Caudate nucleus
1.6 Long-term memory and language learning: a review of theoretical models 1 The declarative-procedural model
1.6.2 Other cognitive models
Skill Acquisition Theory. A model for language processing that was formulated independently of the declarative-procedural model but is related to it is the one outlined in DeKeyser's Skill Acquisition Theory (DeKeyser, 1995, 2007, 2015 for a general overview of skill acquisition theory see Anderson, 2007). In this framework DeKeyser does not mention the memory systems implicated but rather refers to declarative and procedural knowledge, although the two types of knowledge can be understood to denote the information processed and stored in the respective memory systems. DeKeyser considers mainly instructed contexts or situations where second languages are learnt explicitly, i.e. relying on the learners' declarative knowledge of rules and language regularities, or the ability to induce rules.
DeKeyser describes the process of attainment of fluent use of the language through three stages. In the first stage learners rely exclusively on declarative knowledge in language processing/use or may even know the rules of a language without attempting to put them into practice. In the second stage, that DeKeyser calls 'proceduralization', learners draw on declarative and, increasingly, on procedural knowledge as rules start to be practiced "acting on this knowledge, turning it into a behavior, turning knowledge that into knowledge how" (DeKeyser, 2007, p. 953). In the third stage, reached after ample opportunity for reiterated practice of the same or a very similar skill is provided, the practiced language skill gradually becomes fully automatized, i.e. tends to rely completely on procedural knowledge.
As DeKeyser points out it is important to clarify that "turning" declarative into procedural knowledge is a label to describe what is observed behaviorally. However, the
3
The italics are mine. The words in italics were in quotation marks in the original text.
nature of the process is more correctly captured by the idea that at automatized stages of language use both types of knowledge are available, although a shift from reliance on declarative knowledge to reliance on procedural knowledge has occurred. Note that in this model declarative knowledge not only supports proceduralized knowledge but "plays a causal role" in its development (DeKeyser, 2009, p. 126). Finally, similarly to the DP model, the Skill Acquisition model suggests an initial role of declarative processing followed by a more prominent role of procedural processing for increasing amounts of practice.
Usage-based approaches. Another model adopting a cognitive approach to language acquisition/learning is the one proposed in N. Ellis (1994, 2005) and N. Ellis and Wulff (2015). Similarly to the DP model this approach is underpinned by current neurophysiological evidence of the relationship between brain activation and cognition, but proposes a different account of the way language is processed, as well as of the role attention and declarative knowledge play in language acquisition/learning. According to N. Ellis, an individual's first contact with a novel linguistic object primarily involves perceptual priming. Although this type of priming is an unconscious form of learning, attention plays a crucial role already at this stage, because it organizes and unifies the representation of perceptual stimuli that will emerge as a pattern, after a sufficient number of exemplars of the linguistic object is processed at sensory-cortical level.
Involvement of declarative memory is then required to establish any relationships between the linguistic form and the meaning associated to it. The established form- meaning relationship is further primed in subsequent encounters/uses of the linguistic form and feeds into the implicit representation of the stimuli proceduralizing the construction. With other cognitive models (e.g., the DP models) N. Ellis's approach shares the idea that, in order to process language, learners employ domain-general
cognitive mechanisms, i.e., mechanisms that are not exclusive to language learning and processing. A main role for declarative memory in the initial stages of processing is another common point. However, it is important to note that the involvement of procedural memory in this account operates on constructions, i.e. sets of form-meaning pairings, whereby a construction can be any linguistic object ranging from lexis, to morphology, phrase structure or more complex syntactic and pragmatic entities.
In comparison, accounts like the DP model seem to place more focus on a separate role of procedural and declarative memory in processing of rule-based components of language vs. processing of meaning or vocabulary. Finally, N. Ellis's account does not emphasize a competitive relationship between the two memory systems. In his model, declarative memory plays a co-operative role in supporting procedural-memory based language acquisition and learning. Indeed, the main function of declarative memory, as emerges from N. Ellis's account, is to functionally enable and support proceduralized language learning.
Shallow Structure Hypothesis. The Shallow Structure Hypothesis (Clahsen & Felser, 2006; Clahsen & Felser, 2017) is mainly a model of language processing in comprehension. It distinguishes between L1 and L2 language processing proposing that whilst L1 processing consistently relies on morphosyntactic computation, L2 processing in comparison relies to various degrees on nongrammatical information, prioritizing semantics (Clahsen & Felser, 2017, p. 2-3). In a way that is reminiscent of the declarative/procedural distinction in the DP model, the Shallow Structure
Hypothesis suggests that both processing routes (morphology driven vs. heuristics driven) are available to L2 learners, and reliance on the first depends on the level of language proficiency (although the authors exclude that L2 processing can ever become completely native-like).
Recently, the interaction between age of L2 acquisition (age of onset) and language processing has started to be investigated in this theoretical framework under the hypothesis that the age at which the L2 began to be acquired is a predictor of type of L2 processing in adults (Clahsen & Veríssimo, 2016; Veríssimo, Heyer, Jacob, & Clahsen, 2017). In particular, these studies found that whilst derivational (lexically- mediated) priming was not related to age of onset, age of onset predicted priming of inflected forms that are not mediated by a lexeme (grammatical inflection).