentences, participants were faster to process the picture laptop , the two words following the picture, and the sensibility judgments when target entities were pictured in use open laptop rather than not in use closed laptop. However, this in-use facilitation was limited to processing of the picture for the past perfect sentences.
The two words following the picture, as well as the sensibility judgment were no faster when entities were pictured in use rather than not in use for these past perfect sentences. Although it is clear that perceptual simulations are activated during language comprehension, and that grammar serves to constrain these simulations, it is not well understood how this process unfolds in real-time. As mentioned above, it remains unclear what part of a simulation is focused in the case of past perfect sentences.
It was never the agent, but it could be an instrument used to perform an action, a resulting object of the action, or any other entity that participated in the action. Because the perfect aspect focuses the resulting state of a situation [ 1 ], it is perhaps the resulting object that is focused in the case of the perfect aspect completed action.
We asked participants to read imperfective John was using a corkscrew to open the bottle in the restaurant and perfect John had used a corkscrew to open the bottle in the restaurant event descriptions, word-by-word, with the critical instrument word replaced by a picture of that instrument in-use or not in-use, and the critical object word replaced by a picture of the object in its ongoing or resultant state. Response times to process the two pictures in the sentence were measured, as well as accuracy and response times of sensibility judgments to the sentences.
Representing Time in Natural Language | The MIT Press
Before proceeding to a detailed description of the methods, we first provide the motivation for our grammatical choices in testing the aspectual distinction. Furthermore, as the experimental sentences were presented in French to native French speakers, a brief discussion of the aspectual system in French is also warranted. For various reasons and in line with the majority of research on verb aspect , we chose to focus on the past tense.
First, much of our narrative descriptions are in the past tense, as we usually re-tell our experiences after they have happened. Second, as the present can be considered an infinitely small point in time, the present tense does not offer a suitable testing ground for the aspect comparisons.
Finally, the past tense provides a better alignment of tense-aspect systems in English and French. While English and French aspectual systems differ substantially in the present tense, the English imperfective and past perfect find quite similar counterparts in the French past tense. In terms of imperfect options in the past tense, the French imparfait utilisait corresponds almost exactly to the English imperfective was using.
In all cases, the entire situation is described at once the beginning, middle, and especially the end of the event are included in the scope of the description. In addition, the perfect aspect is thought to focus the resultant state of events, whereas the imperfective aspect instead focuses agents, instruments, locations, etc.
However, objects may also be in focus for the imperfective aspect because it is more inclusive of any object or entity relevant to the ongoing action resulting object included. Twenty-eight native French-speaking volunteers, with normal or corrected-to-normal vision took part in this experiment. Both this and the following experiment were conducted in agreement with the code of ethics of the world medical association declaration of Helsinki and were performed under approval Authorization No.
Participants were informed about the experimental procedures before they provided written consent, and they received a financial compensation for their participation. Sixty French sentences were generated to describe a named person using an instrument or tool to act upon an object i. For each of the 60 sentences, 6 versions were created.
Next, for both the perfect and imperfective sentence versions, the instrument word corkscrew was replaced by an image of that instrument either in use corkscrew open in hand at correct orientation or not in use corkscrew closed. Finally, the object word bottle was replaced by an image of that object depicted while the action was in progress bottle being opened or completed uncorked bottle See Fig 1 for example stimuli.
Sentences always included three words after the second picture, so as not to end the sentence on a critical picture stimulus. In addition to the experimental stimuli, 64 similar filler sentences were generated, each with two words replaced by images, as in experimental sentences. The experimental sentences 6 versions of the 60 original sentences were counterbalanced across 2 lists that each presented a given sentence 3 times 3 different versions in 3 separate blocks.