|
|
xii | |
Preface |
|
xiv | |
|
1 Connectionist Natural Language Processing |
|
|
1 | (14) |
|
|
|
|
1 | (4) |
|
|
5 | (4) |
|
|
9 | (1) |
|
|
9 | (6) |
|
|
|
|
15 | (6) |
|
2 Distributed Symbol Discovery through Symbol Recirculation: Toward Natural Language Processing in Distributed Connectionist Networks |
|
|
21 | (28) |
|
|
|
|
|
21 | (1) |
|
Natural language processing: Constraints from the task domain |
|
|
22 | (3) |
|
Dynamic vs. static symbol representations |
|
|
25 | (2) |
|
|
27 | (2) |
|
Encoding semantic networks in DUAL: A distributed connectionist architecture |
|
|
29 | (9) |
|
Other symbol recirculation methods |
|
|
38 | (2) |
|
|
40 | (4) |
|
Variable binding research and symbol formation |
|
|
44 | (1) |
|
|
45 | (1) |
|
|
45 | (1) |
|
|
45 | (4) |
|
3 Representing Meaning Using Microfeatures |
|
|
49 | (26) |
|
|
|
49 | (2) |
|
Microfeature representations in PARROT |
|
|
51 | (6) |
|
Implementation of the microfeature concept within PARROT |
|
|
57 | (5) |
|
|
62 | (1) |
|
Discussion and next steps |
|
|
62 | (5) |
|
|
67 | (1) |
|
|
67 | (1) |
|
Appendix I Outline of the PARROT system |
|
|
68 | (2) |
|
Appendix II Example entries from the lexicon |
|
|
70 | (5) |
|
4 Noun Phrase Analysis with Connectionist Networks |
|
|
75 | (22) |
|
|
|
|
75 | (1) |
|
|
76 | (3) |
|
Learning level: Learning semantic prepositional relationships |
|
|
79 | (5) |
|
Integration level: Integration of semantic and syntactic constraints |
|
|
84 | (3) |
|
A case study for the disambiguation of noun phrases |
|
|
87 | (3) |
|
|
90 | (3) |
|
|
93 | (1) |
|
|
93 | (1) |
|
|
93 | (4) |
|
5 Parallel Constraint Satisfaction as a Comprehension Mechanism |
|
|
97 | (42) |
|
|
|
|
97 | (2) |
|
|
99 | (20) |
|
|
119 | (14) |
|
|
133 | (1) |
|
|
134 | (1) |
|
|
134 | (2) |
|
Appendix I Input and output representations |
|
|
136 | (3) |
|
|
|
|
139 | (2) |
|
|
141 | (2) |
|
6 Self-correcting Connectionist Parsing |
|
|
143 | (26) |
|
|
Introduction: Constrained chaos |
|
|
143 | (5) |
|
|
148 | (6) |
|
|
154 | (3) |
|
|
157 | (6) |
|
|
163 | (2) |
|
|
165 | (1) |
|
|
166 | (1) |
|
|
167 | (2) |
|
7 A Net-linguistic "Earley" Parser |
|
|
169 | (40) |
|
|
|
|
169 | (1) |
|
The basic characteristics of the parser |
|
|
170 | (8) |
|
The representation of parse-information |
|
|
178 | (5) |
|
The Earley parse-list algorithm |
|
|
183 | (4) |
|
|
187 | (15) |
|
|
202 | (1) |
|
|
203 | (6) |
|
PART III REPRESENTATIONAL ADEQUACY |
|
|
|
|
209 | (2) |
|
|
211 | (2) |
|
8 The Demons and the Beast---Modular and Nodular Kinds of Knowledge |
|
|
213 | (40) |
|
|
|
213 | (1) |
|
Structure and habits---The knowledge and the power |
|
|
214 | (6) |
|
A model that learns some morphology |
|
|
220 | (8) |
|
Evidence for nodes unseen---some models that learn to read aloud |
|
|
228 | (12) |
|
The study of statistically available information |
|
|
240 | (9) |
|
Conclusion---Behavioural strategies and mental structures |
|
|
249 | (1) |
|
|
250 | (1) |
|
|
250 | (3) |
|
9 Representational Adequacy and the Case for a Hybrid Connectionist/Marker-parsing Model |
|
|
253 | (20) |
|
|
|
253 | (1) |
|
Representational adequacy |
|
|
254 | (3) |
|
Autonomous semantic networks |
|
|
257 | (3) |
|
ASNs and representational adequacy |
|
|
260 | (7) |
|
|
267 | (3) |
|
|
270 | (1) |
|
|
271 | (1) |
|
|
271 | (2) |
|
10 A Step Toward Sub-symbolic Language Models without Linguistic Representations |
|
|
273 | (44) |
|
|
|
273 | (2) |
|
Basic observations about language |
|
|
275 | (9) |
|
|
284 | (2) |
|
Categorisation and concept formation |
|
|
286 | (11) |
|
|
297 | (7) |
|
|
304 | (1) |
|
|
304 | (5) |
|
|
309 | (1) |
|
|
310 | (1) |
|
|
311 | (1) |
|
|
311 | (2) |
|
|
313 | (4) |
|
PART IV COMPUTATIONAL PSYCHOLINGUISTICS Introduction |
|
|
317 | (144) |
|
|
319 | (2) |
|
11 Connectionist Models of Speech Perception |
|
|
321 | (30) |
|
|
|
321 | (2) |
|
Phonological constraints in speech perception |
|
|
323 | (5) |
|
Interactive activation models |
|
|
328 | (10) |
|
Auditory/visual speech perception |
|
|
338 | (5) |
|
Feedforward connectionist models |
|
|
343 | (5) |
|
|
348 | (1) |
|
|
348 | (1) |
|
|
348 | (3) |
|
12 Connectionism: A New Breed of Bottom-up Model? |
|
|
351 | (22) |
|
|
|
351 | (1) |
|
Interactive activation models |
|
|
352 | (3) |
|
A simple network for word and letter recognition |
|
|
355 | (3) |
|
Elman and McClelland's study |
|
|
358 | (2) |
|
A dynamic net model of word recognition |
|
|
360 | (7) |
|
|
367 | (3) |
|
|
370 | (3) |
|
13 Models of Form-related Priming in Comprehension and Production |
|
|
373 | (36) |
|
|
|
|
|
|
373 | (1) |
|
|
374 | (3) |
|
|
377 | (27) |
|
|
404 | (1) |
|
|
405 | (1) |
|
|
406 | (3) |
|
14 Reading with Attentional Impairments: A Brain-damaged Model of Neglect and Attentional Dyslexias |
|
|
409 | (52) |
|
|
|
|
409 | (2) |
|
|
411 | (10) |
|
Simulations of neglect dyslexia |
|
|
421 | (22) |
|
|
443 | (3) |
|
|
446 | (3) |
|
|
449 | (1) |
|
|
449 | (3) |
|
Appendix 1 PO net dynamics |
|
|
452 | (2) |
|
|
454 | (2) |
|
Appendix 3 Details of AM simulations |
|
|
456 | (1) |
|
Appendix 4 Details of BLIRNET and PO net simulations |
|
|
457 | (4) |
Author Index |
|
461 | (8) |
Subject Index |
|
469 | |