The Latin Project

Abstracts

Paper #1. The Latin Project: Goals and methodology

The goal of The Latin Project is to investigate the relationship among internal (bilingualism, WM capacity) and external (degrees of explicitness in the input) variables in the acquisition of non-primary languages. This first part of the colloquium provides an overview of the motivation and methodology and will serve as an introduction to the papers and discussion that will follow.

The design of the study is experimental and involves 180 L2 learners of Spanish at three different levels of experience. The levels examined are beginners, advanced (majors with study abroad experience) and native-like learners of L2 Spanish. Participants interact with a computer-delivered lesson on Latin's case system and word order that actively engages them in processing meaningful oral and written input, at no point requiring them to produce language. At last year's SLRF, we reported results from the first batch of data, which looked at the most explicit of the treatments, T1 in Figure 1 below. This year, we will present results from the second batch, T2 in Figure 1 below, as well as the interactions. We also incoporate new analyses with new vs. old items, production data (both accuracy and preference).

Figure 1

1 +G+EF+T	2 -G+EF+T	3 +G+IF+T	4 -G+IF+T	5 +G-F+T	6 -G-F+T	7 control	8 –G-F-T
Vocab lesson & quiz	Vocab lesson & quiz	Vocab lesson & quiz	Vocab lesson & quiz	Vocab lesson & quiz	Vocab lesson & quiz	Vocab lesson & quiz	Vocab lesson & quiz
Grammar section	-	Grammar section	-	Grammar section	-	-	-
+EF practice	+EF practice	+IF practice	+IF practice	-F practice	-F practice	-	-F exposure
Language test A B or C	Language test A, B or C	Language test A, B or C	Language test A, B or C	Language test A, B or C	Language test A, B or C	Language test A, B or C	Language test A, B or C

G = grammar explanation EF = explicit feedback IF = implicit feedback T = task-essential practice

These are the RQs guiding the study:

Main factors

1. Do external variables affect L3 acquisition? Can learners use explicit information about the language?

2. Do internal variables such as cognitive capacity affect L3 acquisition? Specifically, does WM span affect the acquisition of an L3?

3. Does bilingualism affect L3 acquisition? Does previous language experience enhance the acquisition of an L3? Specifically, does Spanish-English bilingualism affect L3 acquisition of Latin?

Interactions

4. Do learners with different WM capacity use explicit and implicit evidence differently?

5. Does bilingualism enhance WM capacity? Is there a threshold for those effects?

6. Do learners at different stages of bilingualism use explicit and implicit evidence differently?

7. In responding to the previous question, can differences be explained in terms of differences in WM capacity?

Paper #2. Level of bilingualism and explicitness in the input: Do they relate to L3 acquisition?

Several empirical studies show that prior language experience facilitates L3 acquisition positively in bilingual contexts (i.e., Cenoz & Valencia, 1994; Sanz, 2000; Swain, Lapkin, Rowen, & Hart, 1990). In addition, previous studies that have looked at level of bilingualism as a variable have shown that level of bilingualism is closely related to level of proficiency in a third language (Cenoz, 2001; Lasagabaster, 2000; Muñoz, 2000; Sagasta, 2003). The present study sought to contribute to this line of research by comparing the acquisition of a third language (Latin) by intermediate, advanced and native-like speakers of L2 Spanish exposed to two different explicit treatments to learn to assign semantic functions to noun phrases at the sentence level in Latin.

The advantage of high proficiency levels in L1 and L2 to acquire an L3 follows the pattern predicted by Cummins’s threshold hypothesis (1979) and the common underlying proficiency model (1989). The threshold hypothesis proposes the existence of two thresholds, an upper threshold and a lower threshold. Cognitive advantages should emerge only when a specific level of bilingual proficiency is attained, reaching the second threshold. One of the goals of our study was to determine where, in the continuum towards balanced bilingualism, can the two thresholds be placed.

A group of participants was exposed to a treatment which included grammar explanations, task-essential practice and explicit feedback. Another group of participants was exposed to a less explicit treatment which included only task-essential practice with explicit feedback (i.e., no preemptive grammar explanation). L3 acquisition was measured with a written interpretation task, an aural interpretation task, a grammaticality judgment task and a production task. The results obtained support previous studies that have found positive effects for partial bilingualism (Bialystok, 1988; Yelland, G., Pollard, J., & Mercuri, A., 1993). In addition, the findings suggest that the threshold under which advantages start to emerge depends on factors such as type of treatment and type of L3 measurement. These claims are in line with previous studies that found that the bilingual advantage emerges under conditions that require higher levels of attentional control to process relevant features of the input (Bialystok, 1988, 1999, 2004; Bialystok & Martin, 2003; Nation & McLaughlin, 1986).

Paper #3. WM capacity and explicitness in the input: Do they affect L3 acquisition?

In the current study, we investigated the relationship between working memory capacity and explicitness in the input on L3 acquisition. Researchers agree that working memory (WM) – a system that is concerned with both active processing and transient storage of information (Eysenck & Keane, 2000: 164) – may constrain noticing, which is crucial for L2 learning (Schmidt, 1995 and elsewhere). In addition, it is argued that attentional control and conscious awareness occurs in working memory (see Baddeley, 2000). The efficiency of language learners to acquire a subsequent language may be related to WM capacity, however the findings have not been uniform. Some research has shown that greater WM capacity is favorable for L2 performance (Harrington & Sawyer, 1992; Miyake & Friedman, 1998) and development (Mackey, Philp, Fujii, Egi, & Tatsumi, 2002). However, other research has found no effects, neither positive nor negative, of WM on L2 development (Sagarra, 2000).

Although there may be a relationship between WM and language learning, no studies, except for one, have looked at how L2 participants with different WM capacity perform regarding the input received. Mackey et al. (2002) suggest that learners with higher WM capacity reported more noticing than those with lower WM capacity, which may be beneficial for L2 learning. However, although they looked at the input and the effect of WM capacity, they only focused on one condition—implicit feedback. Different from their study, the current study investigates, with L3 learners, how participants with different levels of WM capacity perform under two types of input conditions—more explicit [+Explicit] and less explicit [-Explicit].

In the current study, bilingual participants were exposed to an L3 under two different conditions ([+/-Explicit]). Their WM capacity was determined by taking a composite of the sentence span and computational digit span test's z-scores, which was then separated into two WM levels [+/-High]. Preliminary analysis of the two conditions show different patterns, leading us to conceive that WM capacity affects L3 acquisition, but the effect, if any, depends on the task, test item (old or new exemplars), and explicitness in the input. These results partially support Harrington & Sawyer (1992), Miyake & Friedman (1998), and Mackey et al. (2002), who found that greater WM capacity is related to more accurate performance or more efficient development. Further analysis will be carried out to further elucidate the nature of said effect.

Paper #4. The Younger the Better? An “Age” Old Question in the Acquisition of Non-primary Language

As we continue to work toward identifying the characteristics of the good language learner, age of onset of language learning is an essential consideration since the traits that characterize a good learner at one age are likely not the same traits that characterize a good learner at another age. While children may enjoy advantages of their cognitive simplicity when it comes to language learning, adults may be able to use the advanced cognitive capabilities of a mature brain to their benefit. Studying third language (L3) acquisition affords the opportunity to examine the age issue from two angles, age of onset of second language (L2) learning and age of onset of L3 learning. Does the age at which one starts to learn a second language set the standard for all subsequent language learning, such that the younger L2 learner will be more successful than an older L2 learner when they are both faced with the task of learning an L3 at the same age? On the other hand, and regardless of age of onset of L2 acquisition, will the wisdom that comes with age prove to be an advantage for older L3 learners or will the effects of cognitive aging prove detrimental to L3 learning success?

This paper, which is an outgrowth of a larger dissertation research project, reports the results of an empirical study that utilized the materials of the Latin Project to investigate how the age variable influences bilingual adults’ ability to acquire an L3. Specifically, the study investigated the relationship between age of onset of L2 and L3 acquisition and success in initial L3 acquisition among US Latino adults. Participants were grouped according to two criteria, level of Spanish-English bilingualism and age of onset of L2 (English) learning. Native Spanish speakers with an intermediate level of English proficiency were subdivided into groups that had begun learning English as children (ages 6-10), adolescents (ages 11-15) and post-adolescents (ages 16-20). A group of native Spanish speakers with native-like English abilities was similarly subdivided according to age of onset of L2 learning. A third grouping of participants included Spanish-English bilinguals who had begun learning both languages simultaneously from birth.

Results are explained within the context of the critical period hypothesis and the Fundamental Difference Hypothesis and interactions among ages of onset, level of bilingualism and cognitive capacity are discussed.

Paper #5. Levels of awareness and L3 learning: A think-aloud protocol analysis

A growing number of studies have employed concurrent think-aloud protocols in an attempt to investigate the role of awareness in language learning through direct measures of learners' attentional processes. In general, they have found that learners who verbalize awareness during L2 tasks demonstrate better post-test performance than those who do not, and furthermore, that higher levels of awareness are associated with more learning (e.g., Leow, 1997, 2000, 2001; Rosa & Leow, 2004; Rosa & O'Neill, 1999). Separately, research has found bilinguals to be more metalinguistically aware than monolinguals (e.g., Bialystok, 1986, 1988) and better at acquiring a third language (e.g., Cenoz & Valencia, 1994; Sanz, 2000; Swain, Lapkin, Rowan, & Hart, 1990). However, questions remain regarding the sources of this bilingual advantage.

The current study, part of a larger project, examined the relationships among levels of bilingualism, awareness of linguistic input, and post-treatment performance during the initial stages of learning Latin as a third language. The 14 participants whose data were analyzed for this portion of the study included near-native adult English-Spanish bilinguals and L1-English university students who were concurrently studying Spanish at the intermediate and advanced levels. They were exposed to computer-based Latin practice with explicit feedback, performed think-alouds during the treatment, and then completed post-tests of written interpretation, aural interpretation, grammaticality judgments, and written production.

Three verbalization coding systems were used to gain insights into the learners' internal thought processes. One, based on Leow (1997), examined levels of awareness; another attempted to tap into the learners' cognitive comparisons between their current state of Latin knowledge and the target language input; and the third explored their attention to form and meaning, both separately and together. The results supported past research in indicating strong, statistically significant positive correlations between frequency of verbalizing awareness at any level and performance on some of the post-tests. The frequency of verbalizing awareness at the level of understanding was also positively related to post-test performances. However, on some tests, moderate to strong negative correlations were found with verbalizing awareness at the lower level of noticing. The multiple methods of coding the think-alouds were useful in providing more fine-grained information about which verbalization tendencies were related to superior performance on which of the various post-tests. Higher frequencies of metalinguistic verbalization were also found to be associated with higher levels of bilingualism, consistent with research showing a bilingual advantage in metalinguistic awareness.

Paper #6. Reactivity: Evidence from L3 acquisition.

This paper investigated the results of employing think aloud protocols for a treatment on a group of bilinguals learning an L3. A number of studies have shown that bilinguals use a variety of processing strategies as compared with monolinguals (e.g., McLaughlin, 1986; Nayak, Hansen, Kreuger, & McLaughlin,1990; Ramsay, 1980). In addition, it has also been suggested that bilingual subjects learning a third language present not only highly developed metalinguistic awareness, cognitive flexibility, and automatization of processes than monolinguals, (Bialystok, 1986, 1988;Bialystok, Craik, Klein, & Viswanathan, 2004; Bialystok & Majumder, 1998, Cromdal, 1999) but also similar levels of creativity (Lasagabaster, 2000). In order to investigate bilinguals’ internal processing, one option is to utilize verbal reports. However, while verbal reports have been widely used in various fields, there have been concerns about its employment with respect to its validity and reliability (e.g., Russo, Johnson, & Stephens, 1989). Hence, the current study will focus on this methodological concern as an attempt to shed some light into the issue of whether thinking aloud can alter cognitive processes that may affect non-primary language learning.

Reactivity was investigated for a group of bilinguals with different levels of proficiency who were learning Latin as their L3 through two different computerized treatments- a grammar explanation and practice with explicit feedback, or only practice with explicit feedback. The effects of reactivity were measured with a written and aural interpretation task, a grammaticality judgment task, and a written production task. Results from the first treatment showed that verbalization did not induce reactivity. However, a positive trend was found in the written interpretation task. Interestingly, the second one showed evidence of reactivity in the aural interpretation and the written production tests. A positive trend was also found in the grammaticality judgment task and the written production task. In contrast to previous studies (e.g., Ericsson & Simon, 1984/1993; Leow & Morgan-Short, 2004; Bowles & Leow, in press), it is argued that verbalization may induce reactivity. The results also suggest that not all types of tasks induce reactivity (e.g., Russo et al., 1989; Stratman & Hamp-Lyons, 1994; William & Davis, 1997). Next, the positive reactive effects show that verbalization may enhance learning. Lastly, the results indicate that verbalization has a different effect on performance, depending on the type of test item. It is posited that familiarity with test items, the number of test items in a test, and working memory capacity may influence the results. In conclusion, the present study cautions researchers over the employment of thinking aloud while investigating bilinguals’ internal cognitive processes.

Abstracts

Paper #1: Acquisition of an L3: Age, Working Memory and Level of Bilingualism

Cristina Sanz, Georgetown University

Contrary to widespread belief among monolingual populations everywhere including the US, research carried out to examine the effects of growing up bilingual has suggested that bilingualism, and, in particular, biliterate bilingualism is related to cognitive efficiency in the acquisition of new knowledge. Specifically, a number of factors involved in language acquisition may be enhanced as a result of the language learning experience of bilinguals.

Such factors have been posited to explain the advantage shown by bilinguals over monolinguals in the acquisition of an L3 both in classroom (Cenoz and Valencia (1994), Sanz 2000) and in laboratory studies (McLaughlin & Nayak, 1989; Nation & McLaughlin, 1986; Nayak, Hansen, Krueger & McLaughlin, 1990). Their comparisons of experienced learners and novices show that experience affects the way in which language is processed. Similarly, Klein's (1995) multilingual subjects showed heightened ability to identify and retain key verbs necessary to trigger the parameter resetting for preposition stranding, the key form to be acquired, suggesting that they are able to pay closer attention to relevant features in the input. Noticing has been posited to determine the amount and quality of linguistic information extracted from the input (Schmidt 1995). According to Robinson (1995), noticing must include not only detection, but rehearsal in short term memory. It is not surprising, then, that a number of scholars in the field are working to investigate the role of working memory (WM) in language acquisition (Ellis 2002). It is yet unclear whether WM is fixed but it is intuitively attractive to think that perhaps experienced language learners have increased WM spans, a cognitive advantage that could explain their superior ability to learn languages.

The present study focuses on the use that beginning, advanced and near-native speakers of L2 Spanish make of explicit and implicit input in the acquisition of a new language. The design is experimental and involves 180 participants who interact with a computer-delivered lesson on Latin's case system and word order. Results from the WM tests are analyzed in relation to linguistic data (accuracy & RT) and learners' language experience in order to elucidate the potential contribution of each factor. Think-aloud protocols, debriefing, and language background questionnaires are also included.

The panel includes four papers: The first lays out the details of the project, the second focuses on working memory, the third focuses on age of L2 acquisition, and the fourth focuses on results from the think alouds.

Summary:
Does cognition benefit from bilingualism, and under what conditions? The panel reports on a series of studies investigating the interaction between two internal variables, namely prior experience (bilingualism) and cognitive capacity (WM, PSTM), and type of input, an external variable

Paper #2: The younger the better? Age of onset of L2 acquisition and L3 learning

Catherine A. Stafford

Language learning appears to be maturationally constrained, and the nature of these constraints is a much-debated issue in first and second language acquisition research. The Critical Period Hypothesis (CPH) suggests that there exists a defined period during which achieving nativelike levels of language proficiency is assured and after which it is not possible. A likely state of affairs is that as age increases, learners' ability to acquire language declines.

The proposed laboratory study examines the interaction among age, cognitive capacity (WM and PSTM capacity) and prior experience. Specifically, the proposed study investigates the effects of these individual variables on participants' ability to acquire rules of Latin morphosyntax.

Three groups of participants formed on the basis of level of Spanish-English bilingualism and age of onset of L2 English acquisition will complete a computerized lesson in Latin sentence processing. The participant groups will be comprised of a. late intermediate L2 English learners (the Intermediate group), b. late L2 English bilinguals at the end state of their L2 development (the End State group) and c. early simultaneous Spanish-English bilinguals (the Simultaneous Bilingual group). All participants will be naïve learners of Latin. Performance on immediate and delayed post-tests (accuracy as well as response times) will be correlated with measures of working memory and phonological short-term memory. In addition, think aloud protocols (TAPs) will be collected and analyzed as a means of exploring the on-line processing of participants in each of the groups as well as the degrees of metalinguistic awareness and attentional control that they exhibit during task performance.

The study makes the following predictions:

1. Individuals with more language experience (i.e., simultaneous bilinguals and adult L2 learners at the end state) will show higher ability to interpret, produce and judge the well-formedness of Latin sentences than individuals with less language experience (i.e., adult intermediate L2 learners);

2. When language experience is held constant, participants with high cognitive capacity will exhibit superior ability to interpret, produce and judge the well-formedness of Latin sentences.

3. When language experience and cognitive capacity are held constant, as the age of onset of L2 acquisition increases, the ability to interpret, produce and judge the well-formedness of Latin sentences will decline.

4. More language experience or higher cognitive capacity may compensate for the negative effect of a later onset of L2 acquisition on the ability to interpret, produce and judge the well-formedness of Latin sentences.

Summary: Three groups of US Latinos' acquisition of L3 Latin is compared (Late intermediate L2 English learners, late L2 English bilinguals at the end state of their L2 development and early simultaneous Spanish-English bilinguals) to investigate maturational constrains in L3 acquisition.

Paper #3: Is performance affected by concurrent think alouds? Evidence from L3

Beatriz Lado, Hui-Ju Lin, Cristina Sanz

Concurrent think alouds have advantages but they also have limitations. One question that is raised is whether or not this method could affect performance. This effect is known as 'reactivity' and has mainly been studied in cognitive research.

Cognitive psychology studies show evidence to support Ericsson and Simon's (1984, 1993) claim that thinking aloud causes reactivity in time (thinking aloud makes participants spend more time completing the tasks). Although there is evidence to support that performance-- either during the actual completion of the activity or in post and delayed tasks--is not necessarily affected by thinking aloud (Karpf, 1972; Johnson & Russo, 1978; Biggs et al., 1993; Brinkman, 1993), some studies showed that the type of task (Russo et al., 1989; Stratman & Hamp-Lyons, 1994; William & Davis, 1997) or the type of instructions asked (Meissner et al., 2001) may influence the presence or lack of reactivity. Studies showing positive reactivity (Gagné & Smith, 1972; Baumann et al, 1993; Baumann et al, 1992 and Ahlum-Heath & Di Vesta, 1986) revealed the need to provide more support for this matter with other empirical studies.

Traditionally, think alouds have been implemented in SLA to study learning strategies (Warren, 1996; Carrell, 1989) including differences in strategies used between L1 and L2 learners (Davis & Bistodeau, 1993). Think alouds have also been used to operationalize attention and awareness in an attempt to investigate the role of these constructs in the acquisition of a second language (Leow 1999, 2000, 2001; Rosa, 1999; Rosa & Leow, 2004; Rosa & O'Neill, 1999). Very little has been done to validate the use of this methodology. Does thinking aloud while exposed to a pedagogical treatment enhance or reduce the effects of the treatment?

The present study expands on previous research on SLA (Morgan-Short, 2004; Bowles & Leow, in progress) that found thinking aloud to be non-reactive for L2 Spanish learners' comprehension, intake, and written production of a morphological or a syntactic target structure. The present study investigates reactivity in think alouds of English L1/Spanish L2 participants exposed to a treatment to learn Latin as an L3. Acquisition of Latin case and word order was measured in written interpretation, aural interpretation, grammaticality judgment, and production pre, post and delayed tests.

Results showed that thinking aloud during a treatment does not hinder L3 acquisition under the conditions proposed. Furthermore, since participants produced mainly metalinguistic think alouds, results can be extended to this type of verbalization.

Summary:

A comparison of two bilingual groups (+ Think aloud/- think aloud) exposed to a treatment to learn L3 Latin showed that thinking aloud during a pedagogical treatment does not hinder L3 acquisition. Since participants produced mainly metalinguistic think alouds, results can be extended to this type of verbalization.

Paper #4: Measuring working memory capacity and its relation to

bilingualism and L3 acquisition

Mariona Anfruns, Beatriz Lado, Hui-Ju Lin, & Almitra Medina

Studies on the acquisition of a third language (L3) (Lasagabaster, 2000; Muñoz, 2000; Sagasta, 2003) have shown that a high level of proficiency in a second language (L2) enhances the acquisition of a third. Sanz (200) proposes a number of factors to explain those effects: intelligence (Peal & Lambert, 1962), metalinguistic awareness (Ben-Zeev,1977; Bialystok, 1991), cognitive flexibility and processing mechanisms (McLaughlin & Nayak, 1989; Nation & McLaughlin, 1986; Nayak, Hansen, Krueger, & McLaughlin, 1990), and cognitive capacity (Miyake and Friedman, 1998) .

The paper reports on an experimental study on the relationship between cognitive capacity operationalized as working memory (WM) capacity and L3 acquisition by English L1/Spanish L2 bilinguals. WM is defined as the ability to simultaneously process and store linguistic information (Miyake and Shah, 1999). Some studies suggest that greater WM capacity is favorable for L2 performance or development (Harrington and Sawyer, 1992; Miyake and Friedman, 1998; Mackey, Philp, Egi, Fujii, and Tatsumi, 2002; Newport, 1990), other suggest that WM has not effects on L2 development (Sawyer, 2000; Juff, 2003). Another issue is whether bilingualism affects cognition, specifically whether knowing a second language affects working memory capacity, and if those effects depend on the degree and type of bilingualism. Abu-Rabia & Siegel (2002) suggest no relationship between bilingualism and WM, while Ransdell, Arecco & Levy (2001) and Da Fontoura & Siegel (1995) found positive effects for bilingualism in working Memory.

WM has been measured by means of comprehension tests. Such tests present a complex task involving storage, retrieval of verbal material, and also comprehension. Their limitations lie in that they are language dependent, and without a recall component, it is measuring processing, not storage (Waters & Caplan, 1996). WM has also been measured by means of digit span tests, both backward and forward. They are language independent, but require little simultaneous storage and processing (Baddeley, 1986). Computational digit tests, also language independent, are complex cognitive tasks requiring multiple steps of processing, their limitations are not evident. Finally, word span tests can measure sort-term storage, but only measure storage, not processing (Waters & Caplan, 1996). Which test is appropriate, then? We tested 1) a comprehension test with a recall component, 2) a computational span test, 3) a digits reversed span test.

Written and aural interpretation and grammaticality judgment data from 31 participants revealed no significant correlations between WM and success of L3 acquisition, nor between WM and level of bilingualism. Interpretation and GJ data (N=31) revealed no significant correlations between WM and L3 acquisition, nor between WM and level of bilingualism. WM was measured by means of a comprehension test with a recall component, a computational span test, and a digits reversed span test.

Main factors

Interactions

Catherine A. Stafford