University of Central Florida Undergraduate Research Journal - Cross-Modal Distraction on Simultaneous Translation: Language Interference in Spanish-English Bilinguals
US tab

Results

There were three groups of participants: all subjects, monolinguals, and bilinguals; the data for each group were analyzed separately. One-way analysis of variance (ANOVA) was conducted for every independent variable. Tests were done at a 95% confidence interval and found to be significant at the p < 0.05 level. Trial 1 was excluded from analysis due to software issues and participants forgetting to press 1 to advance. The control (match) condition was used in analysis because both groups were affected differently. An accuracy score of 1 means no errors were made, resulting in a 100% accuracy response. For example, means equaling 0.98 have an accuracy of 98%, and a higher accuracy score than 0.87. For reaction times, the lower the time (in milliseconds), the faster subjects were to respond. Performance was measured by number of errors (accuracy) and latencies (reaction times). More interference is defined as slower reaction times and lower accuracies.

All Subjects

There were three left-handed subjects (2 bilinguals, 1 monolingual), with handedness having no significant effect on accuracy nor reaction time. Neither caffeine nor gender had an influence on performance. Trial number was significant for both accuracy F(70, 2272) = 1.51, p = 0.005, and reaction time F(70, 2272) = 2.39, p < 0.0001. Four variables from the music tendency questionnaire were tested for influence on performance: frequency of music played when studying, perceived loudness of the music, perceived effects of the music on concentration levels, and if music helped or hurt studying performance. None of these variables had any significant effect on performance. However, there was a significant difference between the performance of bilinguals and monolinguals for both accuracy F(1, 2341) = 36.55, p < 0.0001, and reaction time F(1, 2341) = 40.79, p < 0.0001. These differences are shown in Figures 1 and 2 below.

Figure 1: Average % accuracies for both groups, including all conditions.

Note. Monolinguals (M = 0.99, SD = 0.10); Bilinguals (M = 0.95, SD = 0.22).

Figure 1: Average reaction times for both groups, including all conditions.

Note. Monolinguals (M = 943.79, SD = 1560.54); Bilinguals (M = 1401.01, SD = 1897.13).

Monolinguals

Effects of the two conditions (match or mismatch), ear of the distractor stimuli, and whether the auditory distractor word went to both ears or an individual ear had no significant effects on accuracy or reaction time. Regarding the linguistic side of the experimental setup, four variables were tested for effects on performance: frequency of the screen word, frequency of the audio word, length of the screen word, and length of the audio word. Frequency of the screen or audio word had no significant results for accuracy nor reaction time. Length of screen word had a significant effect on accuracy only F(5, 1201) = 2.61, p = 0.024. A negative correlation was found between accuracy and length of screen word: the longer the word, the lower the accuracy became. Minimum word length was 2 letters (M = 1.00, SD = 0.00), and maximum word length was 7 letters (M = 0.96, SD = 0.20). Neither the length of audio word nor phonetically similar words yielded significant effects on participant’s performance.

Bilinguals

Two variables from the background section of the survey were tested using an ANOVA: proficiency level and first language. Self-proficiency level questions were asked, and the results of the proficiency tests were graded. If a subject correctly responded to more than 50% of questions in any section then they were said to be proficient in that language. There were five levels of scoring for Spanish and three levels of proficiency for English. All subjects scored as highly proficient in the English section. Only five of the 16 bilinguals scored highly proficient in Spanish, while 11 of the 16 self-reported they were highly proficient in Spanish. The questions were based on grammar, and the most appropriate and grammatically correct was considered correct. Half of subjects listed Spanish as their first language, 4/16 listed English, and 4/16 listed both.

Whether the auditory distractor word went to both ears or an individual ear had a significant difference for accuracy only F(1,1134) = 10.96, p = 0.001, with accuracy rates higher for both ear stimuli (M = 0.96, SD = 0.19) than an individual ear (M = 0.92, SD = 0.27). The language of the target word on the screen yielded significance for accuracy F(1, 1134) = 26.66, p < 0.0001, and reaction time F(1, 1134) = 4.72, p = 0.030. Accuracy was higher when the word on screen was in English (M = 0.98, SD = 0.13) than Spanish (M = 0.92, SD = 0.28), and reaction time was lower for English words (M = 1278.65, SD = 1914.046) than Spanish words (M = 1522.93, SD = 1873.85). The language of the auditory distractor word had a significant influence on accuracy only F(1, 1134) = 10.45, p = 0.001. More errors occurred when the language was Spanish (M = 0.93, SD = 0.26) than English (M = 0.97, SD = 0.17). The relationship between the word on screen and the audio word had a significant effect on accuracy only F(1, 1134) = 7.23, p = 0.007, with more errors in the experimental (mismatch) condition (M = 0.93, SD = 0.25) than the control (match) condition (M = 0.97, SD = 0.18). The ear to which the distractor word arrived had a significant effect on accuracy F(2, 1133) = 6.28, p = 0.002, with both ears having the highest accuracy (M = 0.96, SD = 0.19) and the left ear having a higher accuracy (M = 0.93, SD = 0.25) than the right (M = 0.90, SD = 0.30). A post-hoc test of least significant difference (LSD) was computed for the ear variable, with the results listed in Table 1.

Table 1: LSD results for Ear x Accuracy One-Way ANOVA test.

Note. The only significant comparison is 2 (right ear) vs 3 (both ears).

The condition of the word and distractor stimuli had significant effects on the accuracy of the participants (F(3, 1132) = 10.10, p < 0.0001). Conditions are listed in the format (screen word language, auditory word language). The highest accuracy was in the second condition (E, S) with (M = 0.99, SD = 0.08); condition 4 (S,S) produced the most errors (M = 0.91, SD = 0.29). The first condition (E, E) had (a mean accuracy of 0.98, and a standard deviation of 0.14), and the third condition (S, E) had (a mean accuracy of 0.94, and a standard deviation of 0.23). This variable was also significant for reaction time in the LSD posttest for condition 1 (E, E) versus 4 (S, S) only (MD = -261.63, SE = 129.86, p = 0.044), even though ANOVA reaction time was not significant.

Table 2: LSD test for Condition x Accuracy One-Way ANOVA test.

*p < 0.05

**p < 0.01

Note. Only significant differences are 4 (S,S) vs 1 (E,E), and 4 (S,S) vs 2 (E,S).

The same four variables tested for monolinguals were tested for effects on performance in bilinguals. Unlike the monolingual participants, screen word frequency had significant effects on accuracy F(35, 1100) = 2.38, p < 0.0001, and audio word frequency on accuracy F(68, 1067) = 1.64, p = 0.001. The screen words “pasa” (M = 0.75, SD = 0.44), “tiene” (M = 0.83, SD = 0.39), and “eres” (M = 0.86, SD = 0.35) had the lowest accuracy scores. A positive correlation was observed, with higher frequency words having a higher accuracy score. The audio words “quiero” (M = 0.75, SD = 0.45), “pasa” (M = 0.75, SD = 0.45), “casa” (M = 0.75, SD = 0.45) had the lowest accuracy scores. Reaction times for both frequency of screen word and frequency of audio word were not significant. The relationship between the length of screen word and reaction time was not significant, nor was the relationship between length of screen word and accuracy. The length of audio word was also not found to be significant in its relationship to accuracy, nor with reaction time. Four phonetic conditions were named: not phonetically similar, phonetically similar, phonetically similar with same word but audio word in English, and phonetically similar with same word but audio word in Spanish. Phonetic conditions had no significant influence on subjects’ accuracy scores nor reaction times. Yet an LSD test for Phonetic type and Accuracy indicated a significant difference between phonetically similar (M = 0.91, SD = 0.22) and phonetically similar with same word but audio word in Spanish (M = 1.00, SD = 0.00), (MD = - 0.09, SE = 0.05, p = 0.052).

Discussion >>