Foreign language anxiety and self-disclosure analysis as personality traits for online synchronous intercultural exchange practice

Videoconferences are a perfect scenario for autonomous Foreign Language (FL) and intercultural speaking practices. However, it is also a threatening context as learners communicate in an FL, often with a stranger and about personal information and experiences. That may lead to increase Foreign Language Anxiety (FLA) among participants, affecting students’ learning experiences and even provoking drop-outs (Bailey, Onwuegbuzie, & Daley, 2003). This study aims to explore the relationship between FLA and Self-Disclosure (SD) as personality traits. The first indicators of the potential relationship between FLA and SD in online speaking practice were found by Fondo and Erdocia (2018) in which anxious learners showed a tendency to self-disclose as a means to manage their discomfort using the FL. Data was gathered in the first stage of a nine-week-synchronous oral Online Intercultural Exchange (OIE) project between undergraduate business students from the United States, Ireland, Mexico, and Spain.


Introduction
In the last three decades, the affective dimension of the FL learning process has taken prominence in the field of second language acquisition.In this regard, FLA, or the situational and contextual anxiety experienced by language learners, has been one of the most studied learning barriers since Horwtiz, Hortwitz, and Cope coined the term in 1986.Approximately half of the students suffer from FLA when using the FL (Atas, 2015), particularly when speaking.Communication in the FL, as in the L1, involves private information exchange, also known as SD.Indeed, many FL teaching practices are based on inviting the students to share personal information.To our knowledge, no study has looked into this private information sharing effect on students' FLA.The present article analyses student profiles of FLA and their relation to SD, considering both constructs as personality traits in order to understand the role played by both constructs and their effects on OIE practices.

Method
The project was a nine-week OIE through videoconference for Spanish/English speaking practice and intercultural skills development among Business students.
Five universities joined the one-on-one project: Universitat Oberta de Catalunya in Spain, University of Maryland and University of Minnesota from the United States, University of Limerick in Ireland, and Benemérita Universidad Autónoma de Puebla (BUAP) from Mexico.A parallel monolingual intercultural project was run between students from Queens College New York and BUAP.As this project was fully in English, we did not survey New York students for FLA so only the students from Mexico (Spanish speakers) have been included in this study.
One hundred and fourteen students from the five universities mentioned above participated in the project.Every student was provided with a participant code at the beginning of the project in order to assure their anonymity.Participants were paired according to their time compatibility, and had to complete five online sessions in total.
The pre-project questionnaire, which provides the data for this article, was delivered online in the first stage of the project.The questionnaire measures participants' personal levels of FLA using the widely known Foreign Language Classroom Anxiety Scale (FLCAS) (Horwitz, Horwitz, & Cope, 1986) and the FLCAS Spanish version (Pérez-Paredes & Martínez-Sánchez, 2000).In order to assure the reliability of the results, we provided students with a version in their L1: Spanish and English.SD as a personality trait is measured using the General Disclosiveness Scale (GDS) by Wheeless (1978).For this study we translated the GDS into Spanish and validated the translation in dialogue with other experts -to our knowledge this is the first time the GDS has been translated and applied in Spanish.A total of 87 cases were used in data analysis.

Results
We performed a Principal Component Analysis (PCA) of the FLCAS for dimension reduction.This analysis used the varimax rotation and resulted in three components.We named the components as Negative attitudes towards FL learning (Neg.att.), FL classroom participation anxiety (Class.anx.), and Confidence in using the FL (Conf.).
We then performed the k-means clustering technique to the PCA score to create FLA profiles, resulting in a total of six clusters (Table 1).The Anxious partakers profile also contains a similar proportion of the sample, (very low Neg.att.but the highest level of Class.anx.).Finally, the last identified profile is the Insecure, with the lowest levels of Conf.and moderately low levels of Neg.att.
A PCA was not performed on the GDS as its dimensions were established by Wheeless (1978) -in contrast with the FLCAS, which does not have defined dimensions.Wheeless (1978) identified five dimensions of SD: Intent, Amount, Positiveness, Depth, and Honesty.The learners' FLA profiles and SD levels were analysed in their different dimensions (Table 2).
Surprisingly, the Challenged group presents medium levels of SD Amount and high levels of Depth, thus not very different from the Achievers.However, these profiles are opposed regarding Intent, Positiveness and Honesty: Achievers rank high in these dimensions and Challenged low.
The two most common profiles (Reluctant learners and Confidents) show similar levels of SD in all dimensions (high Intent, Positiveness, Depth and Honesty but low Amount).The Insecure profile presents the lowest SD levels in every dimension.

Discussion
This study sheds light on how FLA and SD interact.The fact that Challenged and Insecure profiles have lower levels in SD dimensions point to a connection between low confidence in using the FL and SD, particularly Intent, Honesty (unobservable in interaction), and Positiveness.
In addition, Anxious partakers and Challenged show relatively high tendency to talk about themselves (Amount and Depth), as found in the above mentioned case-study (Fondo & Erdocia, 2018), which are observable in speaking practices and could be misinterpreted as a lack of anxiety as in the case of the Achievers profile.

Conclusion
The study shows that some characteristics and events that influence interactions in speaking practices are not easily observable without a previous personality trait analysis.FLA and SD have shown to be related at an individual level.However, this first stage of personality trait analysis does not clarify yet if self-disclosing is a strategy used by students for FLA reduction nor the role these profiles play in the context of OIE.The next stages of analysis will try to answer this question, diving into the qualitative and quantitative data collected during the project in order to understand, from an ecolinguistic approach, how these profiles affect interaction and how students perceive it.Students, environment, and outcomes will be explored individually and in interaction in order to fully understand how SD and FLA are related during speaking practices.Our aim is to assess the benefits and/or risks of SD in relation to FLA to inform decision-making by the FL teaching community about learning scenarios and situations.

Table 1 .
The two most common FLA profiles seem to us, intuitively, to represent the typical profiles of FL students, each gathering about 25% of the sample each.The first, Reluctant learners, show very low Class.anx.but moderate levels of Neg.att.and Conf.In turn, Confidents have the highest level of Conf.but moderate Class.anx.The confidence in FL use together with moderate negative attitude towards learning seems to us a typical trait in FL students.FL student profiles (clusters) and FLA -general and by dimension (averages of scale items and PCA scores) Two categories seem to be related to extremes in FLA.On the least anxious extreme are those we called Achievers, characterised by the lowest levels of general FLA, Neg.att., and Class.anx., as well as moderate Conf.The moderate confidence level of the Achievers may be related to a more engaged and knowledgeable FL student profile, resulting in the reduction of overconfidence.On the other side of the FLA spectrum, we identified the Challenged profile (highest level of general FLA, Neg.att., moderate Class.anx., and very low Conf.).As expected, the proportion of learners in those categories is marginal (each gathering about 10% of the sample).

Table 2 .
FLCAS clusters and SD levels (average of Likert scales from 1 to 7) by SD dimensions