Continuation Task: A Critical Review and Directions for Future Research

Continuation task has witnessed a tremendous growth of research since it was firstly advanced by Wang, as an effective approach to L2 learning. The basic idea of continuation task is that learners should first read a given text with the end removed, then they should finish the continual writing in a logical and coherent way with imagination on the basis of the given text. In the process of continuation, learners need to not only fully understand the given text, but create new content, imitate and use the words and structures that appeared in the given text, thus producing interactive alignment effects, and improving the efficiency of learning foreign language. Researches on continuation tasks cover key issues regarding to the effectiveness on L2 learning and task-related or learner-related factors in continuation task that influence L2 learning. Since the number of researchers investigating the application of continuation task to L2 teaching and learning continues to grow, it is necessary to assess the current state of the researches in this area and establish future directions of investigation. This review aims to provide an overview of recent researches on continuation task and offer some suggestions for future researches.


Introduction
Language learning is a process where learners are able to use the language of others' [3], so to some degree, language is learned through imitation. More importantly, language learning is achieved through a creative process where learners, who are driven by communication motivation, produce language in complex context and interaction. So this process is bound to be dynamic. Language structures are easy for learners to repeat in their own language production, while the dynamic creation process in language use is difficult to imitate, yet essential for language learning and language use. So learning language through imitation is not equal to rigid repetition, but to follow the dynamic and creative rule for language learning. Therefore, combining imitation and creative use should promote language learning. To achieve this, Wang [1] proposed continuation task, in which learners comprehend the text first and continue to create writing based on the given text. The reading material provides sources for learners to imitate and the continuation task gives learners communication drive to create new content. In order to create a writing coherent to the given text, learners tend to use the previously occurring expressions in their own creation. So what they have comprehended and learned in reading can be creatively reused in production, thus, language is learned through the coupling of language comprehension.

Introduction to Alignment Theory
Why does continuation task facilitate language learning? The mechanism behind lies in the interactive alignment effect [4] derived from the interaction in personal dialogue which serves as the prerequisite of smooth communication. In written language, there also exists such interactive alignment between input (comprehension) and output (production). The concept of alignment did not originate in the field of language acquisition, but appeared in the fields of psychology. Alignment theory was firstly put forward by Pickering & Garrod [4]. They hold that dialogue is the most basic and natural form of language use, and they attribute successful communication to the interactive alignment effect existing in the language use. To further elaborate this, they proposed Interactive Alignment Model, according to which, alignment is defined as the coordination and adaptation of language representations at different levels, namely, linguistic level and situational level between interlocutors in their conversations [4]. It implies that the speakers and listeners of a dialogue negotiate and coordinate with each other, inspire each other, even unconsciously and continuously construct a mutually convergent linguistic and situational model so that the representations at various levels are interconnected and aligned. It is the alignment of linguistic representations at different levels that produces smooth and successful communication.
Pickering & Garrod [4] maintain that alignment of situation models is essential to successful communication, which draws on from Zwaan & Radvansky [5]. According to them, situation models consist of five key dimensions, namely, space, time, causality, intentionality, and reference to main individuals under discussion. interlocutors will form a dynamic concept of situation model according to their situation including time, space, intention, causality, and character. Establishing these dimensions helps the comprehension of the text, which means interlocutors share the similar multidimensional representations of time, space, intention, causality, and character, therefore promoting mutual understanding in dialogues. The successful communication hinges upon the degree to which the interlocutors produce the same or close representation of language within the situation models.
Alignment of situation models is achieved through the alignment at linguistic levels. This works by structural priming [4,6], whereby the interlocutors will more likely to produce an expression using the representation that has been previously activated by a certain expression. Thus, processing an expression that triggers a certain aspect of a situation model will add the possibility that the interlocutor will use an expression congruous with that aspect of the situation model. In this way, "local" alignment of linguistic representation brings about "global" alignment of situation models and vice versa.
To sum up, the Interactive Alignment Model assumes that in communication, interlocutors align their linguistic representations at many levels ranging from the phonological and lexical to the syntactic and semantic, even to discoursal level. And this interactive alignment process depends on the priming mechanisms and a parity of representation for production and comprehension, that is to say, previously encountered and comprehended linguistic representations will be used in the following production, thus enhance the alignment of situation models. In this way, the comprehension and production are coupled, alignment between representations linguistic level and situation models is strengthened, and mutual understanding between interlocutors is achieved.

Alignment in Second Language Acquisition
Different from Pickering and Garrod, who take alignment as a mechanistic account to successful oral communication, Atkinson et al. [7] sees alignment from a socio-cognitive perspective, whereby alignment goes beyond to account for second language learning through integrating environmental and social factors during the alignment process. According to Atkinson et al., alignment under socio-cognitive view is defined as a dynamic adaptation to the mental and physical environment.
The interaction through articulated mind-body-world activities produces a synergistic effect on second language development. That is to say, the dynamic interactions with environment at social level lead to alignment at cognitive level.
Costa et al. [8] conducted a research involving conversations under cross-linguistic context (i.e. L1-L2 conversation or L2-L2 conversation). They found that the alignment effect in L1 conversations functioned in L2 or L1-L2 conversations as well. L2 speakers in a cross-linguistic conversation, whose vocabulary and knowledge of grammar are limited, may find it difficult to express themselves in second language, although it is easier for them to understand other L1 speakers due to the rich context available. In order to fill in the gap between comprehension and production, L2 speaker tended to refer to the expressions that have previously heard from the L1 speaker to guarantee successful communication. This is also known as structural priming, which has been discussed above. In this case, structural priming is regarded as the mechanism of the facilitative role of alignment on second language learning.
As alignment is rooted in interaction, dialogue should be the most beneficial way of interaction, on the one hand. listeners make judgements based on what the speakers say, and speakers react on listeners' feedback. Communication goes smoothly under the alignment of situation model. On the other hand, Wang [9] argued that there are also many drawbacks in dialogue, such as unnecessary repetition, omission and stumble etc., and dialogue possesses fewer syntactic complexity and discourse coherence. More importantly, not every learner gets the opportunity to speak with native speakers. In order to seek an effective approach to second language learning, and inspired by the modality-independent nature of situation models [5], which means that the representations of a certain situation will not be influenced by the different media that represent the situation model, some L2 researchers begin to study alignment in L2 writing. The independent modality of situation models makes it possible that they can be employed in cross-linguistic context and in text comprehension and production. And evidence of alignment effect has been found on L2 collaborative writings [10], proving that writing is more than a cognitive process, instead, it is always a socio-cognitive process involving integrated mind-body-world activities.
Wang expanded Atkinson's idea of alignment between human and environment, and began to explore alignment between L2 learners and reading materials. Wang [1] firstly explored alignment in L2 writing by a continuation task. His study consisted of two experiments: first experiment required two groups of EFL students to finish English continuous writings following an English story and its Chinese version respectively; the second experiment involved writing tasks on two different topics concerning different linguistic experience (students L1 and L2 experience). Wang found that alignment effect manifested itself in the writing performance of English-English continuation task as the writings contains richer lexical items and less errors and students performing the writing task that activates their L1 contextual knowledge produce more errors in the writing. And these findings were later proved again by Wang and Wang [2].
Wang [1] maintains that alignment hinges upon the magnitude of the interaction with the input, and it is also pointed by Pickering and Garrod [4] that alignment can be reinforced and enhanced by the production coupled with comprehension. And this reinforcement and enhancement is achieved through an asymmetry between the linguistic representations in comprehension and production [9]. Since L2 learners have better competence in comprehension than production, by leveling the comprehension and production, alignment expand L2 learners' linguistic representations, resulting in language acquisition. Different from dialogues, which produce unconscious and autonomous alignment, continuation task "forces" learners to align with the situation models in the reading material to ensure content and linguistic coherence with the text, because in order to complete the story, learners must fully understand the given text and construct situation model that consequently leads to alignment of linguistic representation or structural priming. Wang' exploration of alignment into L2 writing expands the facilitative role of alignment from dialogues to texts, which can be wildly applied in second language teaching.
Wang [11] further elaborated how alignment promote second language learning through imitation of language uses in interaction. Imitation here should not be regarded as behavioral repetition, instead, imitation in interaction can be viewed as structural priming or linguistic alignment led by situation models constructed in rich context, so it involves dynamic adaptation between comprehension and production, hence, facilitating second language learning.
To sum up, Interactive Alignment Model not only promotes smooth dialogues, but facilitates second language learning through coupling comprehension and production. In L2 continuous writing task, the asymmetry between linguistic representations in comprehension and production drives learners to align with the reading material, thus structural priming is produced through the interaction with the context, thus, asymmetry is resolved.

Studies Concerning the Effectiveness of Continuation Task in L2 Learning
Since Wang [1] advocated continuation writing task as an effective approach to English learning and manifested the facilitative role of alignment between learners and the reading material, a series of studies concerning the effect of continuation task on L2 development have been conducted. Jiang & Chen [12] contended that facilitative effect of continuation task could not be fully proved simply by the short-term effect and by just examining the accuracy of the production, so they conducted a longitudinal study to explore learners' writing performance of language accuracy, complexity and fluency. They found that accuracy and complexity in continuous writing outperformed those in topic writing. Zhang [13] also explored the longitudinal effectiveness of continuation task in college students. The results indicated that long-term alignment effect did exist and last longer in continuation task than in topic writing task but there was no significantly different sustainable facilitating effect on reducing linguistic errors. And the results of this study coincide with the findings in L2 Chinese learning discovered by Wang and Wang [14], who found frequent repetition of words, chunks and syntactic structures in reading material existed in continuous writings and doing a continuous writing can significantly reduce linguistic errors. Jiang & Tu [15] verified the effectiveness of continuation task on L2 vocabulary learning in form, meaning and use. She concluded that continuation task can better promote L2 vocabulary learning than summary task in meaning and use. Similarly, there were some researches reporting that learners learned English articles better in the continuation task than in a reading task [12,16].
There were also studies concerning the facilitative effect of continuation task on syntactic structures. Zhang [17] reported that higher syntactic pattern density and syntactic complexity were found in writings of continuous task. Wang and Wang [18] studied how continuation task promote learning of some complex syntactic structures (relative clause as an example) by comparing the production of continuation task with and without relative clauses. Their study indicated that the number of relative clauses used in continuous writings largely depended on whether the reading materials contained relative clauses, which again verified the facilitative effect of alignment with reading materials. Wang & Cao [19] took English passive structure as an example to examine whether structures in the reading material of a continuation task led to repetition in the production, and the results gave an affirmative answer as well.
Since alignment can be found in multiple interaction forms and different linguistic levels [4], some researchers went beyond the lexical and syntactic level and started to explore the facilitative effect of continuation task at discoursal level. Zhang and Zhang [20] examined whether different genres of reading material affected the magnitude of alignment and the number of linguistic errors produced in the continuous writing. They found more intensified alignment effect and less errors in continuation task of argumentation in contrast with narration. Miao [21] confirmed the existence of discoursal alignment and the magnitude of discoursal alignment varied under different interactive contexts (internal and external interaction contexts). The study of Peng [22] proved that learners tended to refer to previous expressions to achieve linguistic and content coherence, and the magnitude of coherence accorded with the degree of alignment and comprehension. Yang [23] conducted a qualitative research and found facilitative effect of continuous task on rhetorical devices used in writing. Interestingly, learners who complete continuation task tended to seek deeper comprehension and interpretation of the reading material in order to align with the writing style of the reading materials.

Studies Concerning the Factors Impacting Alignment in Continuation Task
Other researchers turned to focus on the influencing factors of alignment effect in continuation task. Some studies examined the effect of linguistic complexity of the reading material on learners' writing performance in continuation task [24][25][26]. The results of Peng's [24] and Peng et al.'s study [25] showed that learners who read text suitable for their writing ability (simpler for them to comprehend) seemed to produce more fluent and accurate writing than those who read text tailoring to their comprehension ability. However, Xin & Li' study [26] showed a different result: linguistic complexity and textual length did not bring significant increase of accuracy. One possible reason for this may be that learners used most of their cognitive resource to construct meaning, not noticing structures that are less meaning-oriented. Wang & Wang [2] compared effects of different language input on learners' writing performance and results demonstrated that alignment effect intensified when the language input and the production are in the same language form. Besides input of different language forms, different input modes also came into researchers' focus. Zhang et al. [27] reported that continuation after listening can better improve vocabulary acquisition than continuation after writing which suggested that audio-visual mode of input led to much more notice on language input than single audio or visual mode of input.
Zhang & Zhang [20] reported a stronger alignment effect of argumentative continuation over narrative continuation on language accuracy while narrative continuation exceled in improving language fluency of learners production. Another study concerning argumentative continuation also found alignment in argumentative writing [28]. Both of these two studies have shed light on the universality of alignment in facilitating L2 learning and enriched the form of continuation task.
Other researchers began study influencing factors on learners' side. Sun & Wang's study [16] revealed that greater alignment effect of acquisition of English articles was found in learners of high and low L2 proficiency, but no significant improvement was found in learners of medium proficiency because high-level learners can quickly construct situation model and poor learners need to constantly refer to the reading text to bridge the gap between comprehension and production thus promote learning of English articles. As for medium learners, they did not frequently align with the text because it is relatively easier for them to construct situation models but this caused lower frequency of alignment with the text. More importantly, they tended to employ familiar vocabulary and structures to avoid mistakes, thus use of articles, as a difficulty for Chinese English learners, was restrained. Findings by Wang and Cao [19] also showed that lower alignment effect was found in medium learners. They attributed this to the relatively lower enhancement of passive construction in their mental representations. These two studies revealed that learners' internal L2 knowledge greatly influenced the facilitative effect of alignment.
Studies above demonstrate that alignment in continuation task is a kind of autonomous mechanism that brings about implicit learning process, but recently, some researchers discovered that combining explicit instruction [29] or text manipulation [30] with continuation task could lead to greater facilitative effect. Xin [30] explored the influence of enhanced continuation task on L2 grammatical structure acquisition. The results showed that no significant promotion of acquisition of subjunctive mood was observed in ordinary continuation task, while in continuation task under enhanced edition, researchers have found a significant facilitation on the acquisition of subjunctive mood. This indicated that noticing played an important role in strengthening alignment effect, and the results also manifested that continuation task could be a useful tool for teaching grammar structure by inserting target structures in the reading material with some enhanced modification. Gu and Zhou [29] conducted instructive tasks at the interval between reading and writing and the results showed that explicit instruction significantly improved learners' lexical richness and accuracy in the writings.

Conclusion and Future Directions
Empirical studies confirmed the facilitative role of alignment in L2 learning of vocabulary, grammatical structure, discoursal knowledge etc. However, since this approach is still underdeveloped, there are still some problems remaining to be further explored.
Firstly, long-term effectiveness of continuation task should be further examined. The results of current studies support that by doing continuation task, errors in learners continuation writings are significantly reduced [1,2,13,14,20], but this cannot fully prove the facilitative effect of continuation task on L2 learning, because we cannot firmly say the alignment between the reading material and learners' production occurs due to the creative use based on comprehension instead of just by copying the previous expressions. So more longitudinal research is needed to examine the long-term effect of alignment in continuation task. In particular, whether new linguistic knowledge is successfully learned largely depends on its residual in memory and its transfer to new tasks. So more research should be made to trace the learning effect in long-term and examine its application to new forms of tasks.
Additionally, researchers should more carefully consider the potential factors that influence the alignment effect in continuation task. Existing studies mainly focus on task-related factors, such as the complexity of the reading material [24][25][26], language input [1,2] and the genre of reading material [20]. But factors at learners' side is understudied, like learners' awareness of coherence and logic in writing, learners' comprehension of the reading materials, and learner's existing knowledge. Especially, investigating the role learner's L2 proficiency play in the alignment of continuation task would benefit the application and promotion in L2 teaching and learning for the generalizability of current findings to broader populations, since the current studies haven't reached a consensus.
More importantly, also understudied in the influencing factors is how teacher instruction affect the alignment effect. Continuation task provide an internal and autonomous drive for learners to make full use of the existing and new information to create new content, but teacher's explicit instruction inserts an extra boost to learners' awareness of alignment [29]. So more studies related to teacher's instruction, such as the form of instruction that best boosts alignment, the timing of teacher's instruction etc., remain to be further investigated for better application for teaching.
In conclusion, researchers have made great contribution in the 8 years to understand how and under what circumstances continuation task can promote and contribute to L2 learning. As highlighted in this review, continuation task provide a critical means to advance L2 learning and teaching. However, crucial questions remain, providing many avenues for future investigation.