Datasets and R scripts for modelling Czech translation counterparts of Romance causative constructions

README – Supplementary materials

This repository contains the data and code used in the article:

“Predicting translation counterparts in causative constructions”

FILES INCLUDED

DATA
- Causatives_table_full_Spanish.xlsx
  Full annotated dataset for Spanish causative constructions and their Czech translation counterparts.

- Causatives_table_full_Italian.xlsx
  Full annotated dataset for Italian causative constructions and their Czech translation counterparts.

- causatives_es_revised.csv
  Revised Spanish dataset used for statistical modelling, including the variable COMP_CLASS.

- causatives_it_revised.csv
  Revised Italian dataset used for statistical modelling, including the variable COMP_CLASS.

CODE
- brms_causatives_models.R
  R script used to estimate the Bayesian multinomial regression models reported in the paper.
  The models were estimated using the brms package (Stan backend).

MODEL OUTPUT
- brms_summary_causatives.txt
  Summaries of the fitted Bayesian multinomial models for both datasets.

MODEL SPECIFICATION

The models estimate the probability of selecting a Czech translation counterpart (TYPE)
as a function of:

- VALENCY (valency of the base verb)
- COMP_CLASS (class of the complement)

with random effects for:

- VERB (random intercepts and slopes for VALENCY)
- TRANSLATOR (random intercepts)

The models were fitted using Hamiltonian Monte Carlo as implemented in Stan
via the brms package in R.

> summary(brms_it_noanim)
 Family: categorical 
  Links: muB = logit; muC = logit; muD = logit; muE = logit; muF = logit; muX = logit 
Formula: TYPE ~ VALENCY + COMP_CLASS + (1 + VALENCY | VERB) + (1 | TRANSLATOR) 
   Data: it (Number of observations: 1394) 
  Draws: 4 chains, each with iter = 4000; warmup = 1000; thin = 1;
         total post-warmup draws = 12000

Multilevel Hyperparameters:
~TRANSLATOR (Number of levels: 15) 
                  Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS
sd(muB_Intercept)     0.41      0.25     0.03     0.97 1.00     2192     2556
sd(muC_Intercept)     0.76      0.29     0.32     1.44 1.00     4258     5629
sd(muD_Intercept)     0.51      0.35     0.03     1.33 1.00     3006     4035
sd(muE_Intercept)     0.87      0.51     0.10     2.11 1.00     2721     2697
sd(muF_Intercept)     0.39      0.26     0.02     1.00 1.00     2508     3392
sd(muX_Intercept)     0.86      0.33     0.34     1.66 1.00     3778     4805

~VERB (Number of levels: 63) 
                                                Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS
sd(muB_Intercept)                                   1.43      0.22     1.05     1.91 1.00     3231     5719
sd(muB_VALENCYreflexive)                            1.47      0.97     0.08     3.74 1.00     3007     3738
sd(muB_VALENCYtransitive)                           0.73      0.52     0.03     1.93 1.01     1353     2253
sd(muC_Intercept)                                   1.87      0.32     1.29     2.57 1.00     2879     5073
sd(muC_VALENCYreflexive)                            0.85      0.72     0.04     2.65 1.00     2437     3604
sd(muC_VALENCYtransitive)                           0.67      0.53     0.03     1.99 1.00     1439     1685
sd(muD_Intercept)                                   1.97      0.46     1.20     3.05 1.00     3270     5731
sd(muD_VALENCYreflexive)                            1.83      1.39     0.08     5.17 1.00     3972     5000
sd(muD_VALENC . . .