2 appendix

eval output M12, normalised to obs, distance ceiling = outliers removed.

2.1 citetest, method (M12)

To compute distances we queried a corpus for matching conditions where certain (assumed) determiners appear before similar nouns. In M12 (removed)… This distance should give us information structural evidence of how strong these noun occurences are connected, i.e. if a noun appears out of the blue mostly or if it somewhere before has been introduced to the audience. In information structure definitions this would be termed with given and new information (Prince 1981).

2.2 legende

## [1] ", normalised to obs, distance ceiling =  outliers removed"

Table 2.1: model vars
variable	explanation	values
target	corpus	obs,ref
q	condition	a,b,c,d,e,f
det	antecedent POS==DET	TRUE,FALSE
aut_id	author	author hash
lemma	lemma	noun lemma
range	url range of distance devised	1..maxlength(urlthread)
embed.score	semantic similarity score lemma vs. thread	0..1
q:a	query condition	.*
q:b	query condition	this,that,those,these
q:c	query condition	the
q:d	query condition	a,an,any,some
q:e	query condition	my
q:f	query condition	his,her,their,your

2.3 anova analysis

2.3.1 anova plain, formula: [`dist_rel_obs ~ targetqdet`]

##                   Df     Sum Sq    Mean Sq    F value Pr(>F)    
## target             1 6.5314e+09 6531359317 3.2559e+05 <2e-16 ***
## q                  5 3.5058e+08   70116374 3.4953e+03 <2e-16 ***
## det                1 1.5458e+07   15457530 7.7056e+02 <2e-16 ***
## target:q           5 1.8239e+06     364778 1.8184e+01 <2e-16 ***
## target:det         1 6.1232e+06    6123176 3.0524e+02 <2e-16 ***
## q:det              1 3.1197e+04      31197 1.5552e+00 0.2124    
## target:q:det       1 1.8157e+06    1815712 9.0514e+01 <2e-16 ***
## Residuals    2249061 4.5116e+10      20060                      
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

2.3.2 anova of linear regression model: [`anova(summary(lmer))`]

## Type III Analysis of Variance Table with Satterthwaite's method
##                 Sum Sq   Mean Sq NumDF  DenDF    F value    Pr(>F)    
## target           82484     82484     1 155603     4.2727   0.03873 *  
## q              1614197    322839     5 155801    16.7235 < 2.2e-16 ***
## det              54104     54104     1 155344     2.8026   0.09411 .  
## range        250671788 250671788     1 151248 12985.0989 < 2.2e-16 ***
## embed.score    3797399   3797399     1 157460   196.7098 < 2.2e-16 ***
## target:q       1064174    212835     5 155772    11.0251 1.235e-10 ***
## target:det       30708     30708     1 156612     1.5907   0.20722    
## q:det           105369    105369     1 155289     5.4582   0.01948 *  
## target:q:det                                                          
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

2.3.3 linear regression coefficients, formula: [`dist_rel_obs ~ targetqdet+(1|aut_id)+range+embed.score`]

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: eval(expr(lmeform))
##    Data: dfa
## 
## REML criterion at convergence: 2014129
## 
## Scaled residuals: 
##     Min      1Q  Median      3Q     Max 
## -6.5794 -0.3429 -0.0365  0.1456 15.9844 
## 
## Random effects:
##  Groups   Name        Variance Std.Dev.
##  aut_id   (Intercept) 41939    204.8   
##  Residual             19305    138.9   
## Number of obs: 157757, groups:  aut_id, 2512
## 
## Fixed effects:
##                     Estimate Std. Error         df  t value Pr(>|t|)    
## (Intercept)        3.386e+02  4.772e+00  3.367e+03   70.953  < 2e-16 ***
## targetref         -1.375e+01  4.211e+00  3.588e+04   -3.266 0.001093 ** 
## qb                 6.711e+01  1.328e+01  1.553e+05    5.055 4.31e-07 ***
## qc                 4.929e+01  2.865e+00  1.553e+05   17.204  < 2e-16 ***
## qd                 5.424e+01  2.903e+00  1.554e+05   18.686  < 2e-16 ***
## qe                 8.939e+01  4.512e+00  1.555e+05   19.811  < 2e-16 ***
## qf                 7.966e+01  5.550e+00  1.554e+05   14.352  < 2e-16 ***
## detTRUE            5.402e+00  1.346e+00  1.562e+05    4.012 6.01e-05 ***
## range             -3.799e-02  3.334e-04  1.512e+05 -113.952  < 2e-16 ***
## embed.score        6.208e+01  4.426e+00  1.575e+05   14.025  < 2e-16 ***
## targetref:qb      -7.595e+01  3.371e+01  1.563e+05   -2.253 0.024268 *  
## targetref:qc      -4.976e+01  1.388e+01  1.551e+05   -3.585 0.000337 ***
## targetref:qd      -4.986e+01  1.283e+01  1.550e+05   -3.886 0.000102 ***
## targetref:qe      -1.036e+02  3.677e+01  1.575e+05   -2.816 0.004856 ** 
## targetref:qf       2.261e+02  5.835e+01  1.550e+05    3.876 0.000106 ***
## targetref:detTRUE -3.762e+00  2.982e+00  1.566e+05   -1.261 0.207223    
## qb:detTRUE        -3.566e+01  1.527e+01  1.553e+05   -2.336 0.019478 *  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## fit warnings:
## fixed-effect model matrix is rank deficient so dropping 9 columns / coefficients
## Some predictor variables are on very different scales: consider rescaling

2.4 plots

Figure 2.1: compare distances by corpus, normalised to obs, distance ceiling = outliers removed

Figure 2.2: mean distances over query/corpus, normalised to obs, distance ceiling = outliers removed

Table 2.2: mean/median table for M12
target	q	n	mean	median
obs	a	541133	158	73
ref	a	1609781	45	18
obs	b	5540	196	108
ref	b	2407	94	70
obs	c	25564	219	125
ref	c	4550	90	54
obs	d	23803	228	132
ref	d	4498	114	63
obs	e	18067	238	131
ref	e	1417	123	80
obs	f	11464	189	104
ref	f	853	67	55

Figure 2.3: median distances over query/corpus, normalised to obs, distance ceiling = outliers removed

Figure 2.4: distances relation, normalised to obs, distance ceiling = outliers removed

Figure 2.5: distances normalised vs. raw

2.5 REF

literature used and alii…

References

Prince, Ellen F. 1981. “Toward a Taxonomy of Given-New Information.” In Syntax and Semantics: Vol. 14. Radical Pragmatics, edited by P. Cole, 223–55. New York: Academic Press.