This repository contains data files, stimulus materials, and scripts associated with the following psycholinguistic study.
Barr, D. J., & Seyfeddinipur, M. (2010). The role of fillers in listener attributions for speaker disfluency. Language and Cognitive Processes, 25, 441–455.
data_raw/: raw data files in CSV formatstimuli/: stimulus files (Windows BMP image files and WAV audio files)
This section provides R code for importing the raw data files as well as descriptions of the fields within each resulting table. To use the code, load the tidyverse package for R.
library("tidyverse")Contains the coordinates for the two screen regions (areas of interest, or AOIs) where pictures were displayed. Each picture was a 400x400 bitmap image, displayed on a computer screen with a 1024x768 resolution.
aoi <- read_csv(file.path("data_raw", "areas_of_interest.csv"),
col_types = "ciiii")| field | description |
|---|---|
| AOI | name of area of interest (left or right) |
| x1 | upper left X coordinate of image in pixels |
| y1 | upper left Y coordinate of image in pixels |
| x2 | bottom right X coordinate of image in pixels |
| y2 | bottom right Y coordinate of image in pixels |
Information about what images go with which blocks. Each block had three images, labelled A, B, C in the manuscript (see Barr & Seyfeddinipur, 2010, pp. 445–446).
block_images <- read_csv(file.path("data_raw", "block_images.csv"),
col_types = "iic")| field | description |
|---|---|
| block | unique integer identifying blocks |
| image | bitmap image number (e.g., 52 = 52.bmp) |
| imgrole | role the image plays (A, B, C) |
There were four versions of the stimulus materials created for counterbalancing purposes. This file has information about what condition each block was presented in across the four versions.
blocks <- read_csv(file.path("data_raw", "blocks.csv"),
col_types = "iicccccc")| field | description |
|---|---|
| vers | which version of the stimuli (1-4) |
| block | unique integer identifying blocks |
| blocktype | whether the block was critical or noncritical |
| trainspkr | who the speaker was during training |
| testspkr | who the speaker was at test |
| spkr | what speaker condition the block was in (same, diff) |
| pause | what pause condition the block was in (filled, baseline) |
| soundfile | soundfile that played at test |
Trial-by-trial information about training and test trials for each block and each version of the stimulus materials.
versions <- read_csv(file.path("data_raw", "versions.csv"),
col_types = "iiiiiicci")| field | description |
|---|---|
| vers | which version of stimulus materials (1, 2, 3, or 4) |
| block | unique integer identifying each block |
| blocktrial | integer identifying presentation order within the block |
| left | image number in the left area of interest (e.g., 32 means 32.BMP) |
| right | image number in the right area of interest |
| target | position of the target (0 = left, 1 = right) |
| soundfile | name of the wave file (e.g., 4 = 4.wav, 6umE = 6umE.wav) |
| who | identify of the speaker |
| showspeaker | display speaker identity before trial (0 = no, 1 = yes) |
Information about each subject in the experiment.
subjects <- read_csv(file.path("data_raw", "subjects.csv"),
col_types = "ici")| field | description |
|---|---|
| SubjID | unique integer identifying each subject |
| Gender | gender of the subject |
| vers | which version of the materials the subject received |
Information about each trial in the experiment.
trials <- read_csv(file.path("data_raw", "trials.csv"),
col_types = "iiiiiii")| field | description |
|---|---|
| TrialID | integer uniquely identifying each collected trial |
| SubjID | integer uniquely identifying each subject |
| block | block number |
| btrial | trial number within the block |
| tord | trial order |
| RT | response time in ms (from start of audio playback) |
| Selection | which image was selected (0 = left, 1 = right) |
Information about the position of the mouse cursor on each trial on a display with 1024x768 resolution.
mouse <- read_csv(file.path("data_raw", "mouse.csv"),
col_types = "iiii")| field | description |
|---|---|
| TrialID | integer uniquely identifying each collected trial |
| Msec | milliseconds from onset of soundfile |
| X | horizontal position of mouse cursor |
| Y | vertical position of mouse cursor |
library("lme4")
library("tidyverse")
## import the data
subjects <- read_csv(file.path("data_raw", "subjects.csv"),
col_types = "ici")
blocks <- read_csv(file.path("data_raw", "blocks.csv"),
col_types = "iicccccc")
versions <- read_csv(file.path("data_raw", "versions.csv"),
col_types = "iiiiiicci")
trials <- read_csv(file.path("data_raw", "trials.csv"),
col_types = "iiiiiii")
mouse <- read_csv(file.path("data_raw", "mouse.csv"),
col_types = "iiii")
## audiotimings:
## onset of filler/noise: 3056 ms
## onset of noun phrase: 4871 ms
fponset <- 3056L
nponset <- 4871L
## keep only the test trials
crit <- trials %>%
filter(btrial == 5L) %>%
inner_join(subjects %>% select(-Gender), "SubjID") %>%
semi_join(blocks %>% filter(blocktype == "critical"), c("vers", "block")) %>%
inner_join(versions %>% select(vers, block, btrial, target),
c("vers", "block", "btrial")) %>%
mutate(fac = (target * 2L) - 1L,
accurate = Selection == target) %>%
filter(accurate)
## time-lock and calculate distance of cursor from target
mousedist <- mouse %>%
inner_join(crit %>% select(TrialID, fac), "TrialID") %>%
mutate(ms = Msec - fponset,
sdist = fac * (X - 512) / 100L,
x = if_else(abs(sdist) > 1, 1 * sign(sdist), sdist),
win = factor(if_else(ms >= (nponset - fponset),
"expression", "filled interval"),
levels = c("filled interval", "expression"))) %>%
filter(ms >= 0L, ms <= 4471L) %>%
select(TrialID, win, ms, x)
## calculate distance travelled during window
## and match up with block info
dist <- mousedist %>%
group_by(TrialID, win) %>%
filter( (ms == min(ms)) | (ms == max(ms)) ) %>%
mutate(step = paste0("S", row_number())) %>%
select(-ms) %>%
pivot_wider(names_from = step, values_from = x) %>%
mutate(dist = S2 - S1) %>%
ungroup() %>%
inner_join(crit %>% select(SubjID, TrialID, vers, block), c("TrialID")) %>%
inner_join(blocks %>% select(vers, block, spkr, pause),
c("vers", "block")) %>%
select(SubjID, block, win, spkr, pause, dist)descriptives <- dist %>%
group_by(win, spkr, pause) %>%
summarize(mean_dist = mean(dist),
sd_dist = sd(dist))## `summarise()` regrouping output by 'win', 'spkr' (override with `.groups` argument)
| win | spkr | pause | mean_dist | sd_dist |
|---|---|---|---|---|
| filled interval | diff | baseline | 0.04 | 0.28 |
| filled interval | diff | filled | -0.01 | 0.33 |
| filled interval | same | baseline | 0.00 | 0.30 |
| filled interval | same | filled | 0.10 | 0.41 |
| expression | diff | baseline | 0.60 | 0.61 |
| expression | diff | filled | 0.60 | 0.71 |
| expression | same | baseline | 0.70 | 0.56 |
| expression | same | filled | 0.69 | 0.57 |
dat <- dist %>%
mutate(P = if_else(pause == "baseline", -.5, .5),
S = if_else(spkr == "diff", -.5, .5))
mod_fi <- lmer(dist ~ S * P +
(S * P || SubjID) +
(S * P || block),
data = dat %>% filter(win == "filled interval"))## boundary (singular) fit: see ?isSingular
summary(mod_fi)## Linear mixed model fit by REML ['lmerMod']
## Formula: dist ~ S * P + ((1 | SubjID) + (0 + S | SubjID) + (0 + P | SubjID) +
## (0 + S:P | SubjID)) + ((1 | block) + (0 + S | block) + (0 +
## P | block) + (0 + S:P | block))
## Data: dat %>% filter(win == "filled interval")
##
## REML criterion at convergence: 679.2
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -5.6347 -0.2451 -0.0427 0.0870 5.6437
##
## Random effects:
## Groups Name Variance Std.Dev.
## SubjID (Intercept) 0.0059464 0.07711
## SubjID.1 S 0.0253838 0.15932
## SubjID.2 P 0.0003054 0.01747
## SubjID.3 S:P 0.0156026 0.12491
## block (Intercept) 0.0000000 0.00000
## block.1 S 0.0003783 0.01945
## block.2 P 0.0000000 0.00000
## block.3 S:P 0.0004942 0.02223
## Residual 0.0973285 0.31198
## Number of obs: 1075, groups: SubjID, 92; block, 12
##
## Fixed effects:
## Estimate Std. Error t value
## (Intercept) 0.03367 0.01247 2.700
## S 0.03134 0.02590 1.210
## P 0.02969 0.01914 1.551
## S:P 0.15745 0.04079 3.860
##
## Correlation of Fixed Effects:
## (Intr) S P
## S -0.010
## P 0.005 -0.002
## S:P -0.002 0.005 -0.016
## convergence code: 0
## boundary (singular) fit: see ?isSingular
mod_re <- lmer(dist ~ S * P +
(S * P || SubjID) +
(S * P || block),
data = dat %>% filter(win == "expression"))## boundary (singular) fit: see ?isSingular
summary(mod_re)## Linear mixed model fit by REML ['lmerMod']
## Formula: dist ~ S * P + ((1 | SubjID) + (0 + S | SubjID) + (0 + P | SubjID) +
## (0 + S:P | SubjID)) + ((1 | block) + (0 + S | block) + (0 +
## P | block) + (0 + S:P | block))
## Data: dat %>% filter(win == "expression")
##
## REML criterion at convergence: 1859.6
##
## Scaled residuals:
## Min 1Q Median 3Q Max
## -3.5291 -0.6780 0.1039 0.5550 2.8348
##
## Random effects:
## Groups Name Variance Std.Dev.
## SubjID (Intercept) 2.395e-02 1.548e-01
## SubjID.1 S 4.163e-02 2.040e-01
## SubjID.2 P 1.033e-09 3.215e-05
## SubjID.3 S:P 1.200e-03 3.464e-02
## block (Intercept) 5.484e-02 2.342e-01
## block.1 S 1.587e-02 1.260e-01
## block.2 P 2.149e-03 4.636e-02
## block.3 S:P 0.000e+00 0.000e+00
## Residual 2.891e-01 5.377e-01
## Number of obs: 1070, groups: SubjID, 92; block, 12
##
## Fixed effects:
## Estimate Std. Error t value
## (Intercept) 0.640076 0.071435 8.960
## S 0.101991 0.053504 1.906
## P -0.008024 0.035552 -0.226
## S:P -0.012548 0.065966 -0.190
##
## Correlation of Fixed Effects:
## (Intr) S P
## S -0.002
## P 0.003 -0.001
## S:P 0.000 0.007 -0.012
## convergence code: 0
## boundary (singular) fit: see ?isSingular
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent