Toward a connectionist model of recursion in human linguistic performance

By Morten H. Christiansen, N. Chater. 1999

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Natural Language Processing

Citations in the corpus

Inferring Algorithmic Patterns with Stack-Augmented Recurrent Nets (2015, methodology): “Christiansen & Chater (1999) extended these results to grammars with larger vo-cabularies.”
Colorless Green Recurrent Networks Dream Hierarchically (2018, background): “and even context-sensitive grammars (e.g., Elman, 1991, 1993; Rohde and Plaut, 1997; Christiansen and Chater, 1999; Gers and Schmidhuber, 2001; Cartling, 2008).”
COGS: A Compositional Generalization Challenge Based on Semantic Interpretation (2020, background): “Whether un-bounded recursion should be considered as a part of machinery that governs language is a debated issue, the evidence against being the signiﬁcantly degraded human parsing performance on multiply-nested structures (Christiansen and Chater, 1999).”
Expectation-based syntactic comprehension. (2008, methodology): “Many connectionist models of online sentence processing (including Elman 1990, 1991; Christiansen and Chater 1999; Tabor and Tanenhaus 1999; Konieczny and Döring 2003; and a component of the model of Rohde 2002) propose prediction-based metrics of the difficulty of a word wi in its sentence context ”
Doing without schema hierarchies: a recurrent connectionist approach to normal and impaired routine sequential action. (2004, background): “…Hare, Day, & Tyler, 1994), reading (Pacton, Perruchet, Fayol, & Cleeremans, 2001; Plaut, 1999), sentence processing (Allen & Seidenberg, 1999; Christiansen & Chater, 1999; Rohde, 2002; Rohde & Plaut, 1999), implicit learning (Cleeremans, 1993; Cleeremans & McClelland, 1991), dynamic decision…”
Accounting for regressive eye-movements in models of sentence processing: A reappraisal of the Selective Reanalysis hypothesis (2008, background): “…on-line empirical phenomena in the field such as the word-by-word or phrase-by-phrase reading times at different points of a sentence (e.g., Christiansen & Chater, 1999, 2001; Green & Mitchell, 2006; Konieczny & Döring, 2003; Levy, 2008; Lewis, 1993; Lewis & Vasishth, 2005; MacDonald &…”
More than words: Frequency effects for multi-word phrases (2010, unknown): “…to grammar (Bybee, 1998; Goldberg, 1995, 2006; Langacker, 1987, 1988; Tomasello, 2003), connectionist models of learning and processing (e.g. Christiansen & Chater, 1999; Elman, 1990, 1991; MacWhinney, 1998; Rumelhart & McClelland, 1986; Seidenberg, 1994), and exemplar models of…”
Strong systematicity through sensorimotor conceptual grounding: an unsupervised, developmental approach to connectionist sentence processing (2012, methodology): “Following Christiansen and Chater (1999), GPE was defined as the proportion of nodes that predict transitions that are not represented or deviate from the grammar:”
Architectural Bias in Recurrent Neural Networks: Fractal Analysis (2002, methodology): “However, it has been known for some time that when training RNNs to process symbolic sequences, activations of recurrent units display a considerable amount of structural diﬁerentiation even prior to learning (Christiansen & Chater, 1999; Kolen, 1994; Kolen, 1994a; Manolios & Fanelli, 1994).”
Representing word meaning and order information in a composite holographic lexicon. (2007, background): “SRNs can predict the next word in a sequence (Elman, 1990), can identify embedded structure (Elman, 1991) and recursion (Christiansen & Chater, 1999), and can track limited long-range dependencies (Elman, 1995).”
Memory interference during language processing. (2001, background): “They use a simple recurrent network (Christiansen & Chater, 1999b; Elman, 1990) in which learning and processing occur through the connections between simple elements without explicit symbols or rules.”
Syntactic structure assembly in human parsing: a computational model based on competitive inhibition and a lexicalist grammar. (2000, methodology): “† Distributed-connectionist models, based on various types of neural network architectures, in particular on Simple Recurrent Nets (Chater & Christiansen, 1999; Christiansen & Chater, 1999; Elman, 1991; Tabor, Juliano & Tanenhaus, 1997).”
Learn more by training less: systematicity in sentence processing by recurrent networks (2006, background): “4.1 The echo state network Christiansen and Chater (1999) studied the ability of SRNs to process different types of recursive sentence structures and found that even before training these networks were sensitive to structural differences among the sentences.”
Context-free and context-sensitive dynamics in recurrent neural networks (2000, background): “As with human linguistic performance, the network performance is not perfect and degrades with higher m (Christiansen and Chater, 1999).”
Simple Recurrent Networks Learn Context-Free and Context-Sensitive Languages by Counting (2001, background): “The main contribution of this article is to demonstrate that counter solutions for some CFLs and CSLs are reachable with sufficient training.”
On the computational basis of learning and cognition: Arguments from LSA (2002, background): “Neural net models of language take us some distance beyond that level of understanding, primarily by proving that some portions of the process are in principle learnable or executable in some fashion (see, e.g., Christiansen and Chater, 1999; Seidenberg.”
RNNs as psycholinguistic subjects: Syntactic state and grammatical dependency (2018, background): “While we are not interested in directly modeling human processing difﬁculty here, we note that there is a long tradition linking RNN performance to human language processing (El-man, 1990; Christiansen and Chater, 1999; Mac-Donald and Christiansen, 2002; Frank and Bod, 2011).”