Focuses and Trends of the Research on Task-based Language Teaching (1998-2022)

,


Introduction
Task-based language teaching (TBLT) originated in the 1980s. In the past 40 years, researchers have been exploring how TBLT can promote second language acquisition and achieved fruitful results. Interaction hypothesis and focus on form are the main theoretical basis of TBLT research. Long (1983Long ( ,1985Long ( ,1989 believed that interaction in tasks was very important for second language learning, because it not only provided understandable language input, but also triggered learners' attention to language forms in communication. Schmidt (1990Schmidt ( , 2001 pointed out that attention to form was a necessary condition for second language acquisition. Mackey (2007) also believed that meaning negotiation in interaction could help learners construct the connection between language form and meaning faster, thus promoting the development of interlanguage. A meta-analysis of relevant studies by Keck et al. (2006) found that both task-based interaction and corrective feedback had positive effects on second language acquisition. Some researchers have studied TBLT based on sociocultural theories, such as Lantolf (Lantolf & Thorne, 2006;Negueruela & Lantolf, 2006) and Swain (Swain, 2000;Swain & Lapkin, 1995). Within the framework of sociocultural theory, tasks are viewed as social activities that promote language use and create cooperative language learning opportunities. Ellis (2003) pointed out that language development involved the learning of knowledge and the practice of social activities that were the main sources of cognitive function development.
In the late 1990s, researchers represented by Skehan and Robinson proposed and developed the cognitive theory of task research. From the perspective of cognitive psychology, the process of learners' completing tasks is considered to be a process of cognitive processing of attention, memory, reasoning and other information. Therefore, reasonable task design is conducive to easing the cognitive load in tasks and enabling learners to pay more attention to language forms, which has a positive impact on task performance and second language development. In recent years, many researchers have studied TBLT in terms of task performance, task characteristics, task condition, task planning and so on (e.g. Mohammad & Mahsa, 2020;Sanchez & Kalamakis, 2022;Tabari, 2022).
The empirical research on TBLT is productive, but a few studies reviewed previous literature and put forward some suggestions. Jackson and Suethanapornkul (2013) conductued a synthesis and meta-analysis of research on second language task complexity. Long (2016) identified some genuine problems in TBLT and briefly explained research programs under way. These studies, however, are not enough to show the general picture of TBLT research. Therefore, with the mix of quantitative statistics and content analysis, this article analyzes the papers on TBLT published in internationally authoritative second language acquisition academic journals from 1998 to 2022, to identify overall trend, main subjects, focuses and methods of research, thus providing reference and inspiration for future research on TBLT.

Data Sources
In this study, the database Web of Science is used to search for papers on TBLT published in ten internationally renowned second language acquisition academic journals from 1998 to 2022. The search fields are set as "Topic = task-based OR Topic =TBLT OR Topic =task* OR Topic = task-based language teaching OR Topic = task-based language learning". According to the research of Plonsky and Kim (2016), the following ten journals are selected: Applied Linguistics, Language Learning, The Modern Language Journal, System, TESOL Quarterly, Language Teaching Research, Studies in Second Language Acquisition, Annual Review of Applied Linguistics, Language Teaching, Journal of Second Language Writing. The search yields 1682 results, 1362 of which are unrelated to the topic and eliminated, then 320 valid literatures are finally obtained as research samples.

Research Methods and Tools
This study adopts a mix of quantitative and qualitative methods. First of all, such information of research papers as the year, author, title, journal, keyword, abstract, research method, research subject, and target language are saved as text files and formatted. Then Excel is used for descriptive analysis and statistics of high-frequency words. Finally, content analysis is carried out, focusing on four dimensions: overall trend, research subjects, research focuses and research methods. This paper reviews and discusses the research status of TBLT from 1998 to 2022.

Overall Trend
From 1998 to 2022, the number of research papers on TBLT each year has shown a dynamic change in the past 25 years, and generally maintains an upward trend (see Figure 1). Before 2011, the annual number of publications was less than 8, and the development of research in this field was relatively slow. Then the number increased to 16, and decreased slightly in the following three years. In 2015, 2016, and 2017, the number of published papers showed a surge trend, rising year by year, and reached a peak of 41 papers in 2017. In the past five years, 107 papers were published, accounting for 33.4% of the total, indicating that the research has developed steadily. It can be seen that researchers have paid more and more attention to TBLT and affirmed its academic value.

Research Subjects
The research samples of this study include 272 empirical studies on TBLT, covering subjects at different levels.
There are 204 studies on college students (adults), accounting for 75%, 34 papers on middle school students (12.5%), 15 and 12 studies on teachers and children respectively. According to this distribution, the research on TBLT mainly focuses on undergraduates and adults, and pays insufficient attention to primary school students and children. Table 1 shows the information of target language and native language of research subjects in these 272 studies, only 165 of which introduced native language source. It can be found that the majority of subjects are EFL / ESL learners from Asian and European countries, and some are learners of Spanish, German, Chinese and French in English-speaking countries.

Research Focuses
Keywords are generally the summary of main points of a paper. Analyzing the frequency of keywords can help us grasp the hot topics in related research fields. The higher frequency indicates more researchers' attention. In this study, 685 keywords were collected from 320 valid literatures. Some terms with broad meaning were deleted, including "TBLT, task, task-based language teaching, task-based language learning", and words with the same or similar meanings were bracketed together. On this basis, according to Donohue (1973:49), the threshold of high frequency words was determined as 3, then 30 keywords were finally extracted and ranked in the order of word frequency, as shown in Table 2. With the analysis of these 30 high-frequency keywords based on the theoretical framework of Skehan and Robinson, we find that the research hotspots of TBLT from 1998 to 2022 mainly cover five aspects: (1) task performance, (2) task characteristics, (3) task implementation conditions, (4) learner internal factors, (5) integration of TBLT and computer network.  Table 2 shows that task performance research has received the most attention, including keywords 3, 5, 9, 14, 23 and 24. The keyword number three, complexity, accuracy and fluency (CAF), is usually used to analyze participants' task performance. At present, research on task performance of second/foreign language learners mainly includes two aspects.
As for the feature of linguistic forms in learners' task performance, some studies look into a single grammatical or lexical item. For example, Michel (2013) investigated the use of conjunctions by learners in oral tasks of different complexity. Most researchers analyzed linguistic forms in task performance from the perspective of CAF. Abrams and Byrd (2017) studied the effects of pre-writing tasks on L2 writing performance, and the results showed that meaning-focused cooperative pre-writing tasks could improve grammatical accuracy, vocabulary richness, and overall writing quality. Mohammad and Mahsa (2020) empirically explored the relationship between task design and L2 oral performance, and found that task complexity was positively correlated with overall complexity and fluency, and strategic planning and task repetition promoted fluency and syntactic complexity while their combination enhanced accuracy as well.
Some studies examine interactive features, focus on form and other features of non-linguistic form in learners' task performance. For example, Kim (2013) investigated the influence of pre-task planning on attention to form by using the thinking aloud method and the analysis of language-related episodes, and found that pre-task demonstration strategies could help enhance learners' attention to linguistic form during task completion. Gass, Mackey, and Ross-Feldman (2005) compared learners' interaction performance in laboratory and classroom environments, and the results showed that there was no significant difference in the types of interaction between the two environments.
In general, although the study of task performance has achieved fruitful results, researchers pay more attention to oral tasks (keyword 1), but less to writing tasks (keyword 2).

Research on Task Characteristics
The influence of task characteristics on second/foreign language learning is another hot topic in TBLT research in recent years. Different characteristics have been investigated by researchers. Task structure (keyword 28) is a feature that has received early attention (Skehan & Foster,1999), and subsequent research results are consistent with the original findings that task structure is beneficial for improving linguistic accuracy and fluency in declarative oral tasks (e.g. Tavakoli & Foster, 2011;Sanchez & Kalamakis, 2022). Some researchers have also investigated the effect of task characteristics on language learning in terms of task difficulty (keyword 20) and task types (keyword 12) including narrative, problem-solving and decision-making tasks (Kormos, 2011).
With more and more studies on TBLT being based on cognition theory, cognitive task complexity (keyword 4) has become the most concerned task feature at present. There are 50 papers involved, 19 of which are studies in the past five years. Besides, the keyword "Cognition Hypothesis" is also related to task complexity. Skehan and Foster (2001) believed that the attention resources of second language learners were limited, and complexity and accuracy of language production would decrease with the increase of task complexity. Robinson (2001) proposed a multi-attention resource model, emphasizing flexibility in the allocation of attention resources and dividing task complexity into resource-directing and resource-depleting dimensions. He argued that increasing the complexity of resource-directing tasks enhanced the syntactic complexity and accuracy of the language. Based on the above hypothesis, a large amount of research has been generated and the research on oral tasks is significantly more than that on writing tasks (Jackson & Suethanapornkul, 2013). However, the results of the studies are inconsistent. Fukuta and Yamashita (2015) took reasoning as the complexity variable and found that learners' accuracy of oral production in complex tasks was improved, fluency was decreased, but complexity did not change significantly. However, Awwad, Tavakoli, and Wright (2017) found that oral tasks with high reasoning complexity promoted the syntactic complexity and accuracy of the language, but reduced the lexical diversity, and had no significant impact on fluency. Sanchez & Kalamakis (2022) pointed out that resource-directing task complexity had a significant effect on complexity, accuracy, fluency and lexis of written language production. Kuiken and Vedder (2008) also investigated the impact of resource-directing complexity on L2 writing, and the results showed that task complexity could improve learners' language accuracy, but had no significant impact on syntactic complexity.
After reviewing the literature, we find that these studies focus on the impact of task complexity on language production, but the conclusions are not consistent. On the one hand, the dimensions of measuring CAF are different, and on the other hand, there may be a tradeoff among the attention to complexity, accuracy and fluency.

Research on Task Implementation Conditions
Research on task implementation conditions has also received more attention, mainly exploring the correlation of pre-task planning, online planning, or task repetition with task performance (key words 6, 7, 19, 22 and 26).
A large number of studies have investigated the influence of pre-task planning on the language production of second/foreign language learners (e.g. Meraji, 2011;Qin Jie,2022). The results show that pre-task planning is significantly positively correlated with complexity and fluency, but the conclusions on accuracy are mixed. Planning time is another factor affecting L2 production. Mehnert (1998) explored the influence of planning time on oral task performance. He pointed out that within a limited planning time, learners first paid attention to accuracy, then complexity, and fluency would only be improved when the time reached 10 minutes. Ellis and Yuan (2004) conducted a study on the English writing of undergraduates in a Chinese university, and concluded that the 10-minute planning time improved fluency and complexity of the language, but had little effect on accuracy. In addition, Kang and Lee (2019) found that collaborative planning could significantly improve fluency and syntactic complexity, and the effects on lexical complexity were affected by task complexity.
Considering that pre-task planning had little effect on accuracy, Ellis (2005) suggested that online planning should be carried out during task in order to improve language accuracy. Some empirical conclusions show that online planning can indeed improve accuracy and complexity of language (Yuan & Ellis, 2003;Ahmadian, 2012;Tabari, 2022). Both Ellis and Ahmadian interpret the conclusion from the perspective of working memory capacity (keyword 15), arguing that when the time pressure of planning is reduced, learners will release more attention to the language form and monitor the language production, thus increasing the task performance. However, Ahmadian's study also found that the influence of non-stressful online planning on verbal fluency was not as significant as that of stressful online planning. In addition to online planning, interactivity is another condition in task completion. Researchers mainly compare the different effects of interaction on task performance, and the results show that compared with the tasks completed by learners independently, interactive tasks are more conducive to improving task performance (Michel, 2011;Fernandez, 2012).
Task repetition has also received increasing attention as a task condition. Existing studies generally prove that task repetition promotes oral production. For example, Qiu (2020) found that task repetition improved language accuracy, fluency and lexical complexity by repeating oral monologic tasks. Lynch and Maclean (2000) found that task repetition could improve fluency and accuracy; Thai and Boers (2016) examined the repetition of monologue tasks and found a significant positive correlation between task repetition and fluency. More elt.ccsenet.org Vol. 16, No. 9;2023 researchers investigated task repetition from various perspectives. Kim and Tracy-Ventura (2013) compared the different effects of task content repetition and process repetition on learners' language production. The results showed that process repetition improved syntactic complexity, while content repetition had no effect on syntactic complexity, and there was no difference between the two types of repetition in other aspects. Fukuta (2016) examined the effects of task repetition on focus on form in oral tasks, and found that learners focused more on syntactic encoding process and less on conceptualizing process in repetitive tasks. Qiu and Lo (2017) investigated the impact of task repetition on L2 learners' oral task engagement. The research showed that task repetition was negatively correlated with engagement, and learners were more willing to repeat tasks with unfamiliar topics.
In the study of task conditions, most researchers pay attention to the pre-task and during-task stages, while there is relatively little research on the post-task stage which is limited to the effects of post-task transcribing on oral performance (e.g. Foster & Skehan, 2013;Li, 2014). In addition, the effects of task repetition intervals and times on language production also need to be further studied.

Research on Learners' Internal Factors
In Table 2, keywords 13, 16, 17, 18 and 29 show that the internal factors of learners are another focus of TBLT research, which is consistent with the development trend of second language acquisition in the 21st century. Some studies examine individual factors that affect language production and learning in terms of learner's second language proficiency (e.g. Xu & Fan, 2021), learning ability (Kormos & Trebits, 2012), and emotional factors such as anxiety and motivation (e.g. Kang & Kim, 2021).
In recent years, there have been more and more studies on affective factors, among which task motivation and task engagement are considered to be crucial for second language learning (Philp & Duchesne,2016). Therefore, how to improve learners' motivation and engagement in task-based teaching has become an important topic. Murakami, Valvona, and Broudy (2012) explored the relationship between assessment forms and task engagement by using questionnaire surveys before and after the experiment. They found that students' participation in assessment was conducive to improving task engagement, and teacher evaluation together with students' self-assessment and mutual assessment had the most significant impact. Also, content familiarity, high language proficiency, and leadership would have a positive impact on task engagement and motivation (Emmanuell & Tavakoli, 2022;Leeming, 2021). However, Nakamura, Phung, and Reinders (2021) and Mozgalina (2015) found that learners' independent choice in a task was negatively correlated with motivation and engagement, because too many choices would increase learners' pressure to complete the task.
This kind of research mainly investigates learners' language proficiency or affective factors in task-based teaching, but there are few studies on learners' age, cognitive style, and self-efficacy, so the research perspective needs to be expanded.

Research on Integration of TBLT and Computer Network
The 21st century is the age of network information, and the integration of computer-mediated communication (CMC) into task-based teaching has gradually become a hot topic.
Some studies investigate the influence of CMC in TBLT on the negotiation of meaning that is one of the important factors to promote second language development. Smith (2003Smith ( , 2004Smith ( , 2005 conducted a series of studies on the negotiation of meaning in CMC and found that learners would engage in the negotiation of meaning when they had misunderstandings, and the frequency of negotiation was affected by task types. van der Zwaard and Bannink (2014) found that negotiation occurred in both video call and instant message interactions, but there was a significant difference between these two negotiation modes, and the negotiation mode in video call was closer to the face-to-face (FTF) mode. However, Ribeiro & Eslami (2022) found no statistically significant difference in the frequency of negotiation in FTF and text-synchronous-computer-mediated communication (SCMC) interactions.
Attention is another necessary condition for second language acquisition (Schmidt, 2001). Studies have shown that CMC, especially text-chat mode, can promote learners' attention to form (Yilmaz, 2011), and task structure and pre-task language support are positively correlated with focus on form in CMC (Nik, Adams &Newton, 2012). However, some studies have pointed out that there is no significant difference in learners' attention to feedback between online and FTF contexts (Qiu & Bui, 2022).
Additionally, some studies try to find what task factors will affect language production and second language development in CMC. Heift and Rimrott (2012) studied the relationship between task types and language production in online context, and found that meaning-focused tasks could significantly improve language accuracy. Hsu (2017) compared learners' oral production under two planning conditions in CMC, and found that planning with pre-task practice could promote the improvement of complexity and accuracy. Other studies compared effects of task factors in CMC and FTF modes. Baralt (2013) examined the effects of task complexity on second language learning in the FTF and CMC, and found the positive correlation in FTF, but negative in CMC. Qiu & Bui (2022) investigated the effects of pre-task planning on learners' engagement in oral tasks in the FTF and CMC modes. The results showed pre-task planning did not significantly affect the learners' engagement in either the FTF or the CMC condition.
In general, the integration of task-based teaching and computer network plays a positive role in second language learning. However, the limitation of this kind of research is that there are few studies on how the emerging network technology or multiple technologies affect task-based teaching. Meanwhile, how to design tasks that adapt to both the network environment and task-based curriculum needs to be further explored.

Research Methods
According to the statistical results of the research methods of 320 studies, there are three significant characteristics of research methods of TBLT during 1998-2022 (see Table 3 and Figure 2). First, the statistics show that empirical research occupies a mainstream position in TBLT research, with 272 papers, accounting for 85% of the total. There are 48 non-empirical studies, accounting for only 15% of the total. Empirical research has scientific methods, strong operability and more objective results, while non-empirical research is helpful to improve the theory and expand the research perspective. Second, as can be seen from Table 3, most empirical studies on TBLT adopt quantitative methods, among which experimental research is dominant. There is little qualitative research, but its description and analysis of the process of TBLT can make up for the shortcomings of quantitative research to a certain extent. Therefore, more and more studies use both quantitative and qualitative methods. For example, Awwad et al. (2017) used questionnaires and open-ended questions to collect quantitative and qualitative data respectively to explore the relationship between learners' perception of task difficulty and task complexity.
Third, Figure 2 shows that specific research methods and tools are increasingly diversified. Quantitative research tools mainly include experiments, questionnaires and corpora. In addition to interviews and observations, new qualitative research methods are emerging, such as case study, thinking aloud, action research, conversation analysis and so on. Some new research methods and tools, such as eye-movement tracking, have also begun to be applied in TBLT research.

Conclusion and Implications
With the quantitative and qualitative research methods, this paper conducts literature search, statistical analysis and summary of TBLT research in the past 25 years and finds that the number of published papers shows a dynamic increasing trend. These studies examine the influence of TBLT on language production and second language development. The research topics mainly include task characteristics, task conditions, task performance, learner factors, and the integration of TBLT and computer network. The research perspectives are increasingly diversified. Interaction theory, cognitive theory and sociocultural theory have been applied in the research one after another. The research subjects are mainly college students with intermediate English proficiency, while less children and high proficiency learners are examined. Quantitative empirical research is still dominant, while the number of qualitative research or the mix of quantitative and qualitative research is relatively small. This paper has certain implications for future research on TBLT: (1) For previous inconsistent findings, replication studies are encouraged in different environments to verify the effectiveness of task complexity or task performance measurement methods and promote the systematization of research (Long, 2016). (2) Since previous studies mainly focus on intermediate proficiency EFL/ESL in universities, more studies should examine younger learners or advanced learners, and non-English-as-a-foreign-language learners, such as international students learning Chinese in China, are also worthy of attention.
(3) In terms of research topics, the measurement of task complexity, task evaluation, teachers' role, the internal process of language input, absorption and expansion, the moderating effect of individual differences, and the mixed use of speaking and writing modes will all deserve more attention to improve the present research. Besides, the three stages of pre-task, during-task and post-task involve language input, interaction and output, thus forming an interrelated dynamic task ring. However, most of the previous studies focused on the implementation stage of the task, ignoring the correlation between the pre-task and post-task. Therefore, the dynamic development process of each stage in the task ring needs to be further explored. (4) The mix of quantitative and qualitative research methods is suggested in future research, and some advanced technologies such as corpora, eye-movement tracking can be used to make research results more accurate and objective.