kow kuroda's talk at jcla 16 workshop: methodology in cognitive linguistics, 2015
TRANSCRIPT
-
(EBM)
*
162015-09-12 (),
*9 2015/09/06
-
jcla16 ws room3
()
2
-
jcla16 ws room3
NICT/
open
closed
3
-
jcla16 ws room3
()
(EBM)
EBM
EBM
4
-
jcla16 ws room3
9 2015/09/06
URL
clsl.hi.h.kyoto-u.ac.jp/~kkuroda/presentations/kuroda-2015-JCLA16WS-slides.pdf
5
http://clsl.hi.h.kyoto-u.ac.jp/~kkuroda/presentations/kuroda-2015-JCLA16WS-slides.pdf
-
()
-
jcla16 ws room3
=>
7
-
jcla16 ws room3
() 1/3
(e.g., NLP, ) ()
8
-
jcla16 ws room3
() 2/3
(observation) (description) (exaplantion, account)
()
9
-
jcla16 ws room3
() 3/3() (overly) intuition-based authority-based
1, 2
10
-
jcla16 ws room3
(evidence-based medicine: EBM)
(())
11
-
jcla16 ws room3
()
12
-
Evidence-based Medicine
-
jcla16 ws room3
EBM 1/2EBM ()
()
(confounding)
(e.g., )
14
-
jcla16 ws room3
EBM 2/2EBM
15
-
jcla16 ws room3
[1] (vitamin D)
[2] Ignaz Philipp Semmelweiss* (1818-1815)
1 ()
17
*Semmelweis Ignc Flp
-
jcla16 ws room3
1/4 , pp. 82-84
18
-
jcla16 ws room3
2/4 , pp. 82-84
19
-
jcla16 ws room3
3/4 , pp. 82-84
20
-
jcla16 ws room3
4/4 , pp. 82-84
21
-
jcla16 ws room3
1/2 , p. 85
22
-
jcla16 ws room3
2/2 , p. 85
23
-
jcla16 ws room3
1/2
24
-
jcla16 ws room3
2/2
25
-
jcla16 ws room3
Archibold Cochrane (19091988)*
:
:
27
*Tim Harford TED : Trial, error and the Gold complex.
-
jcla16 ws room3
diagnosis
Atul Gawande: Complications.
Lisa Sander: Every Patient Tells a Story.
Jerome Groopman: How Doctors Think.
(big data)
28
-
Evidence-based Medicine
-
jcla16 ws room3
EBM
EBM
: = : /
30
-
jcla16 ws room3
EBM
31
Grade
A
B
C1
C2
D
-
jcla16 ws room3
EBM A, B, C1, C2, D
32
-
jcla16 ws room3
EBM/
33
Level
1a
+ +
1b1
+ +
2a
prospective/concurrent cohort
study +
2b
retrospective/historical cohort
study +
3 retrospective study +
4
5 6
-
jcla16 ws room3
EBM 1a, 1b, 2a, 2b, 3, 4, 5, 6
34
-
jcla16 ws room3
Level 6 Level 5
Level 3
Level 3
35
-
jcla16 ws room3
Seven alternatives to EBM
36
Basis for decisions Marker Measuring device
Unit of measurement
Evidence Ramdomized controlled trial Meta-analysis Odds ratio
Eminence Radiance of white hair LuminometerOptical density
Vehemence Level of stridency Audiometer Decibels
Eloquence (or elegance) /
Smoothness of tongue or nap of
suitTeflometer Adhesion score
Providence Level of religious fervorSextant to measure
angle of genuflaection
International units of
piety
Diffidence Level of gloom Nihilometer Sighs
Nervousness Litigation phobia levelEvery conceivable
testBank
balanceConfidence Bravado Sweat test No sweat
-
jcla16 ws room3
Seven alternatives to EBM 1/2
Isaacs Fitzgerald (1999) EBM :
Eminence based medicine The more senior the colleague, the less importance he or she placed on the need for anything as mundane as evidence. Experience, it seems, is worth any amount of evidence. These colleagues have a touching faith in clinical experience, which has been defined as making the same mistakes with increasing confidence over an impressive number of years. The eminent physicians white hair and balding pate are called the halo effect.
Vehemence based medicine The substitution of volume for evidence is an effective technique for brow beating your more timorous colleagues and for convincing relatives of your ability.
37
-
jcla16 ws room3
Seven alternatives to EBM 2/3
Isaacs Fitzgerald (1999) EBM :
Eloquence based medicineThe year round suntan, carnation in the button hole, silk tie, Armani suit, and tongue should all be equally smooth. Sartorial elegance and verbal eloquence are powerful substitutes for evidence.
Providence based medicine If the caring practitioner has no idea of what to do next, the decision may be best left in the hands of the Almighty. Too many clinicians, unfortunately, are unable to resist giving God a hand with the decision making.
38
-
jcla16 ws room3
Seven alternatives to EBM 3/3
Isaacs Fitzgerald (1999) EBM :
Diffidence based medicine Some doctors see a problem and look for an answer. Others merely see a problem. The diffident doctor may do nothing from a sense of despair. This, of course, may be better than doing.
Nervousness based medicine Fear of litigation is a powerful stimulus to overinvestigation and overtreatment. In an atmosphere of litigation phobia, the only bad test is the test you didnt think of ordering.
Confidence based medicine This is restricted to surgeons (table).
39
-
jcla16 ws room3
EBM7
40
Classification
Eminence-based
/
Vehemence-based
Eloquence (or elegance)-
based
Providence-based
Diffidence-based
Nervousness-based
Confidence-based
-
jcla16 ws room3
Confidence-based surgeon
41
Classification
Eminence-based
/
Vehemence-based
Eloquence (or elegance)-
based
Providence-based
Diffidence-based
Nervousness-based
Confidence-based
/
-
jcla16 ws room3
EBM Evidence-based Linguistics: EBL
EBL
42
-
Evidence-based Linguistics
-
jcla16 ws room3
EBM 1/3
1.
44
-
jcla16 ws room3
EBM 2/32.
EBM
Level 6
45
-
jcla16 ws room3
EBM 3/3
()
46
-
jcla16 ws room3
EBL 1/2
EBL
positives negatives
acceptability judg(e)ment ()
A = f(E x C x R) [A, E, C, R , , , ()]
47
-
jcla16 ws room3
EBL 2/2
=
48
-
jcla16 ws room3
(e.g, )
()
()
indirect negative evidence
(Stefanowitsch 2006, 2008; Clark and Lappin 2009; Yang to appear) =
50
-
jcla16 ws room3
1/3, p. 2
=>
51
-
jcla16 ws room3
2/3
(10)
()
()
()
()
()
, Loc. 1003
52
-
jcla16 ws room3
3/3
()
()
()
()
, Loc. 1003
53
-
jcla16 ws room3
1/2
()
=
54
-
jcla16 ws room3
2/2
()
() W C W
C W
(/)
55
-
EBM
-
jcla16 ws room3
(i) (ii)
1. D
D = BCCWJ
2. D S
3. S T
4. S U
5. T U
A: T = U
B: T U
C: T U
CMTA, B, C
3 4
57
-
jcla16 ws room3
I. , II. , III.
EBM
EBM EBL
58
-
Thank you for your attention.
Dkuji vm za pozornost. Ksznm a figyelmeted. .
-
jcla16 ws room3
References 1/2 (1999): . .
, (2011). (): . In 17, 11391142.
(2012). In , Vol. 27, No. 3 , pp. 326332. 2012.
(2015). : . , .
, D. (2014): . . [Gilbert, D. (2010): Stumbling on Happiness. ]
, K. (2011): . . [Fung (2010): Numbers Rule Your World}, 1st edition. McGraw-Hill Education, 2010]
, M. (2009). . . [McCarthy, Michael (2004). Bayesian Methods for Ecology. ]
, K. J. (2004): : . . [Rothman, Kenneth J. (2002): Epidemiology: An Introduction. ]
60
-
jcla16 ws room3
References 2/2Cochrane, Archibold and Max Blythe (1989). One Mans Medicine: An Autobiography of Professor Archie Cochrane. BMJ Publishing Group.
Isaacs, D. and D. Fitzgerald: Seven alternatives to evidence based medicine. British Medical Journal 319: 1618, 1999.
Kuroda, Kow (2009). Pattern lattice as a model for linguistic knowledge and performance. In Proc. of the 23rd Pacific Asia Conference on Language, Information and Computation, Vol.~1, pp. ,278287.
Stefanowitcsh, Anatol (2006). Negative evidence and the raw frequency fallacy. Corpus Linguistics and Linguistic Theory, 2(1): 61--77.
Stefanowitsch, Anatol (2008). Negative entrenchment: A usage-based approach to negative evidence. Cognitive Linguistics, 19(3): 513531.
Yang, Charles (to appear). Negative knowledge from positive evidence. Language, ??(??):??.
61
-
I, II, III, IV
-
I
-
jcla16 ws room3
64
-
jcla16 ws room3
1/3
65
-
jcla16 ws room3
2/3 (=)
() ()
66
-
jcla16 ws room3
3/3 ()
Part I 67
-
jcla16 ws room3
=>
68
-
jcla16 ws room3
() 1/3
(e.g., NLP, ) ()
69
-
jcla16 ws room3
(aka )
70
-
jcla16 ws room3
()
NLP
()
71
-
jcla16 ws room3
() 2/3
(observation) (description) (exaplantion, account)
()
72
-
jcla16 ws room3
N. Chomsky ()
73
-
jcla16 ws room3
, ,
D
D (proper classification) C
C (proper representation) R
D C/R =
C/D C/D
74
-
jcla16 ws room3
, , 1/3
(confirmation bias)
() T
T
() T
75
-
jcla16 ws room3
1/3 [][] () () ()
DanielGilbert,Loc.1818
[K.K.:]
76
-
jcla16 ws room3
2/3[][][] [Sir Francis Bacon] []
DanielGilbert,Loc.1822
77
-
jcla16 ws room3
3/3 [ Cicero ][][][][]
DanielGilbert,Loc.1818
78
-
jcla16 ws room3
1HIV/
: , p. 84.
79
m 38 37 75
+ n 1 7 8
N = m+n 39 38 77
m/N 0.026 0.184 0.104
-
jcla16 ws room3
1/2
T (e.g., ) (e.g., B3) T
(B2)
positives negatives
(B2)
false positives false positives
80
-
jcla16 ws room3
2/2
Cicero/Bacon
81
-
jcla16 ws room3
2a 1/n
B2:
B3:
C2:
C3:
82
+
m 17 11 28
n 13 9 22
N = m+n 30 20 50
m/N 0.567 0.550 0.560
-
jcla16 ws room3
2b 2/n
B2:
B3:
C2:
C3:
83
+
m 25 17 42
n 5 3 8
N = m+n 30 20 50
m/N 0.833 0.850 0.840
-
jcla16 ws room3
2c 3/n
B2:
B3:
C2:
C3:
84
+
m 25 3 28
n 5 17 22
N = m+n 30 20 50
m/N 0.833 0.150 0.560
-
jcla16 ws room3
2a, b, c Cicero/Bacon 2c
2b
85
-
jcla16 ws room3
1 1/3P: ()
x [ metaphor(x) => cross-domain-mapping(x) ]
P: x x
x [ cross-domain-mapping(x) => metaphor(x)]
86
-
jcla16 ws room3
1 2/3 S, T
S: x [ cross-domain-mapping(x) => metaphor(x) ]
T: x [ cross-domain-mapping(x) => metaphor(x) ]
(P) P
87
-
jcla16 ws room3
1 3/3
P: x x
x [ metaphor(x) => cross-domain-mapping(x)]
P: x x
x [cross-domain-mapping(x) => metaphor(x)]
88
-
jcla16 ws room3
, , 2/3
exclusive classification = hard clustering
A, B, AB, O
89
-
jcla16 ws room3
ABO
Venn (matrix)Hasse
90
+A A
+B AB B
B A O
U
+A +B
: A: B: AB: O
Venn
Hasse
-
jcla16 ws room3
tree structure
91
-
jcla16 ws room3
Formal Concept Analysis (FCA) /
Ganter, B. and R. Wille (1999). Formal Concept Analysis: Mathematical Foundations. Berlin: Springer-Verlag (Translated by C. Franzke).
Ganter, B., G. Stumme, and R. Wille (eds.) (2005). Formal Concept Analysis: Foundations and Applications. Berlin/Heidelberg: Springer.
FCA Concept Explorer 1.3
http://conexp.sourceforge.net/
java -Dfile.encoding=utf-8 -jar conexp.jar
92
http://conexp.sourceforge.net/
-
jcla16 ws room3
, , 3/3
(= )
D C/R
overgeneralizations
93
-
jcla16 ws room3
Construction Grammar
2
Remarks
=
94
-
jcla16 ws room3
A.
B.
A (=)
= ,
B
95
-
jcla16 ws room3
() 2/3
() () () ()
()
96
-
jcla16 ws room3
() 3/3() (overly) intuition-based authority-based
1, 2
97
-
Evidence-based Linguistics
II
-
jcla16 ws room3
EBL
1.
=
2.
99
-
jcla16 ws room3
EBL 1/2
1
R
R () (=)
100
-
jcla16 ws room3
1/31
??
1
a.
b.
c.
d. ??
2
a.
b.
3
a.
b. ?*
c.
d. ?
(bracketing paradox)
101
-
jcla16 ws room3
2/3 1
a. {i. ; ii. ; iii. ; vi. ; v. ; vi. }
b. {i. ; ii. ; iii. }
c.
2
a. {i. ; ii. ; iii. }
b.
c.
102
-
jcla16 ws room3
3/3RT
a.
b. ?
c. ?
d.
b, c ()
103
-
jcla16 ws room3
pattern lattice (PLM)
PLM Kuroda (2009), (2011)
104
-
jcla16 ws room3
EBL 2/2
/ acceptability judg(e)ment =
S
NICT (2008-2010)
105
-
jcla16 ws room3
1 1/6F1
a.
b.
c.
d.
F2
a.
b.
c.
d.
G1
a.
b.
c.
d.
G2
a.
b.
c.
d.
106
-
jcla16 ws room3
1 2/6
16 4 F1, F2, G1, G2 12
i.
ii.
iii.
iv.
i, ii, iii, iv 3, 2, 1, 0
107
-
jcla16 ws room3
1.
2. N=16
3. w
1 108
-
jcla16 ws room3
1 3/6
109
1: E x R r av. stdev
Expression Index r1 r2 r3 r4 r5 r6 r7 r8 r9 r10 r11 r12 r13 r14 r15 r16 av. stdev F1a 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 0.00 G2c 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 1.00 2.88 0.50 G2b 2.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 1.00 2.81 0.54 F2a 3.00 3.00 3.00 3.00 3.00 3.00 3.00 2.00 2.00 3.00 2.00 3.00 2.00 2.00 3.00 2.00 2.63 0.50 F2b 3.00 3.00 3.00 3.00 3.00 3.00 2.00 3.00 2.00 2.00 1.00 3.00 2.00 2.00 2.00 3.00 2.50 0.63 F1b 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 1.00 2.00 2.00 2.00 2.00 1.00 1.00 3.00 2.38 0.81 G1c 3.00 2.00 3.00 3.00 3.00 1.00 3.00 1.00 2.00 2.00 1.00 3.00 2.00 1.00 3.00 1.00 2.13 0.89 F2c 3.00 2.00 3.00 1.00 1.00 1.00 2.00 3.00 2.00 3.00 3.00 2.00 1.00 2.00 2.00 2.00 2.06 0.77 F1c 3.00 3.00 3.00 1.00 2.00 2.00 3.00 3.00 1.00 2.00 1.00 1.00 1.00 1.00 2.00 3.00 2.00 0.89 F2d 3.00 3.00 3.00 2.00 3.00 2.00 2.00 2.00 2.00 1.00 1.00 3.00 1.00 1.00 1.00 2.00 2.00 0.82 G1b 3.00 3.00 3.00 3.00 3.00 2.00 3.00 1.00 2.00 0.00 2.00 1.00 3.00 1.00 1.00 1.00 2.00 1.03 G2d 1.00 3.00 3.00 2.00 3.00 2.00 3.00 3.00 3.00 1.00 2.00 2.00 1.00 2.00 0.00 1.00 2.00 0.97 F1d 3.00 3.00 2.00 1.00 3.00 3.00 2.00 1.00 2.00 1.00 2.00 2.00 0.00 1.00 0.00 2.00 1.75 1.00 G1d 3.00 2.00 2.00 2.00 3.00 1.00 2.00 1.00 2.00 1.00 1.00 2.00 1.00 1.00 1.00 3.00 1.75 0.77 G1a 3.00 1.00 0.00 1.00 1.00 3.00 2.00 1.00 1.00 0.00 0.00 0.00 1.00 3.00 3.00 3.00 1.44 1.21 G2a 2.00 1.00 0.00 1.00 1.00 1.00 1.00 1.00 1.00 0.00 2.00 1.00 0.00 1.00 3.00 1.00 1.06 0.77
av. 2.75 2.56 2.50 2.19 2.56 2.25 2.50 2.13 2.00 1.69 1.81 2.13 1.63 1.75 1.94 2.00stdev 0.58 0.73 1.03 0.91 0.81 0.86 0.63 0.96 0.73 1.14 0.91 0.96 1.02 0.86 1.12 0.89
-
jcla16 ws room3
1 4/6
110
2: R x E F av. stdev
Rater F1a G2c G2b F2a F2b F1b G1c F2c F1c F2d G2d G1b G1d F1d G1a G2a av stdevr3 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 2.00 2.00 0.00 0.00 2.50 1.03r5 3.00 3.00 3.00 3.00 3.00 3.00 3.00 1.00 2.00 3.00 3.00 3.00 3.00 3.00 1.00 1.00 2.56 0.81r4 3.00 3.00 3.00 3.00 3.00 3.00 3.00 1.00 1.00 2.00 2.00 3.00 2.00 1.00 1.00 1.00 2.19 0.91r2 3.00 3.00 3.00 3.00 3.00 3.00 2.00 2.00 3.00 3.00 3.00 3.00 2.00 3.00 1.00 1.00 2.56 0.73r6 3.00 3.00 3.00 3.00 3.00 3.00 1.00 1.00 2.00 2.00 2.00 2.00 1.00 3.00 3.00 1.00 2.25 0.86
r12 3.00 3.00 3.00 3.00 3.00 2.00 3.00 2.00 1.00 3.00 2.00 1.00 2.00 2.00 0.00 1.00 2.13 0.96r7 3.00 3.00 3.00 3.00 2.00 3.00 3.00 2.00 3.00 2.00 3.00 3.00 2.00 2.00 2.00 1.00 2.50 0.63
r10 3.00 3.00 3.00 3.00 2.00 2.00 2.00 3.00 2.00 1.00 1.00 0.00 1.00 1.00 0.00 0.00 1.69 1.14r15 3.00 3.00 3.00 3.00 2.00 1.00 3.00 2.00 2.00 1.00 0.00 1.00 1.00 0.00 3.00 3.00 1.94 1.12r8 3.00 3.00 3.00 2.00 3.00 3.00 1.00 3.00 3.00 2.00 3.00 1.00 1.00 1.00 1.00 1.00 2.13 0.96
r13 3.00 3.00 3.00 2.00 2.00 2.00 2.00 1.00 1.00 1.00 1.00 3.00 1.00 0.00 1.00 0.00 1.63 1.02r9 3.00 3.00 3.00 2.00 2.00 1.00 2.00 2.00 1.00 2.00 3.00 2.00 2.00 2.00 1.00 1.00 2.00 0.73
r14 3.00 3.00 3.00 2.00 2.00 1.00 1.00 2.00 1.00 1.00 2.00 1.00 1.00 1.00 3.00 1.00 1.75 0.86r11 3.00 3.00 3.00 2.00 1.00 2.00 1.00 3.00 1.00 1.00 2.00 2.00 1.00 2.00 0.00 2.00 1.81 0.91r1 3.00 3.00 2.00 3.00 3.00 3.00 3.00 3.00 3.00 3.00 1.00 3.00 3.00 3.00 3.00 2.00 2.75 0.58
r16 3.00 1.00 1.00 2.00 3.00 3.00 1.00 2.00 3.00 2.00 1.00 1.00 3.00 2.00 3.00 1.00 2.00 0.89av 3.00 2.88 2.81 2.63 2.50 2.38 2.13 2.06 2.00 2.00 2.00 2.00 1.75 1.75 1.44 1.06 1.06 1.06stdev 0.00 0.50 0.54 0.50 0.63 0.81 0.89 0.77 0.89 0.82 0.97 1.03 0.77 1.00 1.21 0.77 0.77 0.77
-
jcla16 ws room3
1 5/61 E x R E
1: {F1a, G2c, G2b}
2: {F2a, F2b, F1b, G1c, G1b}
3: {F2c, F1c, F2d, G1d, F1d, G2d}
4: {G1a, G2a}
111
-
jcla16 ws room3
1 6/61 E x R R
1: {r1, r2, r5, , r6}
2: {r16}
3: {r8, r10, r9, , r13, r15}
112
-
jcla16 ws room3
1/4
vs
()
()
22
113
-
jcla16 ws room3
2/4
1
=
E R
ri R e1, e2, (ei E)
114
-
jcla16 ws room3
3/4
R 3
2 = {r16}
115
-
jcla16 ws room3
4/4 2
= ?
E R
ei E r1, r2, (ri R)
116
-
jcla16 ws room3
2 (2015)
NN5
N1-N2
BCCWJ//2
H, M, L
H, M, L 4 XY (X, Y2)
i) X=Y ii)
N1-N2
N2
X, X
117
-
jcla16 ws room3
p
1. S
si ai vi
2. p S si (i) si (ii) si
p si ap vp
3. |ai ap| Ap |vi vp| Vp
4. Ap, Vp H L
Ap < AH, H < Ap < L, L < Ap, Vp < H, H < Vp < L, L < Vp
H: M: L:
118
-
III
-
jcla16 ws room3
confounding
120
-
jcla16 ws room3
1/3
EBM
EBM (Hickery and Roberts 2011) /
121
-
jcla16 ws room3
2/3
/
122
-
jcla16 ws room3
3/3
I. ()
II. ()
NLP(Big Data) I II
123
-
IV
-
jcla16 ws room3
1/2
/
/
125
-
jcla16 ws room3
2/2
(2012):
126
-
jcla16 ws room3
(NLP) (Ling)
127
-
jcla16 ws room3
1/3, pp. 323-324
[][]
[]
128
-
jcla16 ws room3
2/3 [] [][][]MIT[][][][]
129
-
jcla16 ws room3
3/3
[][][]
130
-
jcla16 ws room3
()
()()
//
131
-
jcla16 ws room3
1/2,p.383
[][][][][...][][][][]
132
-
jcla16 ws room3
2/2,p.383
[][]
[][][]
133
-
jcla16 ws room3
1
134
-
jcla16 ws room3
1/2,p.285
[Federalistpapers][][=][=]
135
-
jcla16 ws room3
2/2
,p.285
[][][]
[][][]
136
-
jcla16 ws room3
1
/
137
-
jcla16 ws room3
1/3
,pp.393-394
[][][][][][][...]
138
-
jcla16 ws room3
2/3
,pp.393-394
[...][][][]
139
-
jcla16 ws room3
3/3
,pp.393-394
M[][...]
[...][][...][]
MarkovChain MonteCarlo=MCMC
140
-
jcla16 ws room3
1/2
/
141
-
jcla16 ws room3
2/2
!!
/
142