複雑さ0611

1 定義

1.1 複雑さ：

• ld = 語彙密度(内容語数/総語数から算出)
• ld2 = 語彙密度(内容語数/節数から算出)
• 1AS-unitあたりの節数 = clauseperasunit

1.1.1 新指標

• 1AS-unitあたりの異なり語数 = kotonarucontents_asunit
• 1節あたりの異なり語数 = kotonarucontents_clause
• 1AS-unitあたりの内容語数 = contents_asunit
• 1節あたりの内容語数 = contents_clause

2 Boxplot ——————————————————————-

bxp <- ggboxplot(
  dat, x = "task", y = "wariai", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("Wariai")

bxp <- ggboxplot(
  dat, x = "task", y = "wariai2", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("Wariai2")

bxp <- ggboxplot(
  dat, x = "task", y = "nobe", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("Nobegosu")

bxp <- ggboxplot(
  dat, x = "task", y = "koto", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("Kotonarugosu")

bxp <- ggboxplot(
  dat, x = "task", y = "ld", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("Lexical density")

bxp <- ggboxplot(
  dat, x = "task", y = "ld2", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("Lexical density2")

bxp <- ggboxplot(
  dat, x = "task", y = "jyuzokuclauseperasunit", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("jyuzokuclauseperasunit")

bxp <- ggboxplot(
  dat, x = "task", y = "clauseperasunit", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("clauseperasunit")

3 Comparison between M and H ——-

dat_MH <- dat %>% filter(task == "M"| task == "H")
dat_MH$task <- ordered(dat_MH$task, levels=c("M","H"))

3.1 Boxplot —–

bxp <- ggboxplot(
  dat_MH, x = "task", y = "wariai", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("Wariai MH")

bxp <- ggboxplot(
  dat_MH, x = "task", y = "wariai2", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("Wariai2 MH")

bxp <- ggboxplot(
  dat_MH, x = "task", y = "nobe", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("Nobegosu MH")

bxp <- ggboxplot(
  dat_MH, x = "task", y = "koto", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("Kotonarugosu MH")

bxp <- ggboxplot(
  dat_MH, x = "task", y = "ld", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("Lexical density MH")

bxp <- ggboxplot(
  dat_MH, x = "task", y = "ld2", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("Lexical density 2 MH")

bxp <- ggboxplot(
  dat_MH, x = "task", y = "jyuzokuclauseperasunit", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("jyuzokuclauseperasunit MH")

bxp <- ggboxplot(
  dat_MH, x = "task", y = "clauseperasunit", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("clauseperasunit MH")

3.2 ld MH ——-

3.2.1 Check assumptions ——-

dat_MH %>%
  group_by(task, proficiency) %>%
  identify_outliers(ld)

## # A tibble: 4 × 37
##   task  proficiency id    name      trueid    ld   ld2  koto  nobe asunit clause
##   <ord> <ord>       <fct> <chr>     <fct>  <dbl> <dbl> <dbl> <dbl>  <dbl>  <dbl>
## 1 M     middle      16    FuRao     16      37.1  4.6     38    62      4      5
## 2 M     upper       31    YueZhiho… 32      53.8  5.56    49    93      5      9
## 3 H     upper       4     WuJingyu… 4       37.0  3.64    59   108      6     11
## 4 H     upper       10    QianJing  10      38.3  4.9     64   128      8     10
## # ℹ 26 more variables: jyuzokuclause <dbl>, jyuzokuclauseperasunit <dbl>,
## #   clauseperasunit <dbl>, wariai <dbl>, wariai2 <dbl>, jyodoushi <dbl>,
## #   heiritsu <dbl>, kakari <dbl>, fukujyoshi <dbl>, ka <dbl>,
## #   setsuzokujyoshi <dbl>, kakujyoshi <dbl>, syujyoshi <dbl>, rentai <dbl>,
## #   hijiritsu <dbl>, setubimeishi <dbl>, setsubidoushi <dbl>,
## #   setsubikeiyoushi <dbl>, kotonarucontents <dbl>,
## #   kotonarucontents_asunit <dbl>, kotonarucontents_clause <dbl>, …

dat_MH %>%
  group_by(task, proficiency) %>%
  shapiro_test(ld)

## # A tibble: 6 × 5
##   task  proficiency variable statistic     p
##   <ord> <ord>       <chr>        <dbl> <dbl>
## 1 M     lower       ld           0.883 0.114
## 2 M     middle      ld           0.920 0.289
## 3 M     upper       ld           0.887 0.106
## 4 H     lower       ld           0.972 0.906
## 5 H     middle      ld           0.927 0.351
## 6 H     upper       ld           0.908 0.202

ggqqplot(dat_MH, "ld", ggtheme = theme_bw()) +
  facet_grid(task ~ proficiency)

dat_MH %>%
  group_by(task) %>%
  levene_test(ld ~ proficiency)

## # A tibble: 2 × 5
##   task    df1   df2 statistic     p
##   <ord> <int> <int>     <dbl> <dbl>
## 1 M         2    32     0.120 0.887
## 2 H         2    32     0.322 0.727

3.2.2 Computation ——-

res.aov <- anova_test(
  data = dat_MH, dv = ld, wid = trueid,
  between = proficiency, within = task
)
get_anova_table(res.aov)

## ANOVA Table (type III tests)
## 
##             Effect DFn DFd     F     p p<.05   ges
## 1      proficiency   2  32 1.156 0.327       0.050
## 2             task   1  32 9.666 0.004     * 0.078
## 3 proficiency:task   2  32 0.992 0.382       0.017

interaction.plot(x.factor = dat_MH$task, trace.factor = dat_MH$proficiency, 
                 response = dat_MH$ld, fun = mean,
                 type = "b", legend = TRUE, trace.label = "TASK")

3.2.3 main effects

dat_MH %>%
  pairwise_t_test(
    ld ~ task, paired = TRUE, 
    p.adjust.method = "bonferroni"
  )

## # A tibble: 1 × 10
##   .y.   group1 group2    n1    n2 statistic    df     p p.adj p.adj.signif
## * <chr> <chr>  <chr>  <int> <int>     <dbl> <dbl> <dbl> <dbl> <chr>       
## 1 ld    M      H         35    35     -3.07    34 0.004 0.004 **

3.3 ld2 MH ——-

3.3.1 Check assumptions ——-

dat_MH %>%
  group_by(task, proficiency) %>%
  identify_outliers(ld2)

## # A tibble: 4 × 37
##   task  proficiency id    name      trueid    ld   ld2  koto  nobe asunit clause
##   <ord> <ord>       <fct> <chr>     <fct>  <dbl> <dbl> <dbl> <dbl>  <dbl>  <dbl>
## 1 M     lower       27    WangYing… 28      45.1  6.83    56    91      5      6
## 2 M     middle      24    XuXingsh… 24      49.6  5.7     60   115      9     10
## 3 H     lower       32    WangBiyu… 33      51.2  5.5     34    43      4      4
## 4 H     middle      24    XuXingsh… 24      50    7.12    54   114      6      8
## # ℹ 26 more variables: jyuzokuclause <dbl>, jyuzokuclauseperasunit <dbl>,
## #   clauseperasunit <dbl>, wariai <dbl>, wariai2 <dbl>, jyodoushi <dbl>,
## #   heiritsu <dbl>, kakari <dbl>, fukujyoshi <dbl>, ka <dbl>,
## #   setsuzokujyoshi <dbl>, kakujyoshi <dbl>, syujyoshi <dbl>, rentai <dbl>,
## #   hijiritsu <dbl>, setubimeishi <dbl>, setsubidoushi <dbl>,
## #   setsubikeiyoushi <dbl>, kotonarucontents <dbl>,
## #   kotonarucontents_asunit <dbl>, kotonarucontents_clause <dbl>, …

dat_MH %>%
  group_by(task, proficiency) %>%
  shapiro_test(ld2)

## # A tibble: 6 × 5
##   task  proficiency variable statistic      p
##   <ord> <ord>       <chr>        <dbl>  <dbl>
## 1 M     lower       ld2          0.845 0.0371
## 2 M     middle      ld2          0.939 0.486 
## 3 M     upper       ld2          0.947 0.594 
## 4 H     lower       ld2          0.948 0.622 
## 5 H     middle      ld2          0.907 0.198 
## 6 H     upper       ld2          0.901 0.164

ggqqplot(dat_MH, "ld2", ggtheme = theme_bw()) +
  facet_grid(task ~ proficiency)

dat_MH %>%
  group_by(task) %>%
  levene_test(ld2 ~ proficiency)

## # A tibble: 2 × 5
##   task    df1   df2 statistic      p
##   <ord> <int> <int>     <dbl>  <dbl>
## 1 M         2    32     0.270 0.765 
## 2 H         2    32     2.85  0.0727

3.3.2 Computation ——-

res.aov <- anova_test(
  data = dat_MH, dv = ld2, wid = trueid,
  between = proficiency, within = task
)
get_anova_table(res.aov)

## ANOVA Table (type III tests)
## 
##             Effect DFn DFd     F     p p<.05   ges
## 1      proficiency   2  32 0.395 0.677       0.017
## 2             task   1  32 1.928 0.175       0.018
## 3 proficiency:task   2  32 0.671 0.518       0.013

interaction.plot(x.factor = dat_MH$task, trace.factor = dat_MH$proficiency, 
                 response = dat_MH$ld2, fun = mean,
                 type = "b", legend = TRUE, trace.label = "TASK")

3.4 clauseperasunit MH ——-

3.4.1 Check assumptions ——-

dat_MH %>%
  group_by(task, proficiency) %>%
  identify_outliers(clauseperasunit)

## # A tibble: 7 × 37
##   task  proficiency id    name      trueid    ld   ld2  koto  nobe asunit clause
##   <ord> <ord>       <fct> <chr>     <fct>  <dbl> <dbl> <dbl> <dbl>  <dbl>  <dbl>
## 1 M     lower       13    LuYang    13      52.9  4.09    53    85      7     11
## 2 M     lower       29    LiuChihy… 30      47.7  4.43    41    65      7      7
## 3 M     middle      17    ZhenLinj… 17      40.4  3.83    38    57      2      6
## 4 H     lower       27    WangYing… 28      50.8  3.3     46    65      7     10
## 5 H     lower       32    WangBiyu… 33      51.2  5.5     34    43      4      4
## 6 H     lower       35    JinXinkai 36      47.5  3.45    45    80     11     11
## 7 H     middle      33    ZhangYisu 34      48.5  4.55    60   103      6     11
## # ℹ 26 more variables: jyuzokuclause <dbl>, jyuzokuclauseperasunit <dbl>,
## #   clauseperasunit <dbl>, wariai <dbl>, wariai2 <dbl>, jyodoushi <dbl>,
## #   heiritsu <dbl>, kakari <dbl>, fukujyoshi <dbl>, ka <dbl>,
## #   setsuzokujyoshi <dbl>, kakujyoshi <dbl>, syujyoshi <dbl>, rentai <dbl>,
## #   hijiritsu <dbl>, setubimeishi <dbl>, setsubidoushi <dbl>,
## #   setsubikeiyoushi <dbl>, kotonarucontents <dbl>,
## #   kotonarucontents_asunit <dbl>, kotonarucontents_clause <dbl>, …

dat_MH %>%
  group_by(task, proficiency) %>%
  shapiro_test(clauseperasunit)

## # A tibble: 6 × 5
##   task  proficiency variable        statistic       p
##   <ord> <ord>       <chr>               <dbl>   <dbl>
## 1 M     lower       clauseperasunit     0.880 0.103  
## 2 M     middle      clauseperasunit     0.718 0.00127
## 3 M     upper       clauseperasunit     0.989 1.00   
## 4 H     lower       clauseperasunit     0.923 0.344  
## 5 H     middle      clauseperasunit     0.898 0.149  
## 6 H     upper       clauseperasunit     0.894 0.135

ggqqplot(dat_MH, "clauseperasunit", ggtheme = theme_bw()) +
  facet_grid(task ~ proficiency)

dat_MH %>%
  group_by(task) %>%
  levene_test(clauseperasunit ~ proficiency)

## # A tibble: 2 × 5
##   task    df1   df2 statistic     p
##   <ord> <int> <int>     <dbl> <dbl>
## 1 M         2    32      1.90 0.167
## 2 H         2    32      1.96 0.158

dat_MH %>%
  group_by(task, proficiency) %>%
  get_summary_stats(clauseperasunit,type = "mean_sd")

## # A tibble: 6 × 6
##   task  proficiency variable            n  mean    sd
##   <ord> <ord>       <fct>           <dbl> <dbl> <dbl>
## 1 M     lower       clauseperasunit    11  1.22 0.143
## 2 M     middle      clauseperasunit    12  1.52 0.521
## 3 M     upper       clauseperasunit    12  1.57 0.313
## 4 H     lower       clauseperasunit    11  1.20 0.123
## 5 H     middle      clauseperasunit    12  1.30 0.214
## 6 H     upper       clauseperasunit    12  1.49 0.211

3.4.2 outliers

dat_MH_2 <- dat_MH %>% filter(!(trueid == "17"))

dat_MH_2 %>%
  group_by(task, proficiency) %>%
  shapiro_test(clauseperasunit)

## # A tibble: 6 × 5
##   task  proficiency variable        statistic     p
##   <ord> <ord>       <chr>               <dbl> <dbl>
## 1 M     lower       clauseperasunit     0.880 0.103
## 2 M     middle      clauseperasunit     0.890 0.139
## 3 M     upper       clauseperasunit     0.989 1.00 
## 4 H     lower       clauseperasunit     0.923 0.344
## 5 H     middle      clauseperasunit     0.902 0.194
## 6 H     upper       clauseperasunit     0.894 0.135

ggqqplot(dat_MH_2, "clauseperasunit", ggtheme = theme_bw()) +
  facet_grid(task ~ proficiency)

dat_MH_2 %>%
  group_by(task) %>%
  levene_test(clauseperasunit ~ proficiency)

## # A tibble: 2 × 5
##   task    df1   df2 statistic      p
##   <ord> <int> <int>     <dbl>  <dbl>
## 1 M         2    31      2.84 0.0737
## 2 H         2    31      1.93 0.162

3.4.3 Computation ——-

res.aov <- anova_test(
  data = dat_MH_2, dv = clauseperasunit, wid = trueid,
  between = proficiency, within = task
)
get_anova_table(res.aov)

## ANOVA Table (type III tests)
## 
##             Effect DFn DFd     F     p p<.05   ges
## 1      proficiency   2  31 8.566 0.001     * 0.285
## 2             task   1  31 2.320 0.138       0.020
## 3 proficiency:task   2  31 0.230 0.796       0.004

interaction.plot(x.factor = dat_MH_2$task, trace.factor = dat_MH_2$proficiency, 
                 response = dat_MH_2$clauseperasunit, fun = mean,
                 type = "b", legend = TRUE, trace.label = "TASK")

3.4.4 main effects

dat_MH_2 %>%
  pairwise_t_test(
    clauseperasunit ~ proficiency,
    p.adjust.method = "bonferroni"
  )

## # A tibble: 3 × 9
##   .y.            group1 group2    n1    n2       p p.signif   p.adj p.adj.signif
## * <chr>          <chr>  <chr>  <int> <int>   <dbl> <chr>      <dbl> <chr>       
## 1 clauseperasun… lower  middle    22    22 4.91e-2 *        1.47e-1 ns          
## 2 clauseperasun… lower  upper     22    24 4.84e-6 ****     1.45e-5 ****        
## 3 clauseperasun… middle upper     22    24 4.56e-3 **       1.37e-2 *

bxp <- ggboxplot(
  dat_MH, x = "task", y = "clauseperasunit", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("clauseperasunit MH")

cpa_MH <- bxp + ggtitle("clauseperasunit MH")

3.5 kotonarucontents_asunit MH ——-

3.5.1 Check assumptions ——-

dat_MH %>%
  group_by(task, proficiency) %>%
  identify_outliers(kotonarucontents_asunit)

## # A tibble: 4 × 37
##   task  proficiency id    name      trueid    ld   ld2  koto  nobe asunit clause
##   <ord> <ord>       <fct> <chr>     <fct>  <dbl> <dbl> <dbl> <dbl>  <dbl>  <dbl>
## 1 M     middle      17    ZhenLinj… 17      40.4  3.83    38    57      2      6
## 2 H     lower       32    WangBiyu… 33      51.2  5.5     34    43      4      4
## 3 H     lower       35    JinXinkai 36      47.5  3.45    45    80     11     11
## 4 H     middle      33    ZhangYisu 34      48.5  4.55    60   103      6     11
## # ℹ 26 more variables: jyuzokuclause <dbl>, jyuzokuclauseperasunit <dbl>,
## #   clauseperasunit <dbl>, wariai <dbl>, wariai2 <dbl>, jyodoushi <dbl>,
## #   heiritsu <dbl>, kakari <dbl>, fukujyoshi <dbl>, ka <dbl>,
## #   setsuzokujyoshi <dbl>, kakujyoshi <dbl>, syujyoshi <dbl>, rentai <dbl>,
## #   hijiritsu <dbl>, setubimeishi <dbl>, setsubidoushi <dbl>,
## #   setsubikeiyoushi <dbl>, kotonarucontents <dbl>,
## #   kotonarucontents_asunit <dbl>, kotonarucontents_clause <dbl>, …

dat_MH %>%
  group_by(task, proficiency) %>%
  shapiro_test(kotonarucontents_asunit)

## # A tibble: 6 × 5
##   task  proficiency variable                statistic        p
##   <ord> <ord>       <chr>                       <dbl>    <dbl>
## 1 M     lower       kotonarucontents_asunit     0.944 0.572   
## 2 M     middle      kotonarucontents_asunit     0.669 0.000429
## 3 M     upper       kotonarucontents_asunit     0.935 0.437   
## 4 H     lower       kotonarucontents_asunit     0.882 0.112   
## 5 H     middle      kotonarucontents_asunit     0.859 0.0470  
## 6 H     upper       kotonarucontents_asunit     0.895 0.138

ggqqplot(dat_MH, "kotonarucontents_asunit", ggtheme = theme_bw()) +
  facet_grid(task ~ proficiency)

dat_MH %>%
  group_by(task) %>%
  levene_test(kotonarucontents_asunit ~ proficiency)

## # A tibble: 2 × 5
##   task    df1   df2 statistic     p
##   <ord> <int> <int>     <dbl> <dbl>
## 1 M         2    32     0.521 0.599
## 2 H         2    32     1.55  0.227

3.5.2 Computation ——-

res.aov <- anova_test(
  data = dat_MH, dv = kotonarucontents_asunit, wid = trueid,
  between = proficiency, within = task
)
get_anova_table(res.aov)

## ANOVA Table (type III tests)
## 
##             Effect DFn DFd     F     p p<.05   ges
## 1      proficiency   2  32 3.291 0.050       0.111
## 2             task   1  32 3.795 0.060       0.044
## 3 proficiency:task   2  32 0.209 0.812       0.005

interaction.plot(x.factor = dat_MH$task, trace.factor = dat_MH$proficiency, 
                 response = dat_MH$kotonarucontents_asunit, fun = mean,
                 type = "b", legend = TRUE, trace.label = "TASK")

3.6 kotonarucontents_clause MH ——-

3.6.1 Check assumptions ——-

dat_MH %>%
  group_by(task, proficiency) %>%
  identify_outliers(kotonarucontents_clause)

## # A tibble: 4 × 37
##   task  proficiency id    name      trueid    ld   ld2  koto  nobe asunit clause
##   <ord> <ord>       <fct> <chr>     <fct>  <dbl> <dbl> <dbl> <dbl>  <dbl>  <dbl>
## 1 M     lower       27    WangYing… 28      45.1  6.83    56    91      5      6
## 2 H     lower       32    WangBiyu… 33      51.2  5.5     34    43      4      4
## 3 H     middle      16    FuRao     16      42.5  5.62    64   106      7      8
## 4 H     middle      24    XuXingsh… 24      50    7.12    54   114      6      8
## # ℹ 26 more variables: jyuzokuclause <dbl>, jyuzokuclauseperasunit <dbl>,
## #   clauseperasunit <dbl>, wariai <dbl>, wariai2 <dbl>, jyodoushi <dbl>,
## #   heiritsu <dbl>, kakari <dbl>, fukujyoshi <dbl>, ka <dbl>,
## #   setsuzokujyoshi <dbl>, kakujyoshi <dbl>, syujyoshi <dbl>, rentai <dbl>,
## #   hijiritsu <dbl>, setubimeishi <dbl>, setsubidoushi <dbl>,
## #   setsubikeiyoushi <dbl>, kotonarucontents <dbl>,
## #   kotonarucontents_asunit <dbl>, kotonarucontents_clause <dbl>, …

dat_MH %>%
  group_by(task, proficiency) %>%
  shapiro_test(kotonarucontents_clause)

## # A tibble: 6 × 5
##   task  proficiency variable                statistic         p
##   <ord> <ord>       <chr>                       <dbl>     <dbl>
## 1 M     lower       kotonarucontents_clause     0.849 0.0419   
## 2 M     middle      kotonarucontents_clause     0.919 0.274    
## 3 M     upper       kotonarucontents_clause     0.966 0.864    
## 4 H     lower       kotonarucontents_clause     0.615 0.0000390
## 5 H     middle      kotonarucontents_clause     0.822 0.0167   
## 6 H     upper       kotonarucontents_clause     0.896 0.143

ggqqplot(dat_MH, "kotonarucontents_clause", ggtheme = theme_bw()) +
  facet_grid(task ~ proficiency)

dat_MH %>%
  group_by(task) %>%
  levene_test(kotonarucontents_clause ~ proficiency)

## # A tibble: 2 × 5
##   task    df1   df2 statistic     p
##   <ord> <int> <int>     <dbl> <dbl>
## 1 M         2    32     0.228 0.797
## 2 H         2    32     0.384 0.684

3.6.2 Computation ——-

res.aov <- anova_test(
  data = dat_MH, dv = kotonarucontents_clause, wid = trueid,
  between = proficiency, within = task
)
get_anova_table(res.aov)

## ANOVA Table (type III tests)
## 
##             Effect DFn DFd     F     p p<.05      ges
## 1      proficiency   2  32 0.005 0.995       0.000205
## 2             task   1  32 1.071 0.308       0.011000
## 3 proficiency:task   2  32 0.421 0.660       0.009000

interaction.plot(x.factor = dat_MH$task, trace.factor = dat_MH$proficiency, 
                 response = dat_MH$kotonarucontents_clause, fun = mean,
                 type = "b", legend = TRUE, trace.label = "TASK")

3.7 contents_asunit MH ——-

3.7.1 Check assumptions ——-

dat_MH %>%
  group_by(task, proficiency) %>%
  identify_outliers(contents_asunit)

## # A tibble: 3 × 37
##   task  proficiency id    name      trueid    ld   ld2  koto  nobe asunit clause
##   <ord> <ord>       <fct> <chr>     <fct>  <dbl> <dbl> <dbl> <dbl>  <dbl>  <dbl>
## 1 M     middle      17    ZhenLinj… 17      40.4  3.83    38    57      2      6
## 2 H     lower       35    JinXinkai 36      47.5  3.45    45    80     11     11
## 3 H     middle      24    XuXingsh… 24      50    7.12    54   114      6      8
## # ℹ 26 more variables: jyuzokuclause <dbl>, jyuzokuclauseperasunit <dbl>,
## #   clauseperasunit <dbl>, wariai <dbl>, wariai2 <dbl>, jyodoushi <dbl>,
## #   heiritsu <dbl>, kakari <dbl>, fukujyoshi <dbl>, ka <dbl>,
## #   setsuzokujyoshi <dbl>, kakujyoshi <dbl>, syujyoshi <dbl>, rentai <dbl>,
## #   hijiritsu <dbl>, setubimeishi <dbl>, setsubidoushi <dbl>,
## #   setsubikeiyoushi <dbl>, kotonarucontents <dbl>,
## #   kotonarucontents_asunit <dbl>, kotonarucontents_clause <dbl>, …

dat_MH %>%
  group_by(task, proficiency) %>%
  shapiro_test(contents_asunit)

## # A tibble: 6 × 5
##   task  proficiency variable        statistic       p
##   <ord> <ord>       <chr>               <dbl>   <dbl>
## 1 M     lower       contents_asunit     0.842 0.0332 
## 2 M     middle      contents_asunit     0.802 0.00999
## 3 M     upper       contents_asunit     0.940 0.500  
## 4 H     lower       contents_asunit     0.887 0.127  
## 5 H     middle      contents_asunit     0.859 0.0475 
## 6 H     upper       contents_asunit     0.962 0.818

ggqqplot(dat_MH, "contents_asunit", ggtheme = theme_bw()) +
  facet_grid(task ~ proficiency)

dat_MH %>%
  group_by(task) %>%
  levene_test(contents_asunit ~ proficiency)

## # A tibble: 2 × 5
##   task    df1   df2 statistic     p
##   <ord> <int> <int>     <dbl> <dbl>
## 1 M         2    32     0.811 0.453
## 2 H         2    32     2.45  0.102

3.7.2 Computation ——-

res.aov <- anova_test(
  data = dat_MH, dv = contents_asunit, wid = trueid,
  between = proficiency, within = task
)
get_anova_table(res.aov)

## ANOVA Table (type III tests)
## 
##             Effect DFn DFd     F     p p<.05   ges
## 1      proficiency   2  32 5.470 0.009     * 0.174
## 2             task   1  32 5.227 0.029     * 0.059
## 3 proficiency:task   2  32 0.118 0.889       0.003

interaction.plot(x.factor = dat_MH$task, trace.factor = dat_MH$proficiency, 
                 response = dat_MH$contents_asunit, fun = mean,
                 type = "b", legend = TRUE, trace.label = "TASK")

3.8 contents_clause MH ——-

3.8.1 Check assumptions ——-

dat_MH %>%
  group_by(task, proficiency) %>%
  identify_outliers(contents_clause)

## # A tibble: 4 × 37
##   task  proficiency id    name      trueid    ld   ld2  koto  nobe asunit clause
##   <ord> <ord>       <fct> <chr>     <fct>  <dbl> <dbl> <dbl> <dbl>  <dbl>  <dbl>
## 1 M     lower       27    WangYing… 28      45.1  6.83    56    91      5      6
## 2 M     middle      24    XuXingsh… 24      49.6  5.7     60   115      9     10
## 3 H     lower       32    WangBiyu… 33      51.2  5.5     34    43      4      4
## 4 H     middle      24    XuXingsh… 24      50    7.12    54   114      6      8
## # ℹ 26 more variables: jyuzokuclause <dbl>, jyuzokuclauseperasunit <dbl>,
## #   clauseperasunit <dbl>, wariai <dbl>, wariai2 <dbl>, jyodoushi <dbl>,
## #   heiritsu <dbl>, kakari <dbl>, fukujyoshi <dbl>, ka <dbl>,
## #   setsuzokujyoshi <dbl>, kakujyoshi <dbl>, syujyoshi <dbl>, rentai <dbl>,
## #   hijiritsu <dbl>, setubimeishi <dbl>, setsubidoushi <dbl>,
## #   setsubikeiyoushi <dbl>, kotonarucontents <dbl>,
## #   kotonarucontents_asunit <dbl>, kotonarucontents_clause <dbl>, …

dat_MH %>%
  group_by(task, proficiency) %>%
  shapiro_test(contents_clause)

## # A tibble: 6 × 5
##   task  proficiency variable        statistic      p
##   <ord> <ord>       <chr>               <dbl>  <dbl>
## 1 M     lower       contents_clause     0.845 0.0371
## 2 M     middle      contents_clause     0.939 0.486 
## 3 M     upper       contents_clause     0.947 0.594 
## 4 H     lower       contents_clause     0.948 0.622 
## 5 H     middle      contents_clause     0.907 0.198 
## 6 H     upper       contents_clause     0.901 0.164

ggqqplot(dat_MH, "contents_clause", ggtheme = theme_bw()) +
  facet_grid(task ~ proficiency)

dat_MH %>%
  group_by(task) %>%
  levene_test(contents_clause ~ proficiency)

## # A tibble: 2 × 5
##   task    df1   df2 statistic      p
##   <ord> <int> <int>     <dbl>  <dbl>
## 1 M         2    32     0.270 0.765 
## 2 H         2    32     2.85  0.0727

3.8.2 Computation ——-

res.aov <- anova_test(
  data = dat_MH, dv = contents_clause, wid = trueid,
  between = proficiency, within = task
)
get_anova_table(res.aov)

## ANOVA Table (type III tests)
## 
##             Effect DFn DFd     F     p p<.05   ges
## 1      proficiency   2  32 0.395 0.677       0.017
## 2             task   1  32 1.928 0.175       0.018
## 3 proficiency:task   2  32 0.671 0.518       0.013

interaction.plot(x.factor = dat_MH$task, trace.factor = dat_MH$proficiency, 
                 response = dat_MH$contents_clause, fun = mean,
                 type = "b", legend = TRUE, trace.label = "TASK")

4 Comparison between L and M ——-

dat_LM <- dat %>% filter(task == "M"| task == "L")
dat_LM$task <- ordered(dat_LM$task, levels=c("L","M"))

4.1 Boxplot —–

bxp <- ggboxplot(
  dat_LM, x = "task", y = "wariai", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("Wariai LM")

bxp <- ggboxplot(
  dat_LM, x = "task", y = "wariai2", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("Wariai2 LM")

bxp <- ggboxplot(
  dat_LM, x = "task", y = "nobe", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("Nobegosu LM")

bxp <- ggboxplot(
  dat_LM, x = "task", y = "koto", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("Kotonarugosu LM")

bxp <- ggboxplot(
  dat_LM, x = "task", y = "ld", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("Lexical density LM")

bxp <- ggboxplot(
  dat_LM, x = "task", y = "ld", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("Lexical density 2 LM")

bxp <- ggboxplot(
  dat_LM, x = "task", y = "jyuzokuclauseperasunit", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("jyuzokuclauseperasunit LM")

bxp <- ggboxplot(
  dat_LM, x = "task", y = "clauseperasunit", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("clauseperasunit LM")

4.2 ld MH ——-

4.2.1 Check assumptions ——-

dat_LM %>%
  group_by(task, proficiency) %>%
  identify_outliers(ld)

## # A tibble: 6 × 37
##   task  proficiency id    name      trueid    ld   ld2  koto  nobe asunit clause
##   <ord> <ord>       <fct> <chr>     <fct>  <dbl> <dbl> <dbl> <dbl>  <dbl>  <dbl>
## 1 L     lower       19    ZhouLupi… 19      56.5  3.69    44    85      9     13
## 2 L     lower       25    LiuBoyan  25      46.3  4.17    39    54      5      6
## 3 L     middle      2     JinZeming 2       55.6  3.89    47    63      5      9
## 4 L     middle      15    LiHaiqing 15      53.9  4.77    63   115      9     13
## 5 M     middle      16    FuRao     16      37.1  4.6     38    62      4      5
## 6 M     upper       31    YueZhiho… 32      53.8  5.56    49    93      5      9
## # ℹ 26 more variables: jyuzokuclause <dbl>, jyuzokuclauseperasunit <dbl>,
## #   clauseperasunit <dbl>, wariai <dbl>, wariai2 <dbl>, jyodoushi <dbl>,
## #   heiritsu <dbl>, kakari <dbl>, fukujyoshi <dbl>, ka <dbl>,
## #   setsuzokujyoshi <dbl>, kakujyoshi <dbl>, syujyoshi <dbl>, rentai <dbl>,
## #   hijiritsu <dbl>, setubimeishi <dbl>, setsubidoushi <dbl>,
## #   setsubikeiyoushi <dbl>, kotonarucontents <dbl>,
## #   kotonarucontents_asunit <dbl>, kotonarucontents_clause <dbl>, …

dat_LM %>%
  group_by(task, proficiency) %>%
  shapiro_test(ld)

## # A tibble: 6 × 5
##   task  proficiency variable statistic     p
##   <ord> <ord>       <chr>        <dbl> <dbl>
## 1 L     lower       ld           0.956 0.719
## 2 L     middle      ld           0.931 0.389
## 3 L     upper       ld           0.991 1.00 
## 4 M     lower       ld           0.883 0.114
## 5 M     middle      ld           0.920 0.289
## 6 M     upper       ld           0.887 0.106

ggqqplot(dat_LM, "ld", ggtheme = theme_bw()) +
  facet_grid(task ~ proficiency)

dat_LM %>%
  group_by(task) %>%
  levene_test(ld ~ proficiency)

## # A tibble: 2 × 5
##   task    df1   df2 statistic     p
##   <ord> <int> <int>     <dbl> <dbl>
## 1 L         2    32     1.02  0.373
## 2 M         2    32     0.120 0.887

4.2.2 Computation ——-

res.aov <- anova_test(
  data = dat_LM, dv = ld, wid = trueid,
  between = proficiency, within = task
)
get_anova_table(res.aov)

## ANOVA Table (type III tests)
## 
##             Effect DFn DFd      F     p p<.05   ges
## 1      proficiency   2  32  2.826 0.074       0.102
## 2             task   1  32 11.109 0.002     * 0.111
## 3 proficiency:task   2  32  2.010 0.151       0.043

interaction.plot(x.factor = dat_LM$task, trace.factor = dat_LM$proficiency, 
                 response = dat_LM$ld, fun = mean,
                 type = "b", legend = TRUE, trace.label = "TASK")

4.2.3 main effects

dat_LM %>%
  pairwise_t_test(
    ld ~ task, paired = TRUE, 
    p.adjust.method = "bonferroni"
  )

## # A tibble: 1 × 10
##   .y.   group1 group2    n1    n2 statistic    df     p p.adj p.adj.signif
## * <chr> <chr>  <chr>  <int> <int>     <dbl> <dbl> <dbl> <dbl> <chr>       
## 1 ld    L      M         35    35      3.16    34 0.003 0.003 **

4.3 ld2 MH ——-

4.3.1 Check assumptions ——-

dat_LM %>%
  group_by(task, proficiency) %>%
  identify_outliers(ld2)

## # A tibble: 4 × 37
##   task  proficiency id    name      trueid    ld   ld2  koto  nobe asunit clause
##   <ord> <ord>       <fct> <chr>     <fct>  <dbl> <dbl> <dbl> <dbl>  <dbl>  <dbl>
## 1 L     middle      24    XuXingsh… 24      47.3  6.62    66   112      4      8
## 2 L     upper       23    TuChuyue  23      49.5  8.17    65    99      5      6
## 3 M     lower       27    WangYing… 28      45.1  6.83    56    91      5      6
## 4 M     middle      24    XuXingsh… 24      49.6  5.7     60   115      9     10
## # ℹ 26 more variables: jyuzokuclause <dbl>, jyuzokuclauseperasunit <dbl>,
## #   clauseperasunit <dbl>, wariai <dbl>, wariai2 <dbl>, jyodoushi <dbl>,
## #   heiritsu <dbl>, kakari <dbl>, fukujyoshi <dbl>, ka <dbl>,
## #   setsuzokujyoshi <dbl>, kakujyoshi <dbl>, syujyoshi <dbl>, rentai <dbl>,
## #   hijiritsu <dbl>, setubimeishi <dbl>, setsubidoushi <dbl>,
## #   setsubikeiyoushi <dbl>, kotonarucontents <dbl>,
## #   kotonarucontents_asunit <dbl>, kotonarucontents_clause <dbl>, …

dat_LM %>%
  group_by(task, proficiency) %>%
  shapiro_test(ld2)

## # A tibble: 6 × 5
##   task  proficiency variable statistic      p
##   <ord> <ord>       <chr>        <dbl>  <dbl>
## 1 L     lower       ld2          0.876 0.0929
## 2 L     middle      ld2          0.933 0.409 
## 3 L     upper       ld2          0.828 0.0198
## 4 M     lower       ld2          0.845 0.0371
## 5 M     middle      ld2          0.939 0.486 
## 6 M     upper       ld2          0.947 0.594

ggqqplot(dat_LM, "ld2", ggtheme = theme_bw()) +
  facet_grid(task ~ proficiency)

dat_LM %>%
  group_by(task) %>%
  levene_test(ld2 ~ proficiency)

## # A tibble: 2 × 5
##   task    df1   df2 statistic     p
##   <ord> <int> <int>     <dbl> <dbl>
## 1 L         2    32     0.590 0.560
## 2 M         2    32     0.270 0.765

4.3.2 Computation ——-

res.aov <- anova_test(
  data = dat_LM, dv = ld2, wid = trueid,
  between = proficiency, within = task
)
get_anova_table(res.aov)

## ANOVA Table (type III tests)
## 
##             Effect DFn DFd     F     p p<.05   ges
## 1      proficiency   2  32 1.508 0.237       0.051
## 2             task   1  32 0.608 0.441       0.008
## 3 proficiency:task   2  32 0.324 0.725       0.009

interaction.plot(x.factor = dat_LM$task, trace.factor = dat_LM$proficiency, 
                 response = dat_LM$ld2, fun = mean,
                 type = "b", legend = TRUE, trace.label = "TASK")

4.4 Clauseperasunit MH ——

4.4.1 Check assumptions ——-

dat_LM %>%
  group_by(task, proficiency) %>%
  identify_outliers(clauseperasunit)

## # A tibble: 6 × 37
##   task  proficiency id    name      trueid    ld   ld2  koto  nobe asunit clause
##   <ord> <ord>       <fct> <chr>     <fct>  <dbl> <dbl> <dbl> <dbl>  <dbl>  <dbl>
## 1 L     lower       27    WangYing… 28      50.7  3.8     50    75      3     10
## 2 L     upper       1     Xiaxue    1       46.5  3.65    82   157      9     20
## 3 L     upper       22    ZhangBoh… 22      41.3  4.33    74   126      4     12
## 4 M     lower       13    LuYang    13      52.9  4.09    53    85      7     11
## 5 M     lower       29    LiuChihy… 30      47.7  4.43    41    65      7      7
## 6 M     middle      17    ZhenLinj… 17      40.4  3.83    38    57      2      6
## # ℹ 26 more variables: jyuzokuclause <dbl>, jyuzokuclauseperasunit <dbl>,
## #   clauseperasunit <dbl>, wariai <dbl>, wariai2 <dbl>, jyodoushi <dbl>,
## #   heiritsu <dbl>, kakari <dbl>, fukujyoshi <dbl>, ka <dbl>,
## #   setsuzokujyoshi <dbl>, kakujyoshi <dbl>, syujyoshi <dbl>, rentai <dbl>,
## #   hijiritsu <dbl>, setubimeishi <dbl>, setsubidoushi <dbl>,
## #   setsubikeiyoushi <dbl>, kotonarucontents <dbl>,
## #   kotonarucontents_asunit <dbl>, kotonarucontents_clause <dbl>, …

dat_LM %>%
  group_by(task, proficiency) %>%
  shapiro_test(clauseperasunit)

## # A tibble: 6 × 5
##   task  proficiency variable        statistic        p
##   <ord> <ord>       <chr>               <dbl>    <dbl>
## 1 L     lower       clauseperasunit     0.678 0.000243
## 2 L     middle      clauseperasunit     0.926 0.336   
## 3 L     upper       clauseperasunit     0.761 0.00350 
## 4 M     lower       clauseperasunit     0.880 0.103   
## 5 M     middle      clauseperasunit     0.718 0.00127 
## 6 M     upper       clauseperasunit     0.989 1.00

ggqqplot(dat_LM, "clauseperasunit", ggtheme = theme_bw()) +
  facet_grid(task ~ proficiency)

dat_LM %>%
  group_by(task) %>%
  levene_test(clauseperasunit ~ proficiency)

## # A tibble: 2 × 5
##   task    df1   df2 statistic     p
##   <ord> <int> <int>     <dbl> <dbl>
## 1 L         2    32    0.0444 0.957
## 2 M         2    32    1.90   0.167

dat_LM %>%
  group_by(task, proficiency) %>%
  get_summary_stats(clauseperasunit,type = "mean_sd")

## # A tibble: 6 × 6
##   task  proficiency variable            n  mean    sd
##   <ord> <ord>       <fct>           <dbl> <dbl> <dbl>
## 1 L     lower       clauseperasunit    11  1.49 0.654
## 2 L     middle      clauseperasunit    12  1.70 0.368
## 3 L     upper       clauseperasunit    12  1.62 0.536
## 4 M     lower       clauseperasunit    11  1.22 0.143
## 5 M     middle      clauseperasunit    12  1.52 0.521
## 6 M     upper       clauseperasunit    12  1.57 0.313

bxp <- ggboxplot(
  dat_LM, x = "task", y = "clauseperasunit", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("clauseperasunit LM")

4.4.2 outliers

dat_LM_2 <- dat_LM %>% filter(!(trueid == "22"|trueid == "28"|trueid == "13"|trueid == "17"|trueid == "1"|trueid == "11"))

dat_LM_2 %>%
  group_by(task, proficiency) %>%
  identify_outliers(clauseperasunit)

##  [1] task                    proficiency             id                     
##  [4] name                    trueid                  ld                     
##  [7] ld2                     koto                    nobe                   
## [10] asunit                  clause                  jyuzokuclause          
## [13] jyuzokuclauseperasunit  clauseperasunit         wariai                 
## [16] wariai2                 jyodoushi               heiritsu               
## [19] kakari                  fukujyoshi              ka                     
## [22] setsuzokujyoshi         kakujyoshi              syujyoshi              
## [25] rentai                  hijiritsu               setubimeishi           
## [28] setsubidoushi           setsubikeiyoushi        kotonarucontents       
## [31] kotonarucontents_asunit kotonarucontents_clause contents_asunit        
## [34] contents_clause         contents                is.outlier             
## [37] is.extreme             
## <0 rows> (or 0-length row.names)

dat_LM_2 %>%
  group_by(task, proficiency) %>%
  shapiro_test(clauseperasunit)

## # A tibble: 6 × 5
##   task  proficiency variable        statistic      p
##   <ord> <ord>       <chr>               <dbl>  <dbl>
## 1 L     lower       clauseperasunit     0.943 0.616 
## 2 L     middle      clauseperasunit     0.908 0.231 
## 3 L     upper       clauseperasunit     0.843 0.0630
## 4 M     lower       clauseperasunit     0.963 0.827 
## 5 M     middle      clauseperasunit     0.890 0.139 
## 6 M     upper       clauseperasunit     0.975 0.936

ggqqplot(dat_LM_2, "clauseperasunit", ggtheme = theme_bw()) +
  facet_grid(task ~ proficiency)

dat_LM_2 %>%
  group_by(task) %>%
  levene_test(clauseperasunit ~ proficiency)

## # A tibble: 2 × 5
##   task    df1   df2 statistic      p
##   <ord> <int> <int>     <dbl>  <dbl>
## 1 L         2    26      4.70 0.0182
## 2 M         2    26      2.32 0.118

bxp <- ggboxplot(
  dat_LM_2, x = "task", y = "clauseperasunit", add = "jitter",
  color = "proficiency", palette = "lancet"
)
bxp + ggtitle("clauseperasunit LM")

4.4.3 Computation ——-

dat_LM_2 %>%
  group_by(task, proficiency) %>%
  get_summary_stats(clauseperasunit, type = "mean_sd")

## # A tibble: 6 × 6
##   task  proficiency variable            n  mean    sd
##   <ord> <ord>       <fct>           <dbl> <dbl> <dbl>
## 1 L     lower       clauseperasunit     9  1.31 0.261
## 2 L     middle      clauseperasunit    11  1.70 0.386
## 3 L     upper       clauseperasunit     9  1.36 0.126
## 4 M     lower       clauseperasunit     9  1.18 0.094
## 5 M     middle      clauseperasunit    11  1.38 0.241
## 6 M     upper       clauseperasunit     9  1.48 0.279

res.aov <- anova_test(
  data = dat_LM_2, dv = clauseperasunit, wid = trueid,
  between = proficiency, within = task
)
get_anova_table(res.aov)

## ANOVA Table (type III tests)
## 
##             Effect DFn DFd     F     p p<.05   ges
## 1      proficiency   2  26 6.536 0.005     * 0.197
## 2             task   1  26 2.474 0.128       0.046
## 3 proficiency:task   2  26 3.437 0.047     * 0.119

interaction.plot(x.factor = dat_LM_2$task, trace.factor = dat_LM_2$proficiency, 
                 response = dat_LM_2$clauseperasunit, fun = mean,
                 type = "b", legend = TRUE, trace.label = "TASK")

### Effect of group at each time point

one.way <- dat_LM_2 %>%
  group_by(task) %>%
  anova_test(dv = clauseperasunit, wid = trueid, between = proficiency) %>%
  get_anova_table() %>%
  adjust_pvalue(method = "bonferroni")
one.way

## # A tibble: 2 × 9
##   task  Effect        DFn   DFd     F     p `p<.05`   ges p.adj
##   <ord> <chr>       <dbl> <dbl> <dbl> <dbl> <chr>   <dbl> <dbl>
## 1 L     proficiency     2    26  5.48  0.01 *       0.297  0.02
## 2 M     proficiency     2    26  4.04  0.03 *       0.237  0.06

4.4.4 Pairwise comparisons between group levels

pwc <- dat_LM_2 %>%
  group_by(task) %>%
  pairwise_t_test(clauseperasunit ~ proficiency, p.adjust.method = "bonferroni")
pwc

## # A tibble: 6 × 10
##   task  .y.       group1 group2    n1    n2       p p.signif  p.adj p.adj.signif
## * <ord> <chr>     <chr>  <chr>  <int> <int>   <dbl> <chr>     <dbl> <chr>       
## 1 L     clausepe… lower  middle     9    11 0.00603 **       0.0181 *           
## 2 L     clausepe… lower  upper      9     9 0.726   ns       1      ns          
## 3 L     clausepe… middle upper     11     9 0.0146  *        0.0437 *           
## 4 M     clausepe… lower  middle     9    11 0.0601  ns       0.18   ns          
## 5 M     clausepe… lower  upper      9     9 0.01    *        0.03   *           
## 6 M     clausepe… middle upper     11     9 0.352   ns       1      ns

4.4.5 Visualization: boxplots with p-values

pwc <- pwc %>% add_xy_position(x = "task")

cpa_LM <- bxp + 
  stat_pvalue_manual(pwc, tip.length = 0, hide.ns = TRUE) +
  labs(
    subtitle = get_test_label(res.aov, detailed = TRUE),
    caption = get_pwc_label(pwc)
  )

cpa_LM

4.5 kotonarucontents_asunit LM ——-

4.5.1 Check assumptions ——-

dat_LM %>%
  group_by(task, proficiency) %>%
  identify_outliers(kotonarucontents_asunit)

## # A tibble: 3 × 37
##   task  proficiency id    name      trueid    ld   ld2  koto  nobe asunit clause
##   <ord> <ord>       <fct> <chr>     <fct>  <dbl> <dbl> <dbl> <dbl>  <dbl>  <dbl>
## 1 L     lower       27    WangYing… 28      50.7  3.8     50    75      3     10
## 2 L     upper       22    ZhangBoh… 22      41.3  4.33    74   126      4     12
## 3 M     middle      17    ZhenLinj… 17      40.4  3.83    38    57      2      6
## # ℹ 26 more variables: jyuzokuclause <dbl>, jyuzokuclauseperasunit <dbl>,
## #   clauseperasunit <dbl>, wariai <dbl>, wariai2 <dbl>, jyodoushi <dbl>,
## #   heiritsu <dbl>, kakari <dbl>, fukujyoshi <dbl>, ka <dbl>,
## #   setsuzokujyoshi <dbl>, kakujyoshi <dbl>, syujyoshi <dbl>, rentai <dbl>,
## #   hijiritsu <dbl>, setubimeishi <dbl>, setsubidoushi <dbl>,
## #   setsubikeiyoushi <dbl>, kotonarucontents <dbl>,
## #   kotonarucontents_asunit <dbl>, kotonarucontents_clause <dbl>, …

dat_LM %>%
  group_by(task, proficiency) %>%
  shapiro_test(kotonarucontents_asunit)

## # A tibble: 6 × 5
##   task  proficiency variable                statistic        p
##   <ord> <ord>       <chr>                       <dbl>    <dbl>
## 1 L     lower       kotonarucontents_asunit     0.825 0.0203  
## 2 L     middle      kotonarucontents_asunit     0.929 0.369   
## 3 L     upper       kotonarucontents_asunit     0.878 0.0824  
## 4 M     lower       kotonarucontents_asunit     0.944 0.572   
## 5 M     middle      kotonarucontents_asunit     0.669 0.000429
## 6 M     upper       kotonarucontents_asunit     0.935 0.437

ggqqplot(dat_LM, "kotonarucontents_asunit", ggtheme = theme_bw()) +
  facet_grid(task ~ proficiency)

dat_LM %>%
  group_by(task) %>%
  levene_test(kotonarucontents_asunit ~ proficiency)

## # A tibble: 2 × 5
##   task    df1   df2 statistic     p
##   <ord> <int> <int>     <dbl> <dbl>
## 1 L         2    32    0.0530 0.948
## 2 M         2    32    0.521  0.599

4.5.2 Computation ——-

res.aov <- anova_test(
  data = dat_LM, dv = kotonarucontents_asunit, wid = trueid,
  between = proficiency, within = task
)
get_anova_table(res.aov)

## ANOVA Table (type III tests)
## 
##             Effect DFn DFd     F     p p<.05   ges
## 1      proficiency   2  32 1.471 0.245       0.055
## 2             task   1  32 5.504 0.025     * 0.059
## 3 proficiency:task   2  32 0.096 0.909       0.002

interaction.plot(x.factor = dat_LM$task, trace.factor = dat_LM$proficiency, 
                 response = dat_LM$kotonarucontents_asunit, fun = mean,
                 type = "b", legend = TRUE, trace.label = "TASK")

4.6 kotonarucontents_clause LM ——-

4.6.1 Check assumptions ——-

dat_LM %>%
  group_by(task, proficiency) %>%
  identify_outliers(kotonarucontents_clause)

## # A tibble: 3 × 37
##   task  proficiency id    name      trueid    ld   ld2  koto  nobe asunit clause
##   <ord> <ord>       <fct> <chr>     <fct>  <dbl> <dbl> <dbl> <dbl>  <dbl>  <dbl>
## 1 L     middle      24    XuXingsh… 24      47.3  6.62    66   112      4      8
## 2 L     upper       23    TuChuyue  23      49.5  8.17    65    99      5      6
## 3 M     lower       27    WangYing… 28      45.1  6.83    56    91      5      6
## # ℹ 26 more variables: jyuzokuclause <dbl>, jyuzokuclauseperasunit <dbl>,
## #   clauseperasunit <dbl>, wariai <dbl>, wariai2 <dbl>, jyodoushi <dbl>,
## #   heiritsu <dbl>, kakari <dbl>, fukujyoshi <dbl>, ka <dbl>,
## #   setsuzokujyoshi <dbl>, kakujyoshi <dbl>, syujyoshi <dbl>, rentai <dbl>,
## #   hijiritsu <dbl>, setubimeishi <dbl>, setsubidoushi <dbl>,
## #   setsubikeiyoushi <dbl>, kotonarucontents <dbl>,
## #   kotonarucontents_asunit <dbl>, kotonarucontents_clause <dbl>, …

dat_LM %>%
  group_by(task, proficiency) %>%
  shapiro_test(kotonarucontents_clause)

## # A tibble: 6 × 5
##   task  proficiency variable                statistic      p
##   <ord> <ord>       <chr>                       <dbl>  <dbl>
## 1 L     lower       kotonarucontents_clause     0.888 0.132 
## 2 L     middle      kotonarucontents_clause     0.911 0.218 
## 3 L     upper       kotonarucontents_clause     0.896 0.142 
## 4 M     lower       kotonarucontents_clause     0.849 0.0419
## 5 M     middle      kotonarucontents_clause     0.919 0.274 
## 6 M     upper       kotonarucontents_clause     0.966 0.864

ggqqplot(dat_LM, "kotonarucontents_clause", ggtheme = theme_bw()) +
  facet_grid(task ~ proficiency)

dat_LM %>%
  group_by(task) %>%
  levene_test(kotonarucontents_clause ~ proficiency)

## # A tibble: 2 × 5
##   task    df1   df2 statistic     p
##   <ord> <int> <int>     <dbl> <dbl>
## 1 L         2    32     1.14  0.333
## 2 M         2    32     0.228 0.797

4.6.2 Computation ——-

res.aov <- anova_test(
  data = dat_LM, dv = kotonarucontents_clause, wid = trueid,
  between = proficiency, within = task
)
get_anova_table(res.aov)

## ANOVA Table (type III tests)
## 
##             Effect DFn DFd     F     p p<.05   ges
## 1      proficiency   2  32 0.079 0.924       0.003
## 2             task   1  32 1.921 0.175       0.027
## 3 proficiency:task   2  32 0.552 0.581       0.016

interaction.plot(x.factor = dat_LM$task, trace.factor = dat_LM$proficiency, 
                 response = dat_LM$kotonarucontents_clause, fun = mean,
                 type = "b", legend = TRUE, trace.label = "TASK")

4.7 contents_asunit LM ——-

4.7.1 Check assumptions ——-

dat_LM %>%
  group_by(task, proficiency) %>%
  identify_outliers(contents_asunit)

## # A tibble: 4 × 37
##   task  proficiency id    name      trueid    ld   ld2  koto  nobe asunit clause
##   <ord> <ord>       <fct> <chr>     <fct>  <dbl> <dbl> <dbl> <dbl>  <dbl>  <dbl>
## 1 L     lower       27    WangYing… 28      50.7  3.8     50    75      3     10
## 2 L     middle      24    XuXingsh… 24      47.3  6.62    66   112      4      8
## 3 L     upper       22    ZhangBoh… 22      41.3  4.33    74   126      4     12
## 4 M     middle      17    ZhenLinj… 17      40.4  3.83    38    57      2      6
## # ℹ 26 more variables: jyuzokuclause <dbl>, jyuzokuclauseperasunit <dbl>,
## #   clauseperasunit <dbl>, wariai <dbl>, wariai2 <dbl>, jyodoushi <dbl>,
## #   heiritsu <dbl>, kakari <dbl>, fukujyoshi <dbl>, ka <dbl>,
## #   setsuzokujyoshi <dbl>, kakujyoshi <dbl>, syujyoshi <dbl>, rentai <dbl>,
## #   hijiritsu <dbl>, setubimeishi <dbl>, setsubidoushi <dbl>,
## #   setsubikeiyoushi <dbl>, kotonarucontents <dbl>,
## #   kotonarucontents_asunit <dbl>, kotonarucontents_clause <dbl>, …

dat_LM %>%
  group_by(task, proficiency) %>%
  shapiro_test(contents_asunit)

## # A tibble: 6 × 5
##   task  proficiency variable        statistic       p
##   <ord> <ord>       <chr>               <dbl>   <dbl>
## 1 L     lower       contents_asunit     0.846 0.0375 
## 2 L     middle      contents_asunit     0.940 0.496  
## 3 L     upper       contents_asunit     0.932 0.403  
## 4 M     lower       contents_asunit     0.842 0.0332 
## 5 M     middle      contents_asunit     0.802 0.00999
## 6 M     upper       contents_asunit     0.940 0.500

ggqqplot(dat_LM, "contents_asunit", ggtheme = theme_bw()) +
  facet_grid(task ~ proficiency)

dat_LM %>%
  group_by(task) %>%
  levene_test(contents_asunit ~ proficiency)

## # A tibble: 2 × 5
##   task    df1   df2 statistic     p
##   <ord> <int> <int>     <dbl> <dbl>
## 1 L         2    32   0.00341 0.997
## 2 M         2    32   0.811   0.453

4.7.2 Computation ——-

res.aov <- anova_test(
  data = dat_LM, dv = contents_asunit, wid = trueid,
  between = proficiency, within = task
)
get_anova_table(res.aov)

## ANOVA Table (type III tests)
## 
##             Effect DFn DFd     F     p p<.05   ges
## 1      proficiency   2  32 2.354 0.111       0.090
## 2             task   1  32 5.283 0.028     * 0.052
## 3 proficiency:task   2  32 0.326 0.724       0.007

interaction.plot(x.factor = dat_LM$task, trace.factor = dat_LM$proficiency, 
                 response = dat_LM$contents_asunit, fun = mean,
                 type = "b", legend = TRUE, trace.label = "TASK")

4.8 contents_clause LM ——-

4.8.1 Check assumptions ——-

dat_LM %>%
  group_by(task, proficiency) %>%
  identify_outliers(contents_clause)

## # A tibble: 4 × 37
##   task  proficiency id    name      trueid    ld   ld2  koto  nobe asunit clause
##   <ord> <ord>       <fct> <chr>     <fct>  <dbl> <dbl> <dbl> <dbl>  <dbl>  <dbl>
## 1 L     middle      24    XuXingsh… 24      47.3  6.62    66   112      4      8
## 2 L     upper       23    TuChuyue  23      49.5  8.17    65    99      5      6
## 3 M     lower       27    WangYing… 28      45.1  6.83    56    91      5      6
## 4 M     middle      24    XuXingsh… 24      49.6  5.7     60   115      9     10
## # ℹ 26 more variables: jyuzokuclause <dbl>, jyuzokuclauseperasunit <dbl>,
## #   clauseperasunit <dbl>, wariai <dbl>, wariai2 <dbl>, jyodoushi <dbl>,
## #   heiritsu <dbl>, kakari <dbl>, fukujyoshi <dbl>, ka <dbl>,
## #   setsuzokujyoshi <dbl>, kakujyoshi <dbl>, syujyoshi <dbl>, rentai <dbl>,
## #   hijiritsu <dbl>, setubimeishi <dbl>, setsubidoushi <dbl>,
## #   setsubikeiyoushi <dbl>, kotonarucontents <dbl>,
## #   kotonarucontents_asunit <dbl>, kotonarucontents_clause <dbl>, …

dat_LM %>%
  group_by(task, proficiency) %>%
  shapiro_test(contents_clause)

## # A tibble: 6 × 5
##   task  proficiency variable        statistic      p
##   <ord> <ord>       <chr>               <dbl>  <dbl>
## 1 L     lower       contents_clause     0.876 0.0929
## 2 L     middle      contents_clause     0.933 0.409 
## 3 L     upper       contents_clause     0.828 0.0198
## 4 M     lower       contents_clause     0.845 0.0371
## 5 M     middle      contents_clause     0.939 0.486 
## 6 M     upper       contents_clause     0.947 0.594

ggqqplot(dat_LM, "contents_clause", ggtheme = theme_bw()) +
  facet_grid(task ~ proficiency)

dat_LM %>%
  group_by(task) %>%
  levene_test(contents_clause ~ proficiency)

## # A tibble: 2 × 5
##   task    df1   df2 statistic     p
##   <ord> <int> <int>     <dbl> <dbl>
## 1 L         2    32     0.590 0.560
## 2 M         2    32     0.270 0.765

4.8.2 Computation ——-

res.aov <- anova_test(
  data = dat_LM, dv = contents_clause, wid = trueid,
  between = proficiency, within = task
)
get_anova_table(res.aov)

## ANOVA Table (type III tests)
## 
##             Effect DFn DFd     F     p p<.05   ges
## 1      proficiency   2  32 1.508 0.237       0.051
## 2             task   1  32 0.608 0.441       0.008
## 3 proficiency:task   2  32 0.324 0.725       0.009

interaction.plot(x.factor = dat_LM$task, trace.factor = dat_LM$proficiency, 
                 response = dat_LM$contents_clause, fun = mean,
                 type = "b", legend = TRUE, trace.label = "TASK")

5 結果のまとめ

5.1 LとM（外在性負荷の有無）

5.1.1 語彙の複雑さの結果

• 語彙密度1(内容語数/総語数)では、認知負荷の主効果は有意でした。L > M。

• 1 AS UNITあたりの節数では、交互作用は有意でした(p = .047)。（6人の異常値を消しました） Lの時、下位群 < 中位群 、上位群 < 中位群。Mの時、下位群 < 上位群。

5.2 MとH（内在性負荷の有無）

5.2.1 語彙の複雑さの結果

• 語彙密度1(内容語数/総語数)では、認知負荷の主効果は有意でした。M < H。

• 1 AS UNITあたりの節数では、習熟度は有意でした。下位群 < 上位群。中位群 < 上位群。