The effects of regular and enhanced captions on incidental vocabulary acquisitionon incidental vocabulary acquisition

eliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaelia eliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaelia eliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaelia eliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaelia eliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaeliaelia ELIA 23, 2023, pp. 15-50 DOI: http://dx.doi.org/10.12795/elia.2023.i23.01 15 THE EFFECTS OF REGULAR AND ENHANCED CAPTIONS THE EFFECTS OF REGULAR AND ENHANCED CAPTIONS ON INCIDENTAL VOCABULARY ACQUISITIONON INCIDENTAL VOCABULARY ACQUISITION EFECTOS DE LA SUBTITULACIÓN EN L2 Y DEL REALCE EFECTOS DE LA SUBTITULACIÓN EN L2 Y DEL REALCE TEXTUAL EN LA ADQUISICIÓN INCIDENTAL DE TEXTUAL EN LA ADQUISICIÓN INCIDENTAL DE VOCABULARIOVOCABULARIO Rebeca Finger-Bou Rebeca Finger-Bou Universitat de Barcelona [email protected] Carmen MuñozCarmen Muñoz Universitat de Barcelona [email protected] AbstractAbstract Research has demonstrated that watching audiovisual materials in the target language (L2) through using captions can foster vocabulary learning. Some studies have redirected learners’ attention by enhancing specific parts of those captions, thus increasing their saliency. This study explores the effects of regular and enhanced captions on incidental vocabulary acquisition by L1-Spanish/Catalan learners of English through short exposure to a documentary. It also analyses how vocabulary learning might be affected by previous vocabulary knowledge and language learning aptitude. Two randomly distributed groups were formed. One was provided with 16 The effects of regular and enhanced captions... ELIA 23, 2023, pp. 15-50 DOI: http://dx.doi.org/10.12795/elia.2023.i23.01 regular captions, and the other with enhanced captions (target words in yellow and bold). Vocabulary gains were assessed through pre- and post-tests that tapped into meaning recall, meaning recognition and form recognition knowledge. The results showed that the difference between the pre-test and the post-test was greater in the students with enhanced subtitles, but the difference was not significant between the two groups in the post-test. Vocabulary size emerged as the most significant predictor, but not aptitude. Retrospective questionnaires on participants’ focus of attention reported an emphasis on captions and comprehension. Analyses indicate that paying attention to the enhanced items might have positively affected acquisition and retention. This study has provided new evidence that shows the potential advantage of multimodal input as an accessible pedagogical tool for acquiring languages. Keywords : multimodal input, incidental vocabulary acquisition, enhanced captions, regular captions, individual differences, focus of attention. ResumenResumen Un buen número de estudios ha demostrado que el visionado de material audiovisual en la lengua meta (L2) con la ayuda de subtítulos en la L2 facilita el aprendizaje de vocabulario. Algunos de ellos han dirigido la atención de los participantes realzando partes específicas de esos subtítulos, aumentando así su prominencia. Este estudio explora los efectos de los subtítulos normales y realzados en la adquisición incidental de vocabulario por parte de estudiantes de inglés cuyas lenguas maternas son el español y el catalán, a través de una breve exposición a un documental. También analiza cómo el aprendizaje de vocabulario puede verse afectado por el conocimiento previo de vocabulario y la aptitud para aprender idiomas. Se formaron dos grupos aleatoriamente. Uno visionó el documental con subtítulos normales en inglés y, el otro, con subtítulos realzados (palabras clave en amarillo y negrita). El aprendizaje de vocabulario ELIA 23, 2023, pp. 15-50 DOI: http://dx.doi.org/10.12795/elia.2023.i23.01 17 Rebeca Finger-Bou and Carmen Muñoz se evaluó mediante pruebas previas y posteriores de recordar el significado y la forma, y de reconocer el significado. Los resultados mostraron que la diferencia entre las pruebas era mayor en los estudiantes con subtítulos realzados, pero la diferencia no era significativa entre los dos grupos en la prueba posterior. El tamaño de vocabulario previo resultó ser la variable explicativa más significativa, pero no así la aptitud. Los cuestionarios retrospectivos revelaron atención en los subtítulos y la comprensión. Los análisis indican que prestar atención a los elementos realzados podría haber afectado positivamente a la adquisición y posterior retención de vocabulario. Este estudio proporciona nueva evidencia sobre el beneficio potencial del input multimodal como una herramienta pedagógica accesible para la adquisición de idiomas. Palabras clave : input multimodal, adquisición incidental de vocabulario, subtítulos, subtítulos realzados, diferencias individuales, foco de atención. 1. Literature Review1. Literature Review The massive access to multimodal second language (L2) input in modern times is one of the major reasons that may explain the growing importance of multimodal input in the area of Second Language Acquisition (SLA) (Montero Perez, 2022). Several studies have revealed that language learners are indeed motivated to watch television in an L2 (Peters & Muñoz, 2020). Mayer’s (2014) cognitive theory of multimedia learning ascertains that language learning is greater when information is not only processed in spoken mode but also in written mode, for learners produce mental connections between the aural and the visual information, providing that there is a temporal proximity. In that sense, television programs also supply L2 learners with repeated encounters with both high-frequency and low-frequency words (Rodgers & Webb, 2011), which could potentially fuel L2 vocabulary growth with regular viewing (Feng & Webb, 2019). ELIA 23, 2023, pp. 15-50 DOI: http://dx.doi.org/10.12795/elia.2023.i23.01 18 The effects of regular and enhanced captions... Considering multimodal input as a combination of pictorial information, written verbal information –in the form of captions or subtitles–, and acoustic verbal input (Peters & Muñoz, 2020), multimodal input can enhance language learning whenever all channels, that is, visual and verbal information, are activated simultaneously (Montero Perez, 2022). 1.1. Vocabulary Acquisition through Multimodal Input1.1. Vocabulary Acquisition through Multimodal Input Based on Paivio’s (1986) dual coding theory, Mayer’s (2014) model of multimedia learning proposes that learning is more effective with both words and pictures compared to when words or pictures alone are present. In that sense, research has sought to throw light upon the effects of captions on language acquisition in general and has succeeded in doing so by demonstrating the statistically significant advantage of participants who watch multimodal materials with captions (subtitles in the L2 or original language) or subtitles (in the L1). The use of captions has hence been corroborated to have a positive impact on L2 comprehension (Montero Perez et al., 2014), vocabulary (Pujadas & Muñoz, 2019; Suárez & Gesa, 2019) and grammar (Pattemore & Muñoz, 2020) learning, as word recognition is assisted by the breaking down of speech into separated items. 1.2. Enhanced Captioning1.2. Enhanced Captioning Ample evidence exists confirming vocabulary acquisition through multimodal input and the use of captions (Montero Perez, 2022; Muñoz, 2022). Furthermore, recent studies have intended to redirect and refocus learners’ attention by typographically enhancing specific parts of those captions (Majuddin et al., 2021; Montero Perez et al., 2014; Lee & Révéscz, 2020), as noticing has been widely recognised as ELIA 23, 2023, pp. 15-50 DOI: http://dx.doi.org/10.12795/elia.2023.i23.01 19 Rebeca Finger-Bou and Carmen Muñoz a relevant and essential part of language learning (Schmidt, 1994). When the material salience of single-words is typographically enhanced in captions, learners of an L2 will expectedly pay more attention and learn new L2 vocabulary items (Montero Perez et al., 2015; Puimège et al., 2022). However, research has not revealed a clear advantage of textually enhanced captions over regular or unenhanced captions, and the different techniques that have been used (e.g., highlighting, bolding or using keywords) have shown mixed results (Montero Perez, 2022). 1.3. Individual Differences1.3. Individual Differences In this line of research, it has been found that learner-related factors such as proficiency level (Montero Perez et al., 2013; Suárez & Gesa, 2019), previous vocabulary knowledge (Majuddin, 2020; Montero Perez et al., 2014; Peters & Webb, 2018; Rodgers & Webb, 2019), or working memory (Pattemore & Muñoz, 2020) may impact vocabulary gains and the processing of multimodal input. Particularly, prior vocabulary knowledge is one of the most important factors affecting incidental vocabulary acquisition (Peters & Webb, 2018). In contrast, only a few studies have analysed the association of language learning aptitude with vocabulary learning through multimodal input. In Suárez and Gesa’s (2019) study, for example, aptitude was found to be statistically significant only in the learning of target word meanings, not forms, after exposure to captioned videos. Moreover, the authors also found a main effect for proficiency on the learning scores for both target word forms and meanings. Contrary to the mentioned study, however, Pattemore and Muñoz (2020) did not find any significant effect of the LLAMA tests on grammar construction learning from captioned audio-visual exposure. The authors propose that learners might cease to rely on language learning aptitude when surpassing a certain proficiency threshold, as suggested by Winke (2013). However, previous studies ELIA 23, 2023, pp. 15-50 DOI: http://dx.doi.org/10.12795/elia.2023.i23.01 20 The effects of regular and enhanced captions... have not used enhanced captions and it is still unknown if vocabulary size and aptitude play a similar role as when captions are unenhanced. 1.4. Learners’ Focus of Attention1.4. Learners’ Focus of Attention Learners’ focus of attention while viewing captioned material needs to be investigated. This may be done through retrospective questionnaires that resemble think-aloud verbal protocols (Winke, 2013), which allows researchers to extract subjective and selfreflective information on the conducted experiments. With these conditions in mind, the present study will aim at analysing the effects of caption enhancement on incidental vocabulary acquisition in L1-Spanish/Catalan students of English as a Foreign Language. More specifically, this study pursues to answer the following research questions: 1.5. Research Questions1.5. Research Questions 1. Is there evidence of incidental vocabulary acquisition after viewing a captioned documentary? If so, is the potential learning retained after two weeks? 2. Does the enhancement of captions have an effect on incidental vocabulary acquisition in comparison to regular captions in L1-Spanish/Catalan EFL learners? If so, is the potential learning retained after two weeks? 3. To what extent do previous vocabulary knowledge and language learner’s aptitude, as measured by LLAMA B and D, play a role in potential vocabulary gains through viewing a captioned documentary? ELIA 23, 2023, pp. 15-50 DOI: http://dx.doi.org/10.12795/elia.2023.i23.01 21 Rebeca Finger-Bou and Carmen Muñoz 4. How do enhanced and regular captions affect L1-Spanish/ Catalan EFL learners’ self-reported focus of attention when viewing a captioned documentary? 2. Methodology2. Methodology 2.1. Participants2.1. Participants The participants of this study consisted of 31 L1-Spanish/Catalan learners of English, who were enrolled in different EFL levels at a language school in a small city in the province of Tarragona, Spain. The participants included 18 adolescents and 13 adults, and ages varied from 14 to 64 years old (M = 22.46, SD = 11.11). All participants had the same teacher, who collaborated with the researchers. A background information questionnaire was handed out prior to the experiment so that personal information such as age, sex and previous education could be collected, as well as information on external sources of input, that is, out-of-school exposure to L2 media. Parental consent forms were distributed to all underaged students, whereas adult learners signed to accept their own participation. Two randomly distributed groups were formed. Group 1 was provided with regular captions (RC), and group 2 with enhanced captions (EC). Table 1: Table 1: Descriptive information of participants Age Level Sex Mean SD Min Max B1 B1+ B2 C1 Regular (n = 15) 19.65 6.03 14.00 37.60 2 5 5 3 6 female, 9 male Enhanced (n = 16) 25.09 14.06 15.00 64.00 4 2 6 4 10 female, 6 male All participants (n = 31) 22.46 11.11 14.00 64.00 6 7 11 7 16 female, 15 male ELIA 23, 2023, pp. 15-50 DOI: http://dx.doi.org/10.12795/elia.2023.i23.01 22 The effects of regular and enhanced captions... 2.2. Target Constructions2.2. Target Constructions A total of 21 target words (TWs) were chosen from the script of Viral: The 5G Conspiracy Theory (Livingston, 2020), a 25-minute documentary from the BBC that was released in 2020 in which the conspiracy theories that erupted ever since the beginning of the global pandemic are critically reviewed. Words in the documentary were assessed through LexTutor to extract 21 TWs (from the 1k, 2k, Academic Word List, and OFF types) that appeared at least twice in the audio-visual material. The TWs included in the final analysis were spread, lockdown, harmful, jaw, mad, murderer, opposed, threat, approach, linked, network, remove, appealing, arson, carer, cell, clap, lockdown, mast, ripper, and illness . An enhanced version of the regular captions was created with the application SubtitleEdit (v3.5.18) and embedded on the video with HandBrake (v1.3.0-v1.3.3), where TWs were presented in yellow and bold. Moreover, a virtually equivalent number of words that belonged to the same frequency lists and did not appear in the documentary were selected to function as distractors. TWs and distractors were revised and approved by the participants’ teacher. 2.3. Instruments2.3. Instruments Vocabulary gains were assessed through pre-, immediate post- and delayed post-tests that tapped into meaning knowledge at the level of recall and recognition, to gather information at the two different sensitivities based on Nation’s (2001) nine components of word knowledge. Additionally, immediate post- and delayed post-tests on form recognition were included to assess whether learners remembered seeing TWs on the documentary, as noticing a new word is the first step towards acquisition and it has been suggested that captions generally help learners with both written and aural form recognition and with developing form-meaning connections (Pujadas ELIA 23, 2023, pp. 15-50 DOI: http://dx.doi.org/10.12795/elia.2023.i23.01 23 Rebeca Finger-Bou and Carmen Muñoz & Muñoz, 2020). When taking the tests, TWs and distractors were provided through an audio file recorded with the teacher’s voice that repeated each word twice, whilst the written forms could be read in the paper where participants were to answer, which guaranteed them encountering the same modalities in the tests as those in the multimodal input, and therefore all channels of input were reactivated simultaneously. All tests were piloted by five L1-Spanish/ Catalan learners of English whose ages (M = 36.8, SD = 14.9) ranged very similarly to the study’s participants (M = 22.46, SD = 11.11). Pre-test scores in the two languages were similar (SPA (n = 24) = 72.4% vs CAT (n = 7) = 70.7%)1. Participants’ previous vocabulary knowledge was measured by means of Meara and Miralpeix’s (2015) V_YesNo (v1.01) test. Language learning aptitudes for vocabulary learning and listening for new words, the most relevant to this study, were measured through two of the subtests of Meara and Rogers’s (2019) LLAMA suite of tests: LLAMA B (v3.00) and D (v3.00). LLAMA B consists of a vocabulary learning task in which participants must remember large amounts of words. This subtest measures the users’ ability to attach unfamiliar names to unfamiliar objects LLAMA D, on the other hand, is a phonetic memory subtest, where users must recognise spoken language that they were exposed to a short while earlier (Rogers et al., 2017). An additional test with three comprehension questions that had no relation to the TWs was utilised in the immediate post-test. Furthermore, Likert-scale questionnaires resembling think-aloud verbal protocols (Winke, 2013) were also distributed so as to collect retrospective information on learners’ self-reported focus of attention. 1 A series of chi-squared tests revealed that only for one item ( ripper ) the difference between difficulty indexes in the two languages was statistically significant ( p = .011). Perceptions reflected in the retrospective questionnaire did not account for any extra difficulty when dealing with different items in the meaning recognition tests. ELIA 23, 2023, pp. 15-50 DOI: http://dx.doi.org/10.12795/elia.2023.i23.01 30 The effects of regular and enhanced captions... More specifically, pairwise comparisons between Level and Time show significant differences, on the one hand, between pre-test and delayed post-test only for B1 ( p = .003) and C1 ( p = .018), with B1+ nearly significant ( p = .052) and B2 differences non-significant at all ( p = .816). On the other hand, only the C1 level managed to show significant differences between pre- and immediate post-tests as well ( p = .018). Meaning Recognition At the meaning recognition level, the GLMM showed a significant main effect of Vocabulary Size ( F (1,1947) = 32.152, p < .001), alongside a non-significant main effect of Time ( F (2, 1947) = 1.303, p = .272) and a significant interaction of Vocabulary Size and Time ( F (2,1947) = 3.472, p = .031) (see Table 6). Pairwise comparisons showed significant differences between pre- and immediate post-test as well as pre- and delayed post-test ( p = .014 and p = .021, respectively). A non-significant difference between immediate post- and delayed post-test ( p = .226) further suggests that word knowledge was not significantly lost (see Figure 3). [Level=2] * [Time=1] .300 .482 .623 .533 -.645 1.245 1.350 .525 3.472 [Level=3] * [Time=1] 1.029 .416 2.474 .013.013 .213 1.845 2.799 1.238 6.327 [Level=1] * [Time=2] -.535 .506 -1.057 .290 -1.529 .458 .585 .217 1.581 [Level=2] * [Time=2] -.262 .495 -.529 .597 -1.233 .709 .769 .291 2.031 [Level=3] * [Time=2] .544 .438 1.244 .214 -.314 1.402 1.723 .731 4.065 ELIA 23, 2023, pp. 15-50 DOI: http://dx.doi.org/10.12795/elia.2023.i23.01 31 Rebeca Finger-Bou and Carmen Muñoz Table 6: Results from the GLMM: fixed coefficients for meaning recognition regardless of experimental group Coefficient SE t Sig. 95% Confidence Interval Exp (Coefficient) 95% CI for Exp (Coefficient) Lower Upper Lower Upper V_Size .001 .000 5.286 < .001< .001 .001 .001 1.001 1.001 1.001 Time=1 1.036 .773 1.339 .181 -.481 2.552 2.817 .618 12.833 Time=2 -0.076 .838 -.091 .928 -1.719 1.567 0.927 .179 4.790 V_Size * [Time=1] < .001 < .001 -2.048 .041.041 -.001 < .001 1.000 .999 1.000 V_Size * [Time=2] < .001 < .001 .366 .715 < .001 < .001 1.000 1.000 1.000 Figure 3: Meaning recognition by time n.s. 🡒 p > .05 ** 🡒 p < .05 *** 🡒 p < .001 ELIA 23, 2023, pp. 15-50 DOI: http://dx.doi.org/10.12795/elia.2023.i23.01 32 The effects of regular and enhanced captions... Form Recognition As summarized in Table 7, the GLMM revealed a significant main effect of Vocabulary Size ( F (1, 1298) = 15.439, p < .001), a nonsignificant main effect of Time ( F (1, 1298) = 3.208, p = .074) and a non-significant interaction between these two fixed factors ( p = .516) for form recognition results. Table 7: Results from the GLMM: fixed coefficients for form recognition regardless of experimental group Coefficient SE t Sig. 95% Confidence Interval Exp (Coefficient) 95% CI for Exp (Coefficient) Lower Upper Lower Upper V_Size < .001 < .001 3.881 < .001< .001 < .001 .001 1.000 1.000 1.001 Time=2 1.078 .602 1.791 .074 -.103 2.258 2.938 .902 9.568 V_Size* [Time=2] < .001 < .001 -.650 .516 < .001 < .001 1.000 1.000 1.000 Form recognition differences between immediate post-and delayed post-tests showed a significant reduction of accuracy ( p < .001) from one time to the other (Figure 4). 3.2. The Effects of Enhanced Captions3.2. The Effects of Enhanced Captions The second research question focused on the effects of enhanced captions on L2 vocabulary acquisition in comparison to regular captions. As has been seen in Table 4, all participants gained knowledge no matter their proficiency level, as differences between pre-, immediate post- and delayed post-tests showed a general and significant increase in scores. GLMMs were used to estimate differences between experimental groups. After analysing which independent variables had significant effects, only Vocabulary Size and Time were maintained as fixed factors and Subject and Item as ELIA 23, 2023, pp. 15-50 DOI: http://dx.doi.org/10.12795/elia.2023.i23.01 33 Rebeca Finger-Bou and Carmen Muñoz random intercepts, with Caption (regular, enhanced) as the new included fixed factor. Meaning Recall The GLMM for meaning recall showed significant main effects of Vocabulary Size ( F (1, 1946) = 27.692, p < .001) and Time ( F (2, 1946) = 11.067, p < .001), with non-significant effects of Caption and the interaction between Caption and Time ( p =.759 and p = .289, respectively) (see Table 8). Even though comparisons at specific testing times were nonsignificant between the two groups ( p = .898, p = .394 and p = .890 for pre-, post- and delayed post-test), the pairwise contrasts showed significant differences between both pre- and immediate post-test ( p = Figure 4: (Target) Form recognition by time n.s. 🡒 p > .05 ** 🡒 p < .05 *** 🡒 p < .001 ELIA 23, 2023, pp. 15-50 DOI: http://dx.doi.org/10.12795/elia.2023.i23.01 34 The effects of regular and enhanced captions... .002) and pre- and delayed post-test ( p = .020) in the enhanced captions group, whereas the regular captions group only showed significant differences between pre- and delayed post-tests ( p = .009) (Figure 5). Figure 5: Meaning recall by time by captions n.s. 🡒 p > .05 ** 🡒 p < .05 *** 🡒 p < .001 Table 8: Results from the GLMM: fixed coefficients of meaning recall with caption distinction Coefficient SE t Sig. 95% Confidence Interval Exp (Coefficient) 95% CI for Exp (Coefficient) Lower Upper Lower Upper V_Size .001 < .001 5.262 < .001< .001 .001 .001 1.001 1.001 1.001 Caption=1 .064 .465 .138 .890 -.847 .975 1.066 .429 2.651 Time=1 -.556 .208 -2.675 .008.008 -.964 -.148 .573 .381 .862 Time=2 .224 .212 1.058 .290 -.191 .640 1.251 .826 1.896 [Caption=1] * [Time=1] -.123 .301 -.411 .681 -.713 .466 .884 .490 1.594 [Caption=1] * Time=2] -.463 .305 -1.520 .129 -1.061 .135 .629 .346 1.144 ELIA 23, 2023, pp. 15-50 DOI: http://dx.doi.org/10.12795/elia.2023.i23.01 35 Rebeca Finger-Bou and Carmen Muñoz Meaning Recognition For meaning recognition, the GLMM showed a significant main effect of Vocabulary Size ( F (1, 1941) = 30.718, p < .001), as well as a marginally significant interaction between Vocabulary Size and Time ( F (2, 1941) = 2.916, p = .054) (see Table 9). Table 9: Results of the GLMM: fixed coefficients for meaning recognition with caption distinction Coefficient SE t Sig. 95% Confidence Interval Exp (Coefficient) 95% CI for Exp (Coefficient) Lower Upper Lower Upper V_Size .001 < .001 3.496 < .001< .001 < .001 .001 1.001 1.000 1.001 Caption=1 -1.359 2.063 -0.659 .510 -5.405 2.687 .257 .004 14.690 Time=1 .937 1.110 .845 .398 -1.239 3.114 2.553 .290 22.500 Time=2 -.415 1.258 -.330 .741 -2.882 2.052 .660 .056 7.782 V_Size* [Time=1] < .001 < .001 -1.572 .116 -.001 < .001 1.000 .999 1.000 V_Size* [Time=2] < .001 < .001 .542 .588 < .001 .001 1.000 1.000 1.001 [Caption=1] * [Time=1] -.065 1.562 -.042 .967 -3.129 2.999 0.937 .044 20.061 [Caption=1] * [Time=2] .630 1.701 .370 .711 -2.705 3.965 1.877 .067 52.725 V_Size * [Caption=1] *[Time=1] < .001 < .001 .851 .395 < .001 .001 1.000 1.000 1.001 V_Size * [Caption=1] *[Time=2] < .001 < .001 .162 .871 -.001 .001 1.000 .999 1.001 V_Size * [Caption=1] * [Time=3] < .001 < .001 .525 .599 -.001 .001 1.000 .999 1.001 Pairwise comparisons between experimental groups displayed significant differences between testing times exclusively for the EC ELIA 23, 2023, pp. 15-50 DOI: http://dx.doi.org/10.12795/elia.2023.i23.01 36 The effects of regular and enhanced captions... group ( p = .030 and p = .033 for prevs immediate post-test and prevs delayed post-test, respectively) (Figure 6). That is to say, participants who watched the documentary with enhanced captions significantly increased their score in both the immediate post-test and the delayed post-test and gained a significant amount of knowledge at the meaning recognition level, whereas participants in the regular captions group did not. RC’s scores did not differ significantly between any of the three time points, even if the scores did tend to increase. Figure 6: Meaning recognition by time by captions n.s. 🡒 p > .05 ** 🡒 p < .05 *** 🡒 p < .001 Form Recognition As for the form recognition test, the GLMM showed that both Vocabulary Size ( F (1,1294) = 13.075, p < .001) and the interaction between Caption and Time ( F (1,1294) = 4.707, p = .030) were statistically significant (Table 10). ELIA 23, 2023, pp. 15-50 DOI: http://dx.doi.org/10.12795/elia.2023.i23.01 37 Rebeca Finger-Bou and Carmen Muñoz Table 10: Results from the GLMM: fixed coefficients for form recognition with caption distinction Coefficient SE t Sig. 95% Confidence Interval Exp (Coefficient) 95% CI for Exp (Coefficient) Lower Upper Lower Upper V_Size .001 < .001 3.883 < .001< .001 < .001 .001 1.001 1.000 1.001 Caption=1 2.265 1.394 1.624 .105 -.470 4.999 9.626 .625 148.300 Time=2 2.401 .828 2.900 .004.004 .777 4.025 11.033 2.174 55.981 V_Size * [Time=2] < .001 < .001 -1.723 .085 -.001 < .001 1.000 .999 1.000 [Caption=1] * [Time=2] -2.660 1.226 -2.170 .030.030 -5.066 -.255 .070 .006 .775 V_Size * [Caption=1] * [Time=2] < .001 < .001 .137 .891 -.001 .001 1.000 .999 1.001 V_Size * [Caption=1] * [Time=3] < .001 < .001 -1.366 .172 -.001 .000 1.000 .999 1.000 No significant differences were found at specific testing times ( p = .537 for immediate post-test and p = .230 for delayed post-test). Both experimental groups had significant differences between immediate post- and delayed post-tests ( p = .041 for the regular group and p < .001 for the enhanced group), with the enhanced group scoring higher at the immediate post-test (EC 81.12% vs RC 77.91%) but lower than the regular group at the delayed post-test (EC 61.61% vs RC 69.94%) (Figure 7). In that sense, participants from the RC were able to retain more information than those from the EC. ELIA 23, 2023, pp. 15-50 DOI: http://dx.doi.org/10.12795/elia.2023.i23.01 38 The effects of regular and enhanced captions... Figure 7: Form recognition by time by captions n.s. 🡒 p > .05 ** 🡒 p < .05 *** 🡒 p < .001 3.3 Learners’ Awareness and Self-reported Focus of Attention3.3 Learners’ Awareness and Self-reported Focus of Attention The fourth and final research question focused on EFL learners’ selfreported focus of attention when viewing a captioned documentary. Through a series of retrospective questions, information on different levels of attention was gathered. First of all, as can be seen in Table 11, participants’ self-reported focus of attention was very similar from one experimental group to the other. In fact, a series of independent t -tests revealed that none of the differences between percentages were statistically significant ( p > .05). ELIA 23, 2023, pp. 15-50 DOI: http://dx.doi.org/10.12795/elia.2023.i23.01 39 Rebeca Finger-Bou and Carmen Muñoz Table 11: Participants’ self-reported focus of attention (out of a total of 100%) Captions (%) Audio (%) Image (%) Mean SD Min Max Mean SD Min Max Mean SD Min Max Regular (n = 15) 36.89 18.99 5 80 35.55 15.81 10 70 27.55 15.75 10 70 Enhanced (n = 16) 40.31 18.48 10 70 34.38 14.36 15 65 25.31 13.84 10 50 All participants (n = 31) 38.65 18.50 5 80 34.95 14.84 10 70 26.40 14.59 10 70 Standard deviations of the distribution are high, which indicates that the data is more spread out, or, in other words, that the mean is not that reliable. In general, participants reported to focus more on captions, followed by the audio and the image. Even though the tendency of the EC group is to focus more on captions, probably due to the enhancement of TWs, the difference with the RC group is non-significant. Results showed that the tendency of the RC group, but not of the EC group, was to focus more on the image and the audio . Secondly, as for participants’ self-reported linguistic focus of attention displayed in Table 12, differences among experimental groups were, once again, found non-significant ( p > .05), so participants’ distribution of percentages were statistically similar. The common ranking for all linguistic features in both experimental groups was: general comprehension, new vocabulary, pronunciation, expressions and intonation, in that order. ELIA 23, 2023, pp. 15-50 DOI: http://dx.doi.org/10.12795/elia.2023.i23.01 46 The effects of regular and enhanced captions... and has not found significant differences between experimental groups, although there were significant within-group differences, highlighting the relevance of out-of-classroom exposure to L2 media. Furthermore, this work has also reinforced the importance of individual differences, confirming once again the significance of vocabulary size when learning single-word items, while also studying the non-significant contribution of the LLAMA tests, which has not been widely examined in the context of caption enhancement. Finally, through the retrospective questionnaire, this study has been able to describe participants’ thoughts, perceptions and ideas about the experiment in general, and about the enhancement of captions in particular. The findings reported in this paper should be considered in the light of some limitations. Firstly, this study has not accounted for either a no-captions group or a control group who would not have viewed the documentary. Although the main aim of the study was to compare the two conditions (regular vs enhanced captions), the current findings need to be confirmed by further research including a control group. Secondly, due to the pandemic, until the final session of the experiment, half of the participants conducted the tasks online. Even though this study tried to control for cheating, and some of the websites used did not allow participants to change tabs during tests, there is an extent to which fraud cannot be fully disregarded. Another limitation of the study was that due to practical reasons the number of participants is relatively small. Further research with a larger sample size can provide stronger evidence. Next, this work could have included frequency of occurrence in the different analyses and examine its relationship to learning outcomes, previous vocabulary knowledge and language aptitude, rather than merely focusing on a frequency threshold (Uchihara et al. 2019). Further research (in progress) using the eye-tracking methodology can juxtapose students’ self-reported focus of attention against more objective data to analyse the extent to which students are aware of the way in which captions attract their attention. ELIA 23, 2023, pp. 15-50 DOI: http://dx.doi.org/10.12795/elia.2023.i23.01 47 Rebeca Finger-Bou and Carmen Muñoz This study has some pedagogical implications as well. This work has demonstrated the potential advantage of multimodal input for acquiring languages. In that sense, language teachers could provide students with effective and motivating language learning experiences as some of the participants from this experiment found it “fun,” “good,” and “enjoyable.” Altering Dr Karan Rangarajan’s words from the documentary, “spread knowledge, not the virus” (Livingston, 2020, 00:17:45–00:17:47), we would like to conclude this study by asserting that we can spread culture, knowledge and languages through multimodal input, so go ahead and spread the word! ReferencesReferences Feng, Y., & Webb, S. (2019). Learning Vocabulary Through Reading, Listening, and Viewing: Which Mode of Input is Most Effective? Studies in Second Language Acquisition , 42 (3), 499–523. https://doi.org/10.1017/ S0272263119000494 Gesa, F. (2019). L1 / L2 subtitled TV series and EFL learning: A study on vocabulary acquisition and content comprehension at different proficiency levels. (Doctoral dissertation). University of Barcelona. Lee, M., & Révész, A. (2020). Promoting grammatical development through captions and textual enhancement in multimodal input-based tasks. Studies in Second Language Acquisition , 42 (3), 625–651. https://doi. org/10.1017/S0272263120000108 Livingston, H. (Director). (2020). Viral: The 5G Conspiracy Theory . British Broadcasting Corporation (BBC). Majuddin, E. (2020). Incidental and intentional acquisition of multiword expressions from audio-visual input: The effects of typographically enhanced captions and repetition (Doctoral dissertation). Victoria University of Wellington. Majuddin, E., Siyanova-Chanturia, A., & Boers, F. (2021). Incidental acquisition of multiword expressions through audiovisual materials. ELIA 23, 2023, pp. 15-50 DOI: http://dx.doi.org/10.12795/elia.2023.i23.01 48 The effects of regular and enhanced captions... Studies in Second Language Acquisition, 43(5), 985–1008. https://doi. org/10.1017/S0272263121000036 Mayer, R. E. (2014). Introduction to multimedia learning. In R. Mayer (Ed.), The Cambridge handbook of multimedia learning (pp. 1–24). Cambridge University Press. https://doi.org/10.1017/CBO9781139547369.002 Meara, P. M., & Miralpeix, I. (2015). V_YesNo (v1.01) . Cardiff: Lognostics. Meara, P. M., & Rogers, V. E. (2019). The LLAMA Tests v3. LLAMA_B3 (v3.00) . Cardiff: Lognostics. Meara, P. M., & Rogers, V. E. (2019). The LLAMA Tests v3. LLAMA_D3 (v3.00) . Cardiff: Lognostics. Montero Perez, M. (2022). Second or foreign language learning through watching audio-visual input and the role of on-screen text. Language Teaching , 55 (2), 163-192. https://doi.org/10.1017/S0261444821000501 Montero Perez, M., Peters, E., Clarebout, G., & Desmet, P. (2014). Effects of Captioning on Video Comprehension and Incidental Vocabulary Learning. Language Learning & Technology , 18 (1), 118–141. Montero Perez, M., Peters, E., & Desmet, P. (2015). Enhancing vocabulary learning through captioned video: An eye-tracking study. The Modern Language Journal , 99 , 308–328. https://doi.org/10.1111/modl.12215 Montero Perez, M., Van Den Noortgate, W., & Desmet, P. (2013). Captioned video for L2 listening and vocabulary learning: A meta-analysis. System, 41, 720–739. https://doi.org/10.1016/j.system.2013.07.013 Muñoz, C. (2022). Audiovisual input in L2 learning. Language, Interaction and Acquisition . 13(1): 125-143. https://doi.org/10.1075/lia.22001.mun Nation, P. (2001). Designing the vocabulary component of a language course. In P. Nation, Learning Vocabulary in Another Language (pp. 380–406). Cambridge: CUP. https://doi.org/10.1017/CBO9781139524759.013 Paivio, A. (1986). Mental representations: A dual coding approac h. New York: Oxford University Press. Pattemore, A., & Muñoz, C. (2020). Learning L2 constructions from captioned audio-visual exposure: The effect of learner-related factors. System , 93 , 1–13. https://doi.org/10.1016/j.system.2020.102303 ELIA 23, 2023, pp. 15-50 DOI: http://dx.doi.org/10.12795/elia.2023.i23.01 49 Rebeca Finger-Bou and Carmen Muñoz Pattemore, A., & Muñoz, C. (2022). Captions and learnability factors in learning grammar from audio-visual input. The JALT CALL Journal , (18), pp. 83-109. http://doi.org/10.29140/jaltcall.v18n1.564 Peters, E., & Muñoz, C. (2020). Language Learning from Multimodal Input. Studies in Second Language Acquisition , 42, 489–497. https://doi. org/10.1017/S0272263120000212 Peters, E., & Webb, S. (2018). Incidental Vocabulary Acquisition through Viewing L2 Television and Factors that Affect Learning. Studies in Second Language Acquisition , 40 (3), 551–577. https://doi.org/10.1017/ S0272263117000407 Puimège, E., Montero Perez, M., & Peters, E. (2022). Promoting L2 acquisition of multiword units through textually enhanced audiovisual input: an eye-tracking study. SECOND LANGUAGE RESEARCH . https://doi. org/10.1177/02676583211049741 Pujadas, G. (2019 ). Language learning through extensive TV viewing. A study with adolescent EFL learners . (Doctoral dissertation). University of Barcelona. Pujadas, G., & Muñoz, C. (2019). Extensive viewing of captioned and subtitled TV series: a study of L2 vocabulary learning by adolescents. The Language Learning Journal , 47 (4), 479–496. https://doi.org/10.1080/09571 736.2019.1616806 Pujadas, G., & Muñoz, C. (2020). Examining adolescent EFL learners’ TV viewing comprehension through captions and subtitles. Studies in Second Language Acquisition, 42 (3), 1–25. https://doi.org/10.1017/ S0272263120000042 Rodgers, M. P. H., & Webb, S. (2011). Narrow viewing: The vocabulary in related television programs. TESOL Quarterly , 45 (4), 689–717. https://doi. org/10.5054/tq.2011.268062 Rodgers, M. P. H., & Webb, S. (2019). Incidental vocabulary learning through watching television. ITL—International Journal of Applied Linguistics , 171 (2), 191–220. https://doi.org/10.1075/itl.18034.rod Rogers, V., Meara, P., Barnett-Legh, T., Curry, C., & Davie, E. (2017). Examining the LLAMA aptitude tests. Journal of the European Second Language Association , 1 (1), 49–60. https://doi.org/10.22599/jesla.24 ELIA 23, 2023, pp. 15-50 DOI: http://dx.doi.org/10.12795/elia.2023.i23.01 50 The effects of regular and enhanced captions... Schmidt, R. (1994). Deconstructing consciousness in search of useful definitions for applied linguistics. AILA Review, 11 , 11–26. Suárez, M. D. M., & Gesa, F. (2019). Learning vocabulary with the support of sustained exposure to captioned video: do proficiency and aptitude make a difference? The Language Learning Journal , 47 (4), 497–517. https://doi.org/10.1080/09571736.2019.1617768 Uchihara, T., Webb, S., & Yanagisawa, A. (2019). The Effects of Repetition on Incidental Vocabulary Learning: A Meta-Analysis of Correlational Studies. Language Learning , 69 (3), 559–599. https://doi.org/10.1111/ lang.12343 Winke, P. M. (2013). The Effects of Input Enhancement on Grammar Learning and Comprehension. Studies in Second Language Acquisition , 35 , 323–352. https://doi.org/10.1017/S0272263112000903 First version received: June, 2023 Final version accepted: September, 2023