The Relationship between Language Educators’ Perceptions and Assessment Practices during the COVID-19 Crisis

Abstract

This study investigated language educators’ readiness in coping with language assessment during the global outbreak of the COVID-19 pandemic and the enforced move to Emergency Remote Teaching (ERT). This pandemic prompted debates on language assessors’ roles in adjusting to a New Normal. While a good body of research has investigated the role of teacher assessment perceptions versus assessment behavior for many decades, little is known about the factors that may have impacted language educators’ assessment perceptions and practices during the recent crisis. To address this issue, an online survey was administered to 256 language educators. Pearson correlations and simple linear regression were utilized to determine if the language educators’ perceptions of (1) the official assessment measures, (2) purposes, and (3) their assessment self-efficacy were predictors of their assessment practices during this crisis. The results revealed a total absence of any correlations between these variables. The findings suggest that the assessment accommodations adopted by the teachers were not determined by their assessment perceptions. Other factors such as assessment policy and the assessment culture may have shaped their practices during this crisis.

1. Introduction

The global outbreak of the COVID-19 pandemic has dramatically changed the educational landscape around the world, affecting the way language educators deal with assessment. Educational stakeholders at all educational levels have encountered numerous challenges in ensuring pedagogical continuity in their educational systems, including the way they plan and implement assessment (Ghanbari and Nowroozi 2021). Among the challenges encountered were the difficult planning and administration of end-of-year assessments during this long-lasting crisis (UNESCO 2020a; Clark et al. 2020; Nădrag 2020). These unprecedented conditions prompted debates in various educational disciplines on ways of coping with the “New Normal” (d’Orville 2020; Pacheco 2020) for teachers and other education stakeholders worldwide. The New Normal “forces education further into technologization, a development already well underway” (Pacheco 2020, p. 5), leading to a rethinking of language educators’ preparedness (Gouëdard et al. 2020; Hodges et al. 2020; Hodges and Barbour 2021; Paramour 2020) in terms of their perceptions and their possible impact on their language assessment practices when coping with ERT (Emergency Remote Teaching). The latter refers to the sudden move to remote education because of the distortion of onsite schooling. This has forced many educators worldwide to organize and implement assessments differently (Ockey 2021; Rahim 2020; Reimers and Schleicher 2020; Zhang et al. 2020) while maintaining quality and professionalism (Muhammad and Ockey 2021).

2. Background

Today’s global educational landscape under the COVID-19 crisis calls upon language teachers to use their skills as assessors perhaps more than ever. Research has emphasized teachers’ roles in implementing assessment policy while being at “the frontline as designers and users of language assessments” (Kremmel and Harding 2020, p. 101). The turmoil caused by school closure during the pandemic has initiated a debate on teachers’ role in planning and implementing education (Anderson 2020; Lightfoot 2020; Rahim 2020; UNESCO 2020a, 2020b, 2020c) and language assessment (British Council 2020; Gao and Zhang 2020) worldwide. A study conducted by the British Council (2020) with language teachers and teacher educators in different countries has shown that the forced and unplanned shift to online teaching may be particularly challenging for teachers as they need to develop their skills in effective online assessment. Unlike previous educational emergency research (Salerni and Vaccarelli 2019; MacNeil and Topping 2007) concerned with the question of resilience and “psychological, social, and community dimensions’’ (Vaccarelli 2018, p. 29), the main challenge during the COVID-19 crisis has been to take urgent measures regarding both teaching and assessment during the lockdown (Anderson 2020). The current crisis has had implications mainly on high-accountability systems in the USA (US Department of Education 2020), on high-stakes exams such as the GCSE and the A-level in the UK (Lightfoot 2020; Miller 2020; UK Government 2020), and on end-of-year assessments in the countries affected by the pandemic. The main challenge decision-makers faced was related to the management of scheduled assessments, including high-stakes ones (d’Orville 2020; UNESCO 2020c, 2020d, 2020e, 2020f).

The post-COVID research has raised issues about teachers’ “pedagogical readiness” (Rapanta et al. 2020; Tuah and Naing 2021) in coping with assessment challenges under ERT (Hodges et al. 2020). Limited education (Rahim 2020; Reimers and Schleicher 2020) but almost no language assessment research has pointed to the need for proactive measures to prepare teachers for more principled and effective assessment measures. The unprepared shift from onsite to remote education during ERT was the only possible measure to reduce the destabilizing effect of the COVID-19 crisis on schooling. Educators had to adapt to an online mode for their teaching and assessment alike (Rahim 2020; Reimers and Schleicher 2020). Nevertheless, the lack of theory-based and planned solutions regarding assessment (Iglesias-Pradas et al. 2021) has compelled language educators to implement responsive measures (British Council 2020; Dill et al. 2020; US Department of Education 2020; Paesani 2020; Radcliffe et al. 2020; UNESCO 2020a; Whittle et al. 2020; Woolcock 2020) and assessment accommodations (Marinoni et al. 2020; Reimers and Schleicher 2020; UNESCO 2020c) as an adaptation to this novel situation. Traditionally, assessment accommodations (Abedi 2013) concern changes to meet learners’ special needs (Abedi et al. 2000; Pennock-Roman and Rivera 2011; Taylor 2012; Tsagari and Sperling 2017). The concept of assessment accommodations is used in this study to refer to a strategy, change, or adjustment to the new situation by canceling, postponing, or rescheduling planned exams, moving to online assessment, or opting for alternative modes of assessments (UNESCO 2020a).

Of relevance to the question of teacher preparedness (Anderson 2020; Gouëdard et al. 2020) is teachers’ assessment perceptions (Mäkipää et al. 2021; Rapanta et al. 2020). The ERT context requires the consideration of assessors’ perceptions as a possible determining factor of their behavior from a “teacher as assessor” perspective (Looney et al. 2017). While a plethora of studies has attended to the role of teacher perceptions in determining assessment behavior for many decades (Brown 2004; Scarino 2013), limited post-COVID exploratory research has emerged to bring to light language teachers’ perceptions of online assessment, the assessment methods utilized (Abduh 2021), and the difficulties encountered during the lockdown (Abduh 2021; Ghanbari and Nowroozi 2021). Prior to the current crisis, Zulaiha et al.’s (2020) study revealed a mismatch between English as a foreign language teachers’ assessment perceptions and practices. In the context of ERT, Rapanta et al. (2020) examined teachers’ perceptions of the effective monitoring of students’ learning online and its implications for assessment. Their findings confirm the teachers’ need for more adaptation time and effort for more effective and stress-free assessment practices. In their report, Reimers and Schleicher (2020) reveal compelling findings regarding teachers’ views of the official decisions during the COVID-19 crisis. This study aims at examining language educators’ perceptions of the responsive measures taken and the assessment purposes during the crisis. Such purposes are based on an adapted taxonomy of Brown (2004), including the two main purposes of (a) improving learning through feedback and (b) school accountability.

Among the key aspects of educator perceptions is self-efficacy as a developmental construct (Bandura 1994), which could be a key predictor of educators’ pedagogical practice (e.g., Bandura 1997; Myyry et al. 2021; Poulou et al. 2019). Despite its recognition as a personal construct in educational assessment (Brown 2004; Xu and Brown 2016), limited research has addressed it in the language assessment literature (Levy-Vered and Nasser-Abu Alhija 2015). According to social-cognitive theory, four main factors can shape people’s perceptions of their own efficacy. “Mastery experiences”, including certification, professional development, and years of experience, are the most effective way of developing a strong sense of efficacy (Bandura 1994). Assessment self-efficacy (Bandura 1994; Kruse et al. 2020; Levy-Vered and Nasser-Abu Alhija 2015) is operationally defined for this study as ‘assessors’ personal beliefs about their confidence in planning and performing language assessment tasks and assessment accommodations during the crisis’. Research has shown that assessment self-efficacy can be shaped by assessment literacy and experience (Levy-Vered and Nasser-Abu Alhija 2015). In the context of a crisis such as the current one, teachers’ reported self-efficacy could be associated with the efforts they would deploy “on an activity, how long they will persevere when confronting obstacles, and how resilient they will prove in the face of adverse situations” (Pajares 1996, p. 544). Exceptionally, studies employing quantitative analyses including correlation (Mueller et al. 2008) sometimes failed to reveal any positive relationship between teacher self-efficacy and technology use as an example of classroom practice.

Despite the overall consensus on the significance of teacher factors in determining their practice, little research has focused on such issues during the crisis. Studies accounting for teachers’ roles (Cahapay and Anoba 2021; Ghanbari and Nowroozi 2021), cognition (Gao and Zhang 2020), or teachers’ formative assessment engagement in English as a foreign language settings (Zou et al. 2021) under the crisis have been rather scant. A study using a duoethnographic approach has explored three English language teachers’ assessment literacy (i.e., assessment knowledge and skills) before and after COVID-19 (Tian et al. 2021). The teachers who sought solutions to the post-COVID assessment issues reported an expanded sense of self-questioning, leading to a felt need to develop their assessment literacy. Studies by Ghanbari and Nowroozi (2021) and Watermeyer et al. (2021) have also explored language teachers’ ability and preparedness to deal with the demands of ERT. Jelińska and Paradowski’s (2021) study examined teachers’ engagement during the transition to ERT. Watermeyer et al.’s (2021) survey of teachers’ reactions to ERT provided evidence of these educators’ strong confidence in their ability as both teachers and assessors online. Post-COVID research on language teachers’ experience during ERT has shown that teachers with no prior online teaching experience exerted personal efforts to be capable of teaching remotely (Hodges et al. 2020; Juárez-Díaz and Perales 2021). Juárez-Díaz and Perales (2021) alluded to the teachers’ need to adjust to the assessment conditions of ERT. Even though language teachers from 150 countries demonstrated moderately high confidence in their ability to teach online, some teachers overtly reported that they were rather overwhelmed by the sudden transition to ERT (British Council 2020).

On the basis of the above discussions, this paper will then try to test the following hypotheses:

Hypothesis 1 (H1).

There is a significant relationship between the language educators’ perceptions of the assessment measures mandated by the authorities of their countries and their assessment practices during the crisis.

Hypothesis 2 (H2).

There is a significant relationship between the language educators’ perceptions of assessment purposes during the crisis and their own assessment practices.

Hypothesis 3 (H3).

There is a significant relationship between the language educators’ self-efficacy and their reported post-COVID assessment practices.

3. Materials and Methods

3.1. Research Site and Participants

The study data were collected based on a combination of both purposeful and snowball sampling techniques, with the sample not necessarily representing the larger population of language educators worldwide. A cross-sectional survey questionnaire was developed, piloted, and repiloted to assess the instrument’s internal reliability and validity among the teacher network of the current authors. To reduce the participants’ reluctance and aversion to respond to the survey, a Google Form of the questionnaire was created and disseminated by email or uploaded on the authors’ professional and personal networks during the lockdown from May to July 2020. The researchers managed to recruit a total of 256 participants through a non-probability sampling technique.

The participants’ selection can be justified by their shared “specialized knowledge” (Ruel et al. 2016). Consequently, no generalizable claims can be made. These participants were from four main regions. They were all language educators, 88.7% of whom were language teachers from secondary and higher education, and 11.3% were language educators including teacher trainers and test developers. While 38.4% were from the Middle East and North Africa (MENA) region, 35.1% were from Europe; the remaining participants were from America and other countries around the globe. It is also worth mentioning that 41% had never taught online before the crisis. According to Figure 1, the participants also reported that 40.63% of them “never” and 34.77% “sometimes” taught online prior to the crisis. Question 4 in Section B also revealed that the majority (76.6%) of the respondents used online assessment during ERT (see Figure 1 below).

3.2. Data Collection

To understand language educators’ assessment self-efficacy, it was also necessary to examine their assessment background. The results in

Table 1. The educators’ assessment background.

	Frequency	Relative Frequency
Pre-service training	113	0.863
In-service training	147	0.773
Professional conferences, workshops, seminars, etc.	198	0.667
Experience on the job	221	0.617
Μy colleagues/institution	158	0.574
Online resources such as websites, courses, webinars, etc.	171	0.535
Doing research in this area	137	0.441
Other	20	0.078

below are based on the analysis of the responses to Question 7 of our survey (see Appendix A). They indicate the relative frequency values of the participants’ assessment literacy source. It is clear from the results that the majority (0.863) of the participants developed their assessment literacy mainly thanks to pre-service training, while in-service training occupied the second most important professional assessment development activity (0.773). This was followed by professional development events such as conferences, workshops, and seminars (0.667). The fourth source of assessment literacy development for these language educators was experience on the job (0.617), followed by online resources such as websites, courses, and webinars (0.61). Of equal importance to the development of the participants’ assessment literacy were doing research in assessment (0.49) and in-service training (0.52). Their assessment literacy also developed thanks to collaborating with colleagues and the institution (0.57) in which the participants work.

3.3. The Survey

This study was motivated by COVID-19. It examined three types of associations between the three independent variables of (1) language educators’ perceptions of the assessment measures, (2) their perceptions of assessment purposes, (3) their assessment self-efficacy, and one dependent variable, (4) the variable of language assessment practices during the crisis. The language educators’ questionnaire consisted of four sections (see Appendix A). The items were designed based on the hypotheses guiding this study. Section A was devoted to reporting on the measures taken in the participants’ contexts regarding language assessment under COVID-19. It included one main question allowing the respondents to choose as many options as needed. Section B elicited information on the educators’ assessment practices during the COVID-19 crisis. Question 2, in particular, concerned the assessment accommodations they adopted during the crisis using multiple-choice items, while Question 3 focused on the assessment tasks they used. Question 4 aimed to understand whether these educators conducted online assessments during the crisis using a three-point scale of “yes”, “no”, and “not yet”. Section C, using an agreement Likert scale of five points, elicited information regarding the participants’ perceptions of the language assessment measures taken in their contexts (Question 5), their perceptions of assessment purposes (Question 6), and their perceived assessment self-efficacy (Question 7). The last section collected data on their profiles, including their assessment background (Questions 8 to 17).

3.4. Data Analysis

The data analysis was mainly quantitative. The internal reliability of the survey items comprising Likert scales was examined using Cronbach’s alpha. The findings indicated that the alpha for the total scale was equal to 0.7 and hence deemed acceptable. All statistical analyses were carried out using the Statistical Package for the Social Sciences (SPSS) version 23. In addition to percentages to describe the educational contexts in terms of the assessment measures taken and the participants’ profiles, Pearson correlations and simple linear regression were employed to determine if the participants’ factors of perceptions predicted their assessment practices during this crisis.

A principal components analysis was conducted on the 10 items of the perceptions of language assessment measures during the COVID-19 crisis to determine whether they represent a single construct. An exploratory factor analysis was performed employing a principal component analysis and varimax rotation. In factor extraction, three main criteria were used: (1) a minimum of three items in one factor with an eigenvalue of one or greater; (2) factor loadings less than 0.4 were deleted and not counted in any factor; and (3) items with double loadings were removed. This analysis yielded one factor with an eigenvalue of 2.879, accounting for 47.981% of the total variance for this factor. All items loaded higher than 0.60 on the factor. Factor analysis results indicated that six survey questions clustered together around teachers’ perceptions of the measures taken about high-stakes exams with a Cronbach alpha of 0.767. This indicates that the construct is valid for further analysis. The KMO measure (Kaiser–Meyer–Olkin; Kaiser 1970) of sampling adequacy is 0.756.

For Question 5 regarding these language educators’ perceptions of assessment purposes during the crisis, a principal components analysis was performed on the six items to determine whether they represent a single construct. The analysis yielded one factor with an eigenvalue of 2.153, accounting for 53.833% of the total variance with a Cronbach alpha of 0.708, indicating that the construct is valid for further analysis. The KMO measure of sampling adequacy is 0.683.

Regarding the participants’ perceived assessment self-efficacy, a principal components analysis was equally conducted on the nine items to determine whether they represent a single construct. This analysis yielded two factors, with an eigenvalue of 4.802 for Factor 1, accounting for 53.36% of the total variance and an eigenvalue of 1.129, and a total variance of 12.545 for Factor 2 and a Cronbach alpha of 0.881. The factor analysis results indicated that four survey questions clustered together around the language educators’ self-efficacy. The KMO measure of sampling adequacy is 0.888.

For the last variable of the post-COVID assessment practices, the 27 items received a principal components analysis and resulted in seven factors with eigenvalues of 6.497, 3.031, 2.350, 2.194, 1.942, 1.741, and 1.439 for Factors 1, 2, 3, 4, 5, 6, and 7, respectively. The KMO measure of sampling adequacy is 0.730, and has Cronbach alpha of 0.692. The seven factors were calculated for further analysis.

4. Results

4.1. The Assessment Context in the Different Countries

The statistical analysis below serves first to describe the assessment landscape immediately after the outbreak of the COVID-19 crisis. As displayed in

Table 2. Relative frequency of the responsive assessment measures during the crisis.

Assessment Measures	Frequency	Relative Frequency
Organizing exam(s) online	132	0.515
Postponing the dates of the exam(s) for a later date	93	0.363
Replacing exams with alternative assessments (e.g., projects, essays, etc.)	90	0.351
Reducing the subject content	73	0.285
Offering exam(s) as before	66	0.257
Using written take-home exams	60	0.234
Offering part of the exam(s), e.g., only oral, or only written	57	0.222
Using plagiarism-detection services (e.g., Turnitin)	50	0.195
Reducing the number of subjects tested	42	0.164
Cancelling exam(s) altogether	26	0.101
Using an online proctoring service (=online surveillance, e.g., Proctorio, ProctorU, Examity, etc.)	24	0.093
Using test results from previous semesters/years	24	0.093
Using teacher assessments from previous semesters/years	21	0.082
Conditional admission to the university, complemented by remedial courses upon school reopening	5	0.019
Other	0	0

(see below), the official responsive measures due to school closure indicate that there was a partial shift to organizing exams online with a relative frequency of 0.515 in the participants’ countries. In their answers to the first survey question, the educators reported that the crisis led to postponing exams for a later date (0.363), replacing exams with alternative assessments such as projects and essays (0.351), reducing the subject content (0.285), offering exam(s) as before (0.257), using written take-home exams (0.234), offering part of the exams by opting for either an oral or written form (0.222), using a plagiarism detector for online exams (0.195), reducing the number of subjects tested (0.164), canceling exam(s) altogether in only a few contexts (0.101), and ‘Using an online proctoring service’ (=online surveillance, e.g., Proctorio, ProctorU, Examity, etc.) and ‘Using test results from previous semesters/years’ (0.093). However, ‘Using teacher assessments from previous semesters/years’ represented only 0.082.

4.2. Perceptions of Assessment Measures during the Pandemic

The results in

Table 3. Teachers’ perceptions of assessment measures under COVID-19.

	SD/D *	Neutral	SA/A **
Canceling exams is unacceptable	34%	20.7%	45.3%
Exams are important to learners	28.5%	17%	54.5%
Exams are important for society	35.5%	22.3%	42.2%
Only exams should be maintained for assessment	72%	12%	16%
Exams are important to me as an educator/expert	19.2%	13.3%	67.5%
My country is managing the assessment situation in an effective way	28%	29.3%	41.8%

* SD/D: Strongly Disagree/Disagree, ** SA/A: Strongly Agree/Agree.

below summarize these language educators’ perceptions of the responsive assessment measures under COVID-19. When asked if their countries were managing the assessment situation effectively, the participants were either neutral (29.3%) or disagreed with the statement (28%). In addition, about half of them (45.3%) refused the decision of exam cancellation despite the near-total closures of educational institutions. About half of these educators also believe that exams should be maintained because of their importance to learners (54.5%) and society (42.2%). However, 72% disagree with the decision to maintain only exams. It is important to highlight that 67.5% are against exam cancellation because of the importance they represent to them as language teaching and testing stakeholders.

Table 4. Teachers’ perceptions of assessment purposes during the crisis.

	SD/D	Neutral	SA/A
To create opportunities to provide feedback	2.8%	7.4	95.8%
Because assessment is part of the teaching/learning cycle	4.8%	15.6%	89.8%
Because assessment is mandated in learning curricula	5.8%	15.6%	79.6%
Because of decision making, e.g., progress to the next level, exit requirement, etc.	8.2%	16%	75.8%

displays the participants’ opinions concerning assessment purposes during the crisis. While the majority of opinions (95.8%) reflected an awareness of the importance of assessment during a crisis to create opportunities for feedback, 89.8% agree with the need for assessment because it is part of the teaching and learning cycle. A total of 79.6% reported that they would use assessment during a crisis because it is mandated in learning curricula, and 75.8% think that assessment is needed for decision making, including student progress to the next level or for exit requirements.

4.3. The Language Educators’ Assessment Self-Efficacy

One of the study goals was to understand the role assessment self-efficacy might have played in the study of participants’ assessment accommodations. Thus, the participants were invited to describe their self-efficacy as assessors in Question 5. As displayed in

Table 5. The educators’ self-efficacy about their roles as assessors.

	Designing Test Questions	Designing Scoring Criteria	Scoring Written Papers	Scoring Oral Performance	Giving Students Feedback Based on Assessment Results	Using Self- or Peer Assessment
Not at all confident	2.0%	4.3%	0.4%	0.4%	2.0%	1.6%
Not too confident	8.2%	9.8%	1.6%	2.7%	4.3%	5.9%
Somewhat confident	30.1%	39.1%	6.6%	9.8%	27.7%	10.5%
Moderately Confident	59.8%	46.9%	37.1%	31.6%	66.0%	41.4%
Very confident	2.0%	4.3%	54.3%	55.5%	2.0%	40.6%

, the results show that the educators generally reported having moderate to high self-confidence in their roles as assessors. Around half of them were moderately confident (59.8%) in their ability to design test questions and scoring criteria (46.9%) and give feedback (66%). Their assessment self-efficacy was somewhat higher for scoring written papers, with 37.1% who were moderately confident and 54.3% who were very confident. However, they were equally moderately (41.4%) confident and highly confident (40.6%) in their ability to assess their students via self- or peer assessment.

4.4. The Impact of the Perceptions of the Measures on Language Assessment Practices

This section aims to report whether there is a significant relationship between the language educators’ perceptions of the assessment measures taken by their countries and their assessment practices during the crisis. First, correlation analyses were used to examine this relationship and yielded a correlation of −0.118. Next, simple linear regression was used to test if language educators’ perception of the assessment measures significantly predicted their assessment practices. The fitted regression model was as follows: assessment practices = 81.586 + 1.148 (perception of the assessment measures). The overall regression was statistically insignificant (R² = 0.014, F(1, 254) = 3.575, p < 0.060), indicating that the model explained only 1.4% of the variance in the data that can be explained by the language educators’ perception of the assessment measures. As displayed in

Table 6. Coefficients *.

Model		Unstandardized Coefficients		Standardized Coefficients
		B	Std. Error	Beta	t	Sig.
1	(Constant)	1.686	0.141		11.996	0
	Perceptions of the assessment measures	−0.014	0.007	−0.118	−1.891	0.06

* Dependent variable: assessment practices.

, the language educators’ perceptions of the assessment measures did not significantly predict assessment practices (β = −0.014, p = 0.060).

4.5. Perceptions of Language Assessment Purposes as a Predictor of Assessment Practices

Concerning the correlation between perceptions of assessment purposes during the crisis and assessment practices = 0.069. Simple linear regression was also used to test if the language educators’ perceptions of the assessment purposes during the crisis significantly predicted their assessment practices during this period.

The fitted regression model was as follows: assessment practices = 82.625 + 0.109 (language educators’ perception of assessment purposes during the crisis).

According to

Table 7. Coefficients *.

Model		Unstandardized Coefficients		Standardized Coefficients
		B	Std. Error	Beta	t	Sig.
1	(Constant)	1.273	0.270		4.718	0.000
	Perceptions of assessment purposes during the crisis	0.006	0.011	0.036	0.580	0.562

* Dependent variable: assessment practices.

, the overall regression was statistically insignificant (R² = 0.001, F(1, 254) = 0.337, p < 0.562) and indicated that the model explained only 1% of the variance in the data.

It was found that such perceptions of assessment purposes did not significantly predict the educators’ assessment practices during the crisis (β = −0.006, p = 0.562).

4.6. The Correlation between Self-Efficacy and Assessment Practices

To test the hypothesis of whether there is a significant relationship between the language educators’ assessment self-efficacy and their reported post-COVID assessment practices, simple linear regression was used.

Table 8. Coefficients *.

Model		Unstandardized Coefficients		Standardized Coefficients
		B	Std. Error	Beta	t	Sig.
1	(Constant)	1.153	0.251		4.604	0.000
	Assessment self-efficacy	0.014	0.013	0.069	1.109	0.268

* Dependent variable: AP.

provides a summary of the analysis, showing that assessment practices were the dependent variable and assessment self-efficacy was the predictor variable.

The fitted regression model was as follows: assessment practices = 82.336 + 0.399 (the educators’ perceptions of their assessment self-efficacy). The overall regression was statistically insignificant (R² = 0.005, F(1, 254) = 1.231, p < 0.268), indicating that the model explained only 5% of the variance in the data. It was found that their perceived assessment self-efficacy during the crisis did not significantly predict their assessment practices (β = −0.014, p = 0.268).

5. Discussion

This study was specifically designed to gather evidence concerning the hypotheses about language educators’ perceptions predicting their assessment practices during the crisis. First, the results provided a description of the assessment accommodation characteristics in the various educational contexts investigated. One of the first findings describing the assessment contexts revealed that the assessment measures in their respective countries concerned mainly moving assessment online despite the limited experience these educators had prior to the crisis. This confirms that this New Normal was quite challenging in terms of the teachers’ adaptation to the online assessment mode, as put forth by the majority of researchers during the crisis (Gao and Zhang 2020). Additionally, practical issues were probably at stake not only in the administration of face-to-face assessment but also in online assessment during the crisis, especially because of the teachers’ lack of preparedness in terms of the online assessment.

The findings also revealed that despite the overwhelming nature of the crisis and the absence of clear assessment guidelines, these language educators’ perceptions reflected contradictory perceptions about the assessment measures. Despite the near-total closures of educational institutions during the crisis, two patterns of perceptions appeared in the findings. While about half of these participants believe that exams should be maintained for different professional and accountability reasons, the other half demonstrated more flexibility as to the adoption of alternative assessment methods. The pattern found among these language educators’ perceptions is consistent with previous research findings reflecting beliefs about assessment of learning and assessment for learning as two major assessment paradigms (Brown 2004). Thus, the move to ERT did not systematically compel them to abandon their previous views of the need for formal and summative assessment, most probably in exam-driven assessment cultures. This shows that the crisis did not necessarily lead to a total shift in the language educators’ assessment perceptions as a whole.

It was also found that these educators’ assessment self-efficacy was moderate to high, reflecting their general confidence in their roles as language assessors despite the destabilizing effect of the crisis. This may show that the crisis did not affect their self-perceptions and ability to carry out both formative (i.e., continuous) and summative assessments. These findings are in line with Watermeyer et al.’s (2021) results. However, according to the regression analysis results, assessment self-efficacy did not determine these language educators’ assessment practices despite the forced move to ERT and the importance of self-efficacy in preparing these education stakeholders to face the new assessment demands.

Regarding H1, the findings show that a change in these educators’ assessment practices is not predicted by a change in their assessment perceptions. This finding may confirm that assessors’ perceptions were not necessarily a determining factor of their behavior in the ERT context (Looney et al. 2017). However, this confirms previous research findings (Zulaiha et al. 2020), revealing a possible gap between teachers’ assessment perceptions and practices. Concerning H2, no significant relationship was found between the language educators’ perceptions of assessment purposes and their own assessment practices during the crisis. Likewise, H3 was rejected as no significant relationship between the language educators’ self-efficacy and their reported post-COVID assessment practices was revealed. Unlike previous research findings on the role of self-efficacy (Pajares 1996; Poulou et al. 2019), assessment self-efficacy did not shape these educators’ pedagogical practice and efforts on assessment during the first period of the crisis.

Likewise, these language educators’ perceptions of the assessment measures examined in this study were not found to be predictors of their assessment accommodations and adjustment to the New Normal. Interestingly, the crisis did not generally lead to a conceptual shift in the way these language educators perceived language assessment purposes, despite their confrontation with various assessment challenges imposed by school closure. This may raise more issues about the language teachers’ and teacher trainers’ flexibility in accepting some necessary changes during the crisis. In other words, despite their high assessment self-efficacy, their beliefs that formal assessment through exams had to be maintained may be indicative that they were not fully ready for this crisis (Tuah and Naing 2021). The complexity of the situation might have been intensified by these language educators’ relative openness to adaptations due to the crisis on the one hand and the difficulties of embracing any changes on the other, mainly because of their limited experience in online assessment. Institutional factors such as the subject taught, the learners’ level, the available resources, and the local assessment culture might have also played a role in shaping both their beliefs and practices. The results seem to confirm initial concerns about educational crisis management issues in different contexts worldwide (UNESCO 2020a, 2020b, 2020c).

The absence of any correlations between the assessment perceptions and practices of the participating language educators during the COVID-19 crisis may generally suggest that the accommodations were not dependent on teacher factors such as their assessment self-efficacy as much as other contextual factors. The educational context, prevailing assessment culture, the educators’ language assessment literacy, and technological infrastructure may have played a role in the adoption of the measures and the implementation of the assessment accommodation strategies by the study participants. All these factors may have contributed to the complexity of managing the situation effectively from a language assessment perspective.

6. Conclusions

The findings of the present study have interestingly shown an absence of any relationship between the language educators’ assessment perceptions and practices during the COVID-19 crisis. Researching the questions related to the responsive measures vs. perceptions and practices regarding assessment during crises has been found to be fraught with some challenges. The difficulty in addressing these issues was partly due to the large scale of the crisis and its impact on assessment at various educational levels. In addition to this, the use of a single data collection instrument and quantitative analyses may also need to be further confirmed in more specific educational settings while accounting for contextual factors such as the available technology infrastructure, learner computer literacy, and attitudes to online assessment. This study focused on language educators’ perceptions and practices during their expected adjustment to a demanding New Normal. Future research on assessment during crises should probably deal with the impact of these assessment accommodations on students’ perceptions and performance. This would lead to a better understanding of learners’ needs during the pandemic. Future research should develop an assessment model and/or guidelines to help language educators cope with the crisis more effectively. This could contribute to the adoption and implementation of more informed decisions regarding ERT in general and assessment in particular. From a pedagogical perspective, language educators need to gain more knowledge and skills related to managing assessment for learning sustainability rather than just pedagogical continuity and student resilience in times of crisis. From this perspective, given the long-term nature of the crisis and the need for ERT as a strategic step for learning continuity, teaching and assessment should move beyond mere “psychosocial support” (Anderson 2020) to incorporate more planned research-based assessment strategies at national and international levels.