Measures of declarative memory

Tasks that assess long-term declarative memory are further classified into tasks that tap visual or verbal aspects of information retention. In these tests participants are often exposed to visual or verbal stimuli and they are asked to discriminate or recall the stimuli immediately afterwards and at a delayed time. Relative to tests originally developed in the US or in the UK, a relatively recent battery that is widely used with adult populations is the Wechsler Memory Scales (WMS), a comprehensive set of tests that measures aspects of visual and verbal short- term memory and long-term declarative memory (Wechsler, 1945, latest revision 2009). It has the advantage to provide updated normed data based on a large sample (16-90 y.o.) in its original version and to be available in additional versions adapted for a number of different languages/countries. In the battery, aspects of verbal declarative memory are measured through a Paired Associates task and through a Logical Memory task. In the Paired Associates task participants are asked to memorize and recall pairs of related and unrelated words, whilst in the Logical Memory task participants listen to a short story and are asked to recall it immediately and after a delay.

The Doors and People Test (D&P; Baddeley, Emslie & Nimmo-Smith, 1994; latest revised version 2006) is a test developed in the UK that originally specifically measured long-term episodic memory in adults, tapping both visual and verbal components (Davis, Bradshow & Szabadi, 1999). In its latest version the range of normative data includes children from age 5 and adults, which makes it a particularly useful tool to investigate memory across age groups.

In a number of studies published in the last ten years (Carpenter, 2009; Morgan-Short, Faretta-Stutenberg, Brill-Schuetz, Carpenter, & Wong, 2014, among others) visual declarative memory in adults was also assessed with a computerized version of the Continuous Visual Memory Test (CVMT). The battery was developed by Trahan and Larrabee (1988) in the US, assesses non-verbal visual memory discrimination and recognition and was validated based on the performance of 92 healthy adults and 138 children.

Based on norms for the Italian adult population, a series of tasks along the lines of the ones available in the Wechsler's battery has been developed to assess declarative memory (Bianchi, 2015). Similarly to Wechsler's Logical Memory task, in the Short Story task (Raccontino "Anna Pesenti...") participants memorize and recall a story immediately and after a delay (Novelli, Papagno, Capitani, Laiacona, Cappa, & Vallar, 1986). In its most recent version (Mondini, Mapelli, Vestri, Arcara, & Bisiacchi, 2011) the test was standardized and normed based on a sample of 702 individuals (15-96 years old). Further tests that assess verbal declarative memory include versions of the Paired Associates Test (Zappalà et al., 1995), or tests that require the memorization of lists of words (for a complete review see Dai Pra et al., 2015).

For visual declarative memory Italian tests based on large normative samples include the Rey-Osterrieth Complex Figure Test (Caffarra, Vezzadini, Dieci, Zonato, Venneri, 2002; Carlesimo et al., 2002) and the Modified Taylor Figure Test (Casarotti, Papagno, & Zarino, 2014). In both tests participants are asked to copy a complex abstract figure including a number of elements and to redraw the figure after a delay as precisely as they can recall it.

Materials for the tests of verbal and visual declarative memory used with children are specifically developed for this age group to cater for the age-dependent differences in cognitive development. The Children Memory Scale (CMS; Cohen, 1997) is a comprehensive measure of learning for children aged 5-16, including a total of 9 subtests measuring three domains: Auditory/verbal memory, visual/non-verbal memory and attention/concentration. In particular, recent studies investigating declarative memory in SLI and typically developing child populations (Lum, Gelgic, & Conti-Ramsden, 2010; Conti-Ramsden, Ullman, & Lum, 2015), have deployed two tests from the auditory/verbal component of the CMS. These tests are 'Stories', where children recall events and details of an orally presented story immediately and after a delay, and 'Word Pairs', which tests the ability to recall word pairs over three learning trials immediately and after a delay.

The currently most comprehensive battery to test long-term memory in children for the Italian population is the PROMEA battery (Prove di Memoria e Apprendimento per l'Età Evolutiva [Developmental Memory and Learning Tests]) developed by Vicari (2007). The battery was validated on a sample of 709 Italian children from 5 to 11 years of age and includes tests of verbal and visual aspects of declarative memory, priming and working memory. Verbal declarative memory is assessed with a word retention task and a short story task similar to the one adopted in the CMS battery. Visual/spatial declarative memory is assessed with a picture discrimination task and a spatial learning task, where children memorize and recall the position of pictures of objects on a four-space matrix (see Chapter 6 for details). 1.7.2 Measures of nondeclarative memory and procedural learning ability

The Serial Reaction Times Task (SRT) was developed specifically to tap the implicit learning of new associations and sequence learning (Niessen & Bullemer,

1987). In the original version of the paradigm participants are asked to react as fast as they can to a visual stimulus appearing in different locations on a computer screen by pressing corresponding keys on a keyboard. A random block of trials (baseline) is followed by a series of blocks where a fixed sequence of 10 trials is presented, followed by a final random block. A decrease in reaction times (RTs) in fixed- sequence blocks compared to random presentations is expected if learning of the sequence has occurred by the end of training. As the paradigm established itself in the study of implicit sequence learning it became clear that some participants were able to, at least partially, learn the stimuli sequence explicitly (as evidenced by the fact that they were able to recall parts of the sequence after the experiment).

In order to control for this confound new versions of the paradigm were developed where the detection of the implicit sequence through explicit learning strategies was made more difficult, for example by alternating random blocks with sequence blocks (Meulemans, Van der Linden, & Perruchet, 1998) or by presenting sequence patterns that included random trials (Alternating Serial Reaction Task, ASRT; Howard & Howard, 1997).

Compared to similar tasks, the ASRT in particular has provided a paradigm that is more reliable in filtering out learning effects due to declarative strategies (Song, Howard, & Howard, 2007). In this task the repeated sequence is "hidden" and alternates with a series of random events, so that in each sequence of 8 items a fixed item is followed by a random item, for instance following the pattern 1r2r4r3r. According to Hedenius (2013) the ASRT presents at least two additional advantages compared to the original paradigm. First it allows a clearer separation of general motor skill learning from sequence learning in repeating sequences, due to the possibility of comparing RT decreases relative to the complete 8-item sequence with

the decreases relative to the hidden 4-item sequence. Secondly, it allows continuous assessment of RTs as they progressively decrease, with an on-going comparison of RTs in random and nonrandom parts of the sequence across blocks (Hedenius, 2013, p. 44).

Variations of the SRT task have been used extensively in the literature to detect implicit learning (Barker, 2012; Jackson, Jackson, Harrison, Henderson, & Kennard, 1995; Knopman & Niessen, 1987; Robertson, 2007; Smith, Siegert, McDowall & Abernethy, 2001), as well as a behavioral measure of procedural memory functioning in both adults and children (Brill-Schuetz & Morgan-Short, 2014; Ferraro, Balota & Connor, 1993; Hedenius, 2013; Janacsek, Fiser, & Nemeth, 2012; Lum et al., 2010, among others). In some studies, measures of motor control (in the case of Lum et. al., 2010 the MOT, from the CANTAB battery) were administered alongside SRT tasks in order to filter out the effect of motor-control in the evaluation of procedural memory scores.

Another family of tests widely used in the investigation of procedural learning ability are probabilistic tasks such as the Weather Prediction Task (WPT; for the use of this task in studies investigating adult L2 learning see for example Carpenter, 2008; Morgan-Short et al., 2014). In this task participants are asked to guess a weather outcome (sun or rain) based on the presentation of a series of cues. After each trial the participants are given feedback (correct/incorrect). Unbeknownst to them, each cue is assigned a fixed probability to give rise to one or the other weather outcome, so that increasing accuracy in the prediction constitutes evidence of learning of the underlying implicit probability pattern (for a probabilistic tasks adapted for use with children, see Mayor-Dubois, Zesiger, Van der Linden, & Roulet-Perez, 2016). Also in the case of probabilistic tasks, studies have found that the use of explicit strategies

may represent a potential confound, included in studies with children. Mayor-Dubois and colleagues for example found that performance in the second part of a probabilistic task was related to the deployment of explicit learning strategies, and increasingly so in older children compared to younger children (Mayor-Dubois et al., 2016, p. 729). In Morgan-Short et al. (2014) and other studies, scores obtained assessing parts of the Tower of London task (TOL) have also been used to calculate composite scores of procedural learning ability, although the task itself has been deployed to assess planning and executive function.

In terms of their neuroanatomical correlates, a number of clinical and neuroimaging studies have shown that both motor/cognitive sequence learning and probabilistic tasks implicate the involvement of striatal and cortical areas in the brain. Specifically, sequence motor learning (SRT tasks) has been associated to the activation of the putamen, and cognitive learning (probabilistic tasks) has been associated to the activation of the caudate nuclei, (for a review of studies see Mayor- Dubois et al., 2016).

The first section of this chapter has presented a currently widely adopted dual model of the architecture of long-term memory, as well as our current understanding of the localization of long-term memory functions in the human brain according to this model. This initial neuropsychological background served to introduce a number of (neuro)cognitive models of (second) language acquisition that assume or are compatible with a dual representation of long-term memory (declarative and procedural memory). Finally, in order to set the basis for the methodological choices made in the experimental studies, the third section of the chapter was devoted to a detailed review of normed behavioral batteries and tasks that have been used as

measures of declarative and procedural learning ability in SLA studies and in the neuropsychological literature.

2. Language learning processes, language instruction and the nature