Altmann, G.; Steedman, M. (1988). Interaction with Context during Human Sentence Processing, Cognition, 30:191-238.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....a modification relation, or something else , and what characterizes the attachments that escape prediction 2. Pragmatically ambiguous PPs Although structural and lexical rules alone do not suffice to disambiguate all kinds of PPs, discourse modelling is viewed as computationally costly (cf. [1]) The debate over resolution strategies is not simply about practicality, but rather, at stake is the notion of what exactly it means for a PP to attach. This paper defends discourse level strategies by arguing that a certain PP attachment anbiguity, sentential rs. verb phrase (S PP rs. VP PP) ....

Altmann, G. and M. Steedman 1988. Interaction with context during human sentence processing, Cognition, 30(3).

....be a garden path effect in the first sentence, but not in the second, whilst with the alternative context we might expect the opposite effect. Crain and Steedman [CS85] proposed a mechanism for using context to help make attachment decisions, which was later reformulated by Altman and Steedman [AS88] as the Principle of Referential Support: An NP analysis which is referentially supported will be favoured over one that is not. This means that if the context makes one interpretation more likely than the other, this interpretation will be favoured. For example, the NP the woman is ....

....it will not be referentially supported. On the other hand, the NP the woman with a blue hat is only referentially supported if there is more than one woman in the universe of discourse and one of these women is distinguished from the rest by virtue of her having a blue hat. Altman and Steedman [AS88] performed similar experiments to Crain s, using pairs of sentences supported by one of two contexts. In the VP supporting context, only one possible referent for the object of the verb was introduced, whilst in the NP supporting context, two possible referents were introduced. The idea is that ....

[Article contains additional citation context not shown here]

G. Altmann and M. Steedman. Interaction with context during human sentence processing. Cognition, 30:191--238, 1988.

.... Rene Descartes 1.1 Motivation Structural (or syntactic) disambiguation in sentence analysis is still a central problem in natural language processing. To resolve ambiguities completely, we would need to construct a human language understanding system (Johnson Laird, 1983; Tsujii, 1987; Altmann and Steedman, 1988). The construction of such a system would be extremely di#cult, however, if not impossible. For example, when analyzing the sentence I ate ice cream with a spoon, 1.1) a natural language processing system may obtain two interpretations: I ate ice cream using a spoon and I ate ice cream and a ....

Altmann, Gerry and Mark Steedman. 1988. Interaction with context during human sentence processing. Cognition, 30:191--238.

....disambiguate speech, we must be incorporating higher level knowledge of some type, whether it be syntactic, semantic, or prosodic. In order to be available for incorporation into our disambiguation decisions, however, it must be the case that this information is computed incrementally. CS85] and [AS88] argue that even reference needs to be computed incrementally to adequately explain human language processing. For example, consider the two sentences in (5) reproduced from [AS88, p.218] Processing times for the two sentences di ered signi cantly depending on whether readers were presented ....

....example, consider the two sentences in (5) reproduced from [AS88, p.218] Processing times for the two sentences di ered signi cantly depending on whether readers were presented with a situation in which there was more than one safe, or one in which there were multiple ways to blow up the safe. [AS88] claim that this data refutes theories based on minimal attachment but can be explained by incremental access to reference; context constrains our initial interpretation of the sentence, and if this interpretation is incompatible with the correct parse we have to pause and re analyze, much like ....

[Article contains additional citation context not shown here]

Gerry Altmann and Mark J. Steedman. Interaction with context during human sentence parsing. Cognition, pages 191-238, 1988.

....of this automatic disambiguation system: Inputs What informatin does it require Outputs What do we want in return Upper bounds How successful can we expect the system to be There is a lot of information that the system might conceivably require to make its decision. Some authors [Altman and Steedman, 1988] have suggested that nothing less than a model of the world and the discourse structure is required. Happily it turns out that the system can achieve reasonable accuracy without such extensive information, but the question about exactly how much context should be included remains. The sentence ....

....2. Previous Work 9 since an interpretation relating the two foodstuffs is more probable. There is considerable overlap between semantic an lexical approaches, and some lexically based approaches reviewed in that also use semantic information. 2. 4 Using pragmatics and discourse information [Altman and Steedman, 1988] suggested that PP ambiguity can only be solved if we use additional pragmatic and discourse information. They propose a set of principles, for example: A Priori Plausibility If an interpretation is more plausible in terms either of general knowledge about the world, or of specific knowledge ....

Altman, G. and Steedman, M. (1988). Interaction with context during human sentence processing. Cognition, 30:191--238.

....ambiguity so common in language Some researchers (e.g. Frazier and Fodor 1978, Mitchell Corley and Garnham 1992 and references therein) have argued that the sentence processor embodies structurally defined criteria such as Minimal Attachment and Late Closure. Others (e.g. Crain and Steedman 1985, Altmann and Steedman 1988, Trueswell and Tanenhaus 1991) have argued that ambiguity resolution decisions are made online, rather quickly, based on the relative sensibleness of the available analyses. Psycholinguistic research of this issue has focused on a limited collection of structures such as: 1) a. The horse raced ....

Altmann, Gerry and Mark J. Steedman. 1988. Interaction with Context during Human Sentence Processing. Cognition 30.

....either reinforcing or weakening them. Experiment 1 Why would theta transmission be more difficult in passive sentences than in actives Let us consider the active structure first, which is shown in Figure 1. Assume that interpretations are built more or less incrementally, from left to right (Altmann Steedman, 1988; Sedivy, Tanenhaus, Chambers, Carlson, 1999; Traxler, Bybee, Pickering, 1997) The sequence the man is analyzed as the subject of the sentence, and perhaps provisionally as some sort of a proto agent (Dowty, 1991) that is, the comprehender likely immediately accesses knowledge indicating ....

Altmann, G., & Steedman, M. (1988). Interaction with context during human sentence processing. Cognition, 30, 191-238.

....perform structural disambiguation independently from the other forms of disambiguation Schubert and Pelletier 1982, Allen 1987, Hobbs and Shieber 1987, Alshawi 1992. There is evidence, however, that structural disambiguation interacts at least with reference interpretation Crain and Steedman 1985, Altmann and Steedman 1988 and a lot of the recent work on statistical parsing relies on the hypothesis that lexical interpretation Semantic Ambiguity and Perceived Ambiguity 41 6 Discussion I have suggested that to develop a theory of discourse interpretation that is consistent with what we know about the problem of ....

Altmann, G. T. M., and M. Steedman. 1988. Interaction with Context during Human Sentence Processing. Cognition 30:191--238.

....theories. Modern theories of disambiguation fall into two classes, those based on global metrics and those based on local structural heuristics. The global models choose some single global metric (such as prefer the most semantically plausible interpretation (Crain Steedman 1985; Kurtzman 1985; Altmann 23 Steedman 1988; Charniak Goldman 1988) to choose among structures. The syntactic heuristic models use a simple syntactic heuristic or combination of heuristics such as choose the syntactically simplest interpretation (Kimball 1973; Frazier Fodor 1978; Wanner 1980; Shieber 1983; Pereira 1985; Kaplan 1972; ....

ALTMANN, GERRY T. M., & MARK J. STEEDMAN. 1988. Interaction with context during human sentence processing. Cognition 30.191--238.

....due to the restricted communication pathways. In single path parsing, control strategies such as minimal attachment (Frazier, 1987) correspond to specifying a fixed in advance scheme for ambiguity resolution 5 . More flexible control structures underly weakly interactive models such as Altmann and Steedman s (1988), which assume that variable content (e.g. referential context) can affect the control of parsing processes. The rigidity of traditional control structures in computer science led Newell (1962, 1973) to consider production systems as an alternative control scheme. In a production system, the ....

Altmann, G. & Steedman, M. (1988). Interaction with context during human sentence processing. Cognition, 30:191--238.

....explain Moore and Moser s puzzle; it would explain why contributor:core constructions as in A so B are surprisingly frequent, and demonstrably e ective. We conclude with a general moral, which links this suggestion back to the discussion of the supposedly causal interpretation of full stops. As Altmann and Steedman (1988) have said: there is no such thing as a null context. ....

Altmann, G. T. M. and Steedman, M. J. (1988) Interaction with context during human sentence processing. Cognition, 30, 191-238.

....of very specific associations. The processing organization that emerges here is similar to other proposals in which a syntax module generates alternative structures in parallel, and semantic and contextual knowledge sources decide among them via a restricted interface (Warner Glass, 1987; Altmann Steedman, 1988). As Altmann (1988) points out, by allowing fine grained control over a syntax module (such that parsing decisions can be influenced at the level of the word) it is possible to predict interactive effects while still preserving the modularity hypothesis. These models, along with NL Soar, make ....

.... The processing organization that emerges here is similar to other proposals in which a syntax module generates alternative structures in parallel, and semantic and contextual knowledge sources decide among them via a restricted interface (Warner Glass, 1987; Altmann Steedman, 1988) As Altmann (1988) points out, by allowing fine grained control over a syntax module (such that parsing decisions can be influenced at the level of the word) it is possible to predict interactive effects while still preserving the modularity hypothesis. These models, along with NL Soar, make clear the problem of ....

Altmann, G. & Steedman, M. (1988). Interaction with context during human sentence processing. Cognition, 30:191--238.

.... that shares structure requires careful bookkeeping of which link goes with which interpretation (Lombardo, this volume) Finally, note that none of these arguments against parallel, breadth first parsing are arguments against parallel generation of syntactic alternatives at ambiguous points (Altmann Steedman, 1988). These are orthogonal issues. The arguments are against maintaining and pursuing distinct interpretations in parallel. The distinction can be seen clearly in the Soarbased comprehension model of Lewis (1993a) This architecture separates the parallel generation and selection of potential parsing ....

....mechanism and the open control structure means that NL Soar can account for both the structural regularities captured by the syntactic metrics of garden path difficulty (e.g, Pritchett, 1992; Gibson, 1991) and the flexible, context sensitive ambiguity resolution of interactive theories (e. g, Altmann Steedman, 1988). The open control of parsing also means that once NL Soar s on line repair process has failed, it can engage in a deliberate recovery process by carefully rereading the problematic sentence. Carefully is given a precise computational definition in Soar: in effect, it means that Soar is forcing ....

Altmann, G. & Steedman, M. 1988. Interaction with context during human sentence processing. Cognition , 30, 191-238.

....is generally construed to be the less likely scenario. Again, surrounding discourse may bias the likelihood in a particular direction. For example, take the sentences: 6) The florist sent the flowers (was very pleased) 7 Many of these contradictory findings are themselves under dispute. See Altmann and Steedman (1988) for example. 17 (7) The performer sent the flowers (was very pleased) Without the optional ending, both sentences take the simple active form where the florist performer is the sender; with the additional three words the main noun loses the active sender s role, resulting in the reader being ....

Altmann, G.T.M. & Steedman, M.J. (1988) Interaction with Context During Human Sentence Processing. Cognition, 30, 191-238.

....We propose it to be grounded in structural complexity. 5 Many other proposals that have discussed the resolution of MV RR ambiguity are not equipped to explain the difference between verb classes within the same structural ambiguity. For example, referential theory (Crain and Steedman, 1985; Altmann and Steedman, 1988) proposes that RRs are particularly difficult because they require more presuppositions than the main verb alternative. This is due to the restrictive nature of the RR, which presupposes a set of elements in the discourse from which the RR picks one. The suggestion is that the HSPM follows a ....

Altmann, G. and M. Steedman (1988). Interaction with context during human sentence processing. Cognition 30 (3), 191--238.

No context found.

Altmann, G. T. M. & Steedman, M. (1988) Interaction with context during human sentence processing. Cognition, 30, 191-238.

.... the present account of indefinites might appear to mix derivation and evaluation in a dangerous way, this is in fact what we would expect from a monotonic semantics that supports the use of incremental semantic interpretation to guide parsing, as humans appear to (see Crain and Steedman 1985 and Altmann and Steedman 1988). Further support for a non quantificational analysis of indefinites can be obtained from the observation that certain nominals that have been talked of as quantifiers entirely fail to exhibit scope alternations of the kind just discussed. One important class is the non specific or nongroup ....

Altmann, Gerry, and Mark Steedman. 1988. "Interaction with Context During Human Sentence Processing." Cognition, 30, 191--238.

No context found.

Altmann, G.; Steedman, M. (1988). Interaction with Context during Human Sentence Processing, Cognition, 30:191-238.

No context found.

Altmann, G., & Steedman, M. (1988). Interaction with context during human sentence processing. Cognition, 30, 191--238.

No context found.

G. Altmann and M. Steedman, "Interaction with context during human sentence processing," Cognition, vol. 30, pp. 191--238, 1988.

No context found.

Altmann, G. & Steedman, M. (1988). Interaction with context during human sentence processing. Cognition, 30, 191-238.

No context found.

Altmann, G., & Steedman, M. (1988). Interaction with context during human sentence processing. Cognition, 30, 191--238.

No context found.

Altmann, G. T. M., & Steedman, M. J. (1988). Interaction with context during human sentence processing. Cognition, 30, 191--238.

No context found.

Altmann, Gerry, and Mark Steedman. 1988. Interaction with context during human sentence processing. Cognition, 30, 191-238.

No context found.

G. Altmann and M. Steedman. Interaction with context during human sentence processing. Cognition, 30:191# 238, 1988.

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC