() ( ) Modified on 2009/05/24, 2008/09/17, 15, 12, 11, 10, 09 Created on 2008/07/02 1 1) () ( ) (exgen Excel VBA ) 2)3) 1.1 ( ) () : : (1) ( ) ( ) (2) / (1) (= ) (2) (= () =) 4)5) () ( ) () (=) (1) : ( ) (=) (2) / : ( ) ( ) 1) 9 (2008/09/16) exgen-macros URL: http://clsl.hi. h.kyoto-u.ac.jp/ kkuroda/tools/ exgen-macros-current.xls 3) exgen (exgen-samples.xls) URL: http: //clsl.hi.h.kyoto-u.ac.jp/ kkuroda/ tools/exgen-samples-current.xls 4) Web : 5) [9] i) (Bayes Bayes ) ii) 1
1 ( ) : / ( ) ( ) ( ) ( ) 1.2 X1 X2 X3 V X1={,,... }, X2={,,... },... 2 exgen-macros-vn.xls (n ) 2.1 (3) a. (e.g., X2 X1 V ) (= ) b. (e.g., X1, X2, V) (e.g., X2 = {,,... }) 6) A1 Pattern Template 1 2.2 (4) (3) a. Excel [] [] Generate Examples b. generated-expressions c. 1 2 3 1: 3.1 (5) a. b. 6) 65,000 2
2 : 120 (= 2 6 5 1 2) [] c. d. (6) X2 X1 3.2 (8) : (5) (6) X1, X2 (7) a. Pattern = { X2 X1 V } b. X1 = {,,, } c. X2 = {,,, } d. V = {, } X1, X2 (= ) 3 7) 7) X1= (8) a. 1.0 [Perfect = Perfectly Good]: e b. 0.5 [Fair = Fairly Good]: e e c. 0.1 [Poor = Hard to accept]: e e (= ) X1 3
3 ( exgen Index, Equiv, Acceptability ) d. 0.0 [Bad = Unacceptable]: e e ( ) 1 (= 0) 8) 1.0 0.1 1.0 0.5 9) : (9) : (10) 8) 9) 0.5 1.0 4
4 4.1 4.1.1 : (11) / ( ) e e A(e) e i A(e i ) e j A(e j ) d = A(e i ) A(e j ) 4.1.2 10) n E= { e 1, e 2,..., e n } A Ā (e i, e j ) (i) A(e i ) < A(e j ) (ii) A(e i ) A(e j ) (iii) A(e i ) > A(e j ) ( ( ) ) ( A(e i ) < A(e j ) A(e j ) < A(e k ) A(e i ) < A(e k ) ) ( 10) 2009/05/24 ) () 4.2 (= ) (= ) ( ) ( ) : (12) e A(e) a. e ( ) L (e.g., L, L) m(l) b. e C m(c) f (m(l), m(c)) A(s) = f (m(l), m(c)) (13) A(s) [0, 1] ( a. A(s) = 0 b. A(s) = 1.0 ) 11) F (Fisher ) 12) e ( ) 0 < L (e) < 1 e ( ) 13) 0 < C (e) < 1 F 2 L(e) C(e) L (e)+c (e) A(e) ( L 0 C 0): (14) a. L(e) = C(e) = 1.0 A(e) = 1.0 b. L(e) = C(e) = 0.5 A(e) = 0.5 14) c. L(e) = 0.5 C(e) = 1.0 A(e) 0.67 d. L(e) = 0.1 C(e) = 0.1 A(e) 0.1 ( ) 11) A(e) = 1.0 A(e) = 0 = 1.0 0 A(e) = 1.0 A(e) = 0 12) M(L) M(C) 13) L(e) M(L(e)) e e M(C) (degree of fitness) 14) 0 < d 1.0 L(e) = C(e) = d A(e) = d 5
4.3 exgen : (15) a. ( 1.0 10% 30% ) b. ( 15) ) (15a) 5 X1 X2 X3 V (16) ( 4 ): (16) a. b. c. d. e. 125 1.0 [= ] 13.6% 0.5 [= ] 16.8% 0.1 [=] 15.2% 0 [= ] 54.4% 1.0 0.5 30% 16) 15) 16) exgen (15) exgen 4.4 4.4.1 (15) (risk of overgeneration) ( (15a) ) ( ) 17) : ( ) 4.4.2? (collocations) ( ) 18) 17) 18) L. Wittgenstein ( ) Z. S. Harris (distributional hypothesis) [4] 6
4 exgen 5 : X1 X2 X3 V 125 (= 1 5 5 5) {1.0, 0.5, 0.1, 0 } 7
19) ( ) 4.5 2 (generation of expressions) (lexical insertions) [1, 6] (= ) 20) 4.5.1 (usage-based model of grammar) 6 exgen exgen ( ) 6 7 (replacement) (vacuous) 6 (subcategorization) * exgen i) ii) / 21) 22) ( ) ( ) 4.5.2 (selectional restrictions/preferences) Wittgenstein 19) ( ) 20) Minimalist Program [2] 21) n ( ) 22) Z. S. Harris [5] ( ( ) ) 8
6 (nesting) 7 6 nesting [vacuous (subcategorization) ] 9
exgen MDS PCA (Principal Component Analysis) ICA (Independent Component Analysis) 23) ( exgen ) (17) a. () ( ) b. () ( ) (17a) (17b) : (mechanism of linguistic memories) ( [3] [7, 8] ) ( ) s (i) s (ii) s A : (18) a. ((Pattern: X1 V1 ; Pattern: X1 X2 V1 ;... ), (X1: ; X1: ;... ), (X2: ;... ), (V1: ; V1: ;... ),... ) b. (Cartesian product) P c. P 24) 5 exgen 23) [14, 13, 10] 24) [11] [11] [12] [1] N. Chomsky. Aspects of the Theory of Syntax. MIT Press, Cambridge, MA, 1965. [2] N. Chomsky. The Minimalist Program. MIT Press, Cambridge, MA., 1995. [3] W. Daelemans and A. van den Bosch. Memory-based Natural Language Processing. Cambridge Unversity Press, Cambridge, UK, 2005. [4] Z. S. Harris. Distributional structure. Word, 10(2-3):146 162, 1954. Reprinted in Fodor, J. A and Katz, J. J. (eds.), Readings in the Philosophy of Language, pp. 33 49. Englewood Cliffs, NJ: Prentice-Hall. [5] Z. S. Harris. A Theory of Language and Information. Oxford University Press, 1991. [6] J. D. McCawley. Prelexical syntax. In P. A. M. Seuren, editor, Semantic Syntax, pages 29 42. Oxford University Press, London, 1971. 10
[7] R. Port. Toward a rich phonology, 2006. A 4-page summary of the argument in Words, symbols and rich memory, prepared for the ESCA Experimental Linguistics Conference in Athens Greece, Sep., 2006. Available online at http://www.cs.indiana.edu/ port/ pap/toward_rich_phonology.final.pdf. [8] R. Port. How are words stored in memory? beyond phones and phonemes. New Ideas in Psychology, 25(2):143 170, 2007. The original version, entitled Words, symbols and rich memory, available online at http://www.cs.indiana. edu/ port/pap/words.symbols.rich. memory.snglsp.aug8.pdf]. [9] A. Stefanowitsch. Negative entrenchment: A usagebased approach to negative evidence. Cognitive Linguistics, 19(3):513 531, 2008. [10] and. y x :. In 6 Conference Handbook, pages 157 160, 2005. [11]. [x ]. http: //clsl.hi.h.kyoto-u.ac.jp/ kkuroda/ papers/remarks-on-x-katta.pdf, 2007. [12]. :., 36(11):24 34, 2007. : http://clsl.hi. h.kyoto-u.ac.jp/ kkuroda/papers/ la-with-rich-memory-full.pdf. [13],,, and. : x y. In 22, pages 253 55 (Q 38), 2005. [ : http://clsl. hi.h.kyoto-u.ac.jp/ kkuroda/papers/ frames-attract-readings-jcss22.pdf]. [14],,,, and. : Berkeley FrameNet and Beyond. In 5, pages 558 578, 2005. 11