1. The concepts involved in Assessment for Learning

Learning
The shift in thinking about assessment discussed in the previous section was related to the developments in how learning was conceptualised. Watkins (2003) listed three view of learning: behaviourist, cognitive constructivist and socio-cultural. The former is concerned with stimulus-response learning and in FL teaching is manifested in audiolingual approaches. The second perspective focuses on the importance of processing information and on individual characteristics (see also Section 2.2.1). The third perspective understands learning as externally mediated through social interaction. The proponents of this perspective argue that socially mediated interaction, which can be affected by individual and cultural factors, constitutes the means for developing individual knowledge and learning (Shepard, 2006).
In criterion-referenced contexts, where attainment targets for the course/year are provided, the expected outcomes of learning are predefined. However, learners could presumably take different trajectories to achieve those outcomes. Hence, the ongoing learning that happens during the term is viewed as a non-linear process, specific outcomes of which may not be possible to (pre-) determine. Carr (2008) refers to such outcomes as ‘fuzzy’ (p. 37). This is consistent with the understanding oflearning as a dynamic process, not a static performance. However, it seems that the process of learning and performance are not necessarily mutually exclusive terms. For instance, Sadler (2007) considered confident, accurate and independent performance to be a way of demonstrating the outcomes of learning. The important feature of performance that demonstrates learning is its dynamic nature; it can be adapted and improved, hence demonstrating progress in learning. Dweck (2000) distinguished between learning and performance goals. She argued that the former focus on developing one’s own skills and understanding, while the latter focus on ‘winning positive judgements of your competence and avoiding negative ones’ (p. 15). Evidently, the understanding of the term performance by Dweck (ibid.) differs from that of Sadler (ibid.). The understanding of learning adopted in the current study encompasses the setting and achieving of learning goals as defined by Dweck (ibid.) and adopts Sadler’s (ibid.) interpretation that the process of learning can be demonstrated through performance. Language assessment that focuses solely on performance with reference to predetermined outcomes is referred to as convergent (Pryor & Crossouard, 2008). This can be contrasted with the type of assessment that promises to move learning forward, which is divergent (Pryor & Crossouard, 2008). It aims to discover what a learner knows and can do to subsequently use that insight to support learning. To sum up, it is argued here that learning can be demonstrated by performance and facilitated by setting learning goals that may be supported by assessment that is divergent in nature, although presumably constrained by the curriculum. The notion that learning could be facilitated by assessment has attracted considerable research attention, especially following the claims about the potential of AfL to raise achievement (Black & Wiliam, 1998). Black and Wiliam (ibid.) reviewed 250 articles about studies of formative assessment and concluded that it does improve learning. Their findings were criticised by Bennett (2011) and Dunn and Mulvenon (2009), who questioned the validity of comparing the diverse studies that were included in this metaanalysis or the methodological rigour of those studies. Nevertheless, the Black and Wiliam (ibid.) article inspired further research. The outcomes of that research are complex and although some views are that FA and AfL may benefit the learning process (e.g. Earl, 2012; Ruiz-Primo & Furlak, 2006; Sly, 1999) there is little empirical evidence to support such claims (Bennett, 2011; Dunn & Mulvenon, 2009). Another significant issue was also highlighted, namely that there is not a commonly shared definition of what exactly FA and AfL mean (definitional issue, Bennett, 2011). Bennett (ibid.) argued that without a well-established and clear theoretical framework, it is impossible to conduct research that would provide conclusive empirical evidence to support the claims proposing the positive impact of AfL on raising achievement, as suggested by Wiliam (2009) and others. This issue is further complicated by inconsistent uses of the terms FAand AfL. As these are of key importance to the present study, the next section discusses the distinction between AfL and FA.