A unified, self-organizing model of garden path phenomena, center-embedding phenomena, and interference effects
نویسنده
چکیده
Two phenomena in sentence processing, the difficulty associated with garden path sentences and the difficulty associated with center embeddings, have generally been ascribed to different features of the parsing mechanism. For example, garden path phenomena are claimed to stem from biases (possibly structural, lexical, discourse, etc.) that cause the parser to favor a parse that eventually turns out to be wrong (e.g., Frazier & Rayner, 1982; MacDonald et al., 1994). Center embedding difficulty has been ascribed to the strain of simultaneously holding in memory the states of multiple partially completed constituents (Gibson, 1998). The current proposal follows Vosse & Kempen (2000)'s observation that that both phenomena arise naturally in a system in which perception of words spawns the activation of phrase-tree fragments which interact in a self-organizing fashion to form larger tree structures. This kind of model treats garden path phenomena as a kind of interference effect: in such cases, one fragment gets the jump on another early in a sentence, but later arriving information allows the initially squashed choice to make a comeback bid, the battle for dominance delaying stabilization. Likewise, because the system allows all words in a sentence to interact with one another, a sequence of similar noun phrases (e.g., "the dog the cat the rat...") followed by a sequence of verbs that can combine with any of them ("...chased bit died") leads to considerable interference and often failure of the parse. Moreover, several recent research projects have found evidence for the novel claim of the self-organizing perspective that all words can interact with each other: Van Dyke et al. (2005) found that that the semantic compatibility of a noun in an embedded clause with the verb phrase in the matrix clause affects eye movements (1); Tabor et al. (2004) found that the parser experiences interference from locally coherent structures even if they are incompatible with the already existing primary parse (2).
منابع مشابه
Interference in short-term memory: the magical number two (or three) in sentence processing.
Many theories have been proposed to explain difficulty with center embedded constructions, most attributing the problem to some kind of limited-capacity short-term memory. However, these theories have developed for the most part independently of more traditional memory research, which has focused on uncovering general principles such as chunking and interference. This article attempts to gain s...
متن کاملRecent developments in the NL-Soar garden path theory
This report describes a theory of garden path phenomena that is emerging from work on NL-Soar, a computational model of language comprehension embedded within the Soar architecture. The theory is constrained by a corpus of two kinds of sentences: garden paths (GP), which reveal the limitations of human comprehension in dealing with local ambiguities, and non-garden-paths (NGP), which reveal its...
متن کاملLink Prediction using Network Embedding based on Global Similarity
Background: The link prediction issue is one of the most widely used problems in complex network analysis. Link prediction requires knowing the background of previous link connections and combining them with available information. The link prediction local approaches with node structure objectives are fast in case of speed but are not accurate enough. On the other hand, the global link predicti...
متن کاملCompare the Phenology of Fritillaria, Bellevallia and Muscari species in National Botanical Garden of Iran
Phenology is one of the important issues in ecology, which studies various stages of plant life and its relation with climatic and ecological conditions. Useful information for the production cycle is obtained by evaluating the various stages of native species development. In this research, the phenology of 17 species of Fritillaria, Bellevallia and Muscari in the National Botanical Garden of I...
متن کاملSelf-organizing maps in sequence processing
Models are abstractions of observed real world phenomena or processes. A good model captures the essential properties of the modeled phenomena. In the statistical learning paradigm the processes that generate observations are assumed unknown and too complex for analytical modeling, thus the models are trained from more general templates with measured observations. A substantial part of the proc...
متن کامل