Understanding how explicit information (EI) and practice shape second language development remains an active goal of instructed SLA research. Evidence indicates that, in some cases, EI benefitted learning very little (Sanz & Morgan-Short, Reference Sanz and Morgan-Short2004; Stafford, Bowden, & Sanz, Reference Stafford, Bowden and Sanz2012; VanPatten & Oikkenon, Reference VanPatten and Oikkenon1996). In other cases, EI accelerates the point at which learners begin to interpret the input more accurately, increasing the efficiency of practice (Fernández, Reference Fernández2008; Henry, Culman, & VanPatten, Reference Henry, Culman and VanPatten2009; VanPatten & Borst, Reference VanPatten and Borst2012; VanPatten, Collopy, Price, Borst, & Qualin, Reference VanPatten, Collopy, Price, Borst and Qualin2013). Research has focused on the roles played by EI and practice in L2 learning, especially the effects of different types of EI and/or practice (DeKeyser & Prieto Botana, 2015), such as EI about L1 and L2 properties for crosslinguistically problematic target features (Spada, Lightbown, & White, Reference Spada, Lightbown, White, Housen and Pierrard2005).
To this end, McManus and Marsden (Reference McManus and Marsden2017) examined whether providing explicit instruction about L1 form-meaning connections would benefit the L2 learning of the French Imparfait, a late-acquired form by English-speaking learners because of complex L1-L2 form-meaning mapping differences (Howard, Reference Howard, Labeau and Larrivée2005; Izquierdo & Collins, Reference Izquierdo and Collins2008; McManus, Reference McManus2013, Reference McManus2015).Footnote 1 In a 12-week Pretest, Posttest, Delayed Posttest design, 50 English-speaking learners of L2 French were assigned to two treatments (L2-only and L2+L1) and a test-only control. The L2-only group received a core treatment of EI about the L2 and L2 task-essential practice. The L2+L1 group received the exact same core L2 treatment (as received by L2-only) as well as EI about the L1 and L1 task-essential practice. EI was delivered (a) prepractice (approximately five minutes) and (b) during the task-essential practice following incorrect answers. The prepractice EI used video, image, and sound to depict conceptual-semantic information (e.g., an ongoing event of a man eating an apple), followed by aural and written forms and information about how to interpret their meaning(s), that is, information about form-meaning connections. EI and practice were delivered using laptops (for a full description of the instruction, see McManus and Marsden, Reference McManus and Marsden2017, and IRIS for materials). Pretest, Posttest, and Delayed Posttest measures were online (self-paced reading, SPR) and offline (context-sentence matching tests, CMT, in listening and reading), measuring sensitivity to the Imparfait’s ongoing and habitual meanings, in which a French stimulus either matched or mismatched the ongoing/habitual meaning of a preceding event description in English (see [3] for example stimuli). Results showed no significant improvement over time for the Control group in speed and accuracy of interpreting the Imparfait. The L2-only group’s gains were mostly in offline tests at Post (not Delayed) in matched trials, with negligible effects in mismatched trials. Improvement up to Delayed was found in CMT-Read for the Imparfait’s ongoing meaning (matched trials). Improvement up to Post was found for CMT-Listen (habitual, matched) and SPR (ongoing, matched), but these were lost by Delayed. The L2+L1 group, in contrast, improved in all measures up to Delayed (six weeks after the intervention), in both offline and online outcome measures, in matched and mismatched trials. Critically, SPR results demonstrated sensitivity to aspectual anomalies, as evidenced by slower reading times (RTs) in mismatched trials for only the L2+L1 group.
While some previous research has provided EI about the L1 (but without L1 practice), with mixed learning outcomes (González, Reference González2008; Horst, White, & Bell, Reference Horst, White and Bell2010; Spada et al., Reference Spada, Lightbown, White, Housen and Pierrard2005), McManus and Marsden (Reference McManus and Marsden2017) is understood to be the first to have employed L1 task-essential practice. It is possible that L1 task-essential practice contributed to the online and offline benefits observed because the effects of L1 EI alone remain inconclusive. However, the extent to which L1 practice without L1 EI would result in learning benefits similar to those observed in the original study remains unclear. The present study contributes to that gap by means of a partial replication using the exact same procedure and outcome measures as in the original study. Similarly to McManus and Marsden (Reference McManus and Marsden2017), our intervention provided L2 EI plus L2 practice. It additionally included L1 task-essential practice, like McManus and Marsden’s (Reference McManus and Marsden2017) L2+L1 group, but, unlike that group, it did not include any EI about the L1. Critically, this design allowed us to investigate whether providing L1 task-essential practice without L1 EI would result in the same learning gains as found for L1 EI plus L1 task-essential practice.
MOTIVATION FOR THE PRESENT STUDY
A small number of studies have provided EI about both L2 and L1, finding some offline benefits for learning L2 morphosyntax. González (Reference González2008) compared L2+L1 EI with L2-only EI for lexical terminativity. L1 EI described how terminativity operates in Dutch (L1). Participants used that EI to construct rules about the (un)grammaticality of Dutch sentences. For the L2 EI, participants first described what they knew about Spanish past tenses before receiving EI about terminativity and perfectivity in Spanish (L2). That EI was then used to construct rules about the (un)grammaticality of Spanish sentences. Pre-Post cloze test results showed that L2+L1 EI was more beneficial than L2-only EI. However, interpretation of the results is inconclusive because the test (a written composition, identical at Pretest and Posttest) did not test knowledge types other than those taught during the intervention (largely controlled, offline, explicit) without Delayed Posttests. Spada et al.’s (Reference Spada, Lightbown, White, Housen and Pierrard2005) L1-L2 contrastive EI explicitly directed learners’ attention to L1-L2 differences and similarities. The EI was complemented with whole-class games and activities in the L2 (but not in L1) that required learners to produce the target feature. Data from a range of task types (passage correction, oral production, judgments) indicated an advantage for L2+L1 EI. However, further investigation is required because between-group differences at Pretest may have affected the results and, importantly, L1 EI was not used systematically or to the same extent by all the teachers in the study, and no L1 practice was provided. In sum, previous research providing L1 EI has indicated potential advantages. Thus, the extent to which the learning gains observed by McManus and Marsden (Reference McManus and Marsden2017) were due to providing L1 EI, L1 practice, or their combination remains unclear.
An additional motivation for the current study relates to the nature of the outcome measures to date. (Although see Andringa & Curcic, Reference Andringa and Curcic2015; Dracos, Reference Dracos2013; Henry, Reference Henry2015; Marsden, Williams, & Lui, Reference Marsden, Williams and Liu2013, using eye-tracking, SPR, and cross-modal priming respectively; but these studies did not incorporate L1 information or practice.) That is, research into L1 EI has not measured instructional effects on online processing, which is particularly relevant given that many interventions have aimed to influence input processing.
In terms of potential learning mechanisms at play, McManus and Marsden (Reference McManus and Marsden2017) argued that the L1 EI may have helped establish explicit knowledge about L1 form-meaning mappings, specifically that in English perfectivity and habituality can be expressed by the same morpheme (-ed), but ongoingness is expressed by a different morpheme (-ing). However, French form-meaning mappings are different: Habituality and ongoingness can be expressed by the same morpheme (-ait), but perfectivity is expressed by a different morpheme. As the task-essential practice required learners to make explicit, meaning-based judgements about stimuli, the practice was claimed to have helped proceduralize explicit knowledge about L1 and L2 form-meaning mappings (DeKeyser, Reference DeKeyser, VanPatten and Williams2015). Considering evidence of L1 activation during L2 processing and production (see Runnqvist, Gollan, Costa, & Ferreira, Reference Runnqvist, Gollan, Costa and Ferreira2013; Sanoudaki & Thierry, Reference Sanoudaki, Thierry, Thomas and Mennen2014), this knowledge may have helped the L2+L1 group parse and/or anticipate cues needed to disambiguate the Imparfait’s habitual from ongoing meanings more accurately and faster than the L2-only group. Given the preceding, it is unclear whether EI and practice play different roles. For example, L1 practice alone may have been sufficient to establish such knowledge, or, alternatively, it may not have provided any additional benefit.
In sum, to examine the role of L1 practice in L2 learning, the present replication examines the effect of providing L1 practice without L1 EI. This allowed us to examine the extent to which providing L1 EI with practice or L1 practice only—both alongside a core L2 treatment of L2 EI and practice—resulted in similar learning gains, online and offline.
In terms of predicted outcomes, providing L1 task-essential practice alone could result in similar learning gains as L1 EI plus L1 practice, that is, results would pattern with McManus and Marsden’s (Reference McManus and Marsden2017) L2+L1 group. This outcome would be accounted for by, for example, (a) learners being able to induce L1 EI from the L1 practice (see DeKeyser, Reference DeKeyser, VanPatten and Williams2015; Marsden & Chen, Reference Marsden and Chen2011; Sanz & Morgan-Short, Reference Sanz and Morgan-Short2004), that is, this account assumes that L1 EI can benefit L2 learning; or (b) L1 EI does not, in fact, affect or interact with L2 knowledge/processing mechanisms, but practice in interpreting the L1 does.
Our hypothesis that L1 EI could interact with accruing L2 knowledge is compatible with views that foreground a role for L1 in L2 processing (Ellis, Reference Ellis2006; O’Grady, Reference O’Grady2013), including evidence of L1 influence in L2 sentence processing (Roberts & Liszka, Reference Roberts and Liszka2013; Tokowicz & Warren, Reference Tokowicz and Warren2010) and convergence of L1 and L2 systems (Brown, Reference Brown2015; Brown & Gullberg, Reference Brown and Gullberg2013). We additionally assume roles for L2 EI plus L2 practice in L2 learning (DeKeyser, Reference DeKeyser, VanPatten and Williams2015). Given that the preceding may be particularly relevant for features with crosslinguistic differences in form-meaning mappings and processing routines, like the target feature investigated here, it is possible that the L1 EI in McManus and Marsden (Reference McManus and Marsden2017) could have directed learners’ attention to L1 form-meaning mappings and L1-entrenched processing routines to improve the effectiveness of the L1 practice. L1 practice, in turn, could promote remapping of L2 form-meanings and more accurate and efficient L2 processing during the L2 EI and practice. In short:
1. Results that pattern like those from McManus and Marsden’s (Reference McManus and Marsden2017) L2+L1 group would suggest a less important role for L1 EI, indicating that the benefits observed for the L2+L1 group were not (uniquely) attributable to the intervention’s L1 EI component. This would be compatible to some extent with previous findings that providing EI, albeit about the L2, was not necessary for offline performance (Marsden & Chen, Reference Marsden and Chen2011; Sanz and Morgan-Short, Reference Sanz and Morgan-Short2004) or beneficial for online processing (Andringa and Curcic, Reference Andringa and Curcic2015).
2. Results that pattern more closely to those from McManus and Marsden’s (Reference McManus and Marsden2017) L2-only group would suggest a more important role for L1 EI, indicating that the benefits observed for the L2+L1 group were, at least in part, attributable to the intervention’s L1 EI component or the combination of L1 EI plus L1 practice. This would corroborate findings by González (Reference González2008), Horst et al. (Reference Horst, White and Bell2010), and Spada et al. (Reference Spada, Lightbown, White, Housen and Pierrard2005) that L1 EI can be beneficial. It would also be compatible to some extent with the findings of Henry et al. (Reference Henry, Culman and VanPatten2009) and VanPatten and Borst (Reference VanPatten and Borst2012) that EI, albeit about L2, made practice more efficient.
METHOD
PARTICIPANTS
Participants were 19 English-speaking learners of French as a foreign language in semester two of a four-year bachelor of arts honors degree in French, recruited from two large universities in England. Every participant was an English native speaker, had completed A-level French (a British, subject-based, secondary school leaving qualification, equivalent to level B2 in the Common European Framework of Reference for Languages), and had not spent more than six weeks in a French-speaking country. Participants’ mean age was 19, and the mean time spent in a French speaking country was two weeks. These are very similar participant characteristics to the participants in McManus and Marsden (Reference McManus and Marsden2017).
DESIGN
The study design, timescale, procedure, and outcome measures were exactly as reported by McManus and Marsden (Reference McManus and Marsden2017). The design included a Pretest (week 1), Posttest (week 5), and Delayed Posttest (week 12), and the treatment was delivered in four 45-minute sessions over three weeks, totaling 3.5 hours. Sessions 1 and 2 were delivered in week 2, session 3 in week 3, and session 4 in week 4. As in the original study, none of the participants received any instruction between posttests.
INSTRUCTIONAL TREATMENT
The treatment included L2 EI and task-essential form-meaning mapping practice in interpreting the French Imparfait. It also included task-essential practice in interpreting equivalent English forms for ongoingness and habituality. These treatment components are briefly described in the following text (see McManus and Marsden [Reference McManus and Marsden2017] for descriptions and examples and IRIS for full materials, www.iris-database.com).
EI about the L2 was provided prepractice (approximately 5 minutes) and during the task-essential practice activities following incorrect answers. The amount of L2 EI received during practice was very similar across the treatment groups in McManus and Marsden (Reference McManus and Marsden2017) (see Table S3, Supplementary Materials). Identical to McManus and Marsden (Reference McManus and Marsden2017), the prepractice EI depicted conceptual-semantic information using a short video, image, or sound file of events. Then the appropriate aural and written forms were presented, and information given about how to interpret their meaning. No L1 EI was provided, either prepractice or during practice.
Immediately following the L2 prepractice EI, participants undertook forced-choice, task-essential practice that focused attention on meaning contrasts expressed by different verb forms. Images plus bracketed infinitives pinpointed the verb form to which participants should attend. The stimulus’s meaning was selected from two fixed options in training sessions 1–3 (as in, e.g., VanPatten, Reference VanPatten2002) and three in session 4. Practice lasted approximately 40–50 minutes per session. L1 (k = 112) and L2 (k = 384) practice items were interweaved to promote attention to L1 and L2 form-meaning mappings. Sample sentences are shown in (a) for past ongoingness and (b) for past habituality. The French items drew on 48 lexical verb types, each occurring eight times with Imparfait: four for reading (two habitual, two ongoing); four for listening (two habitual, two ongoing). The English items drew on 14 lexical verb types, which were also balanced across listening/reading and habitual/ongoing.
(1) Elle … mangeait un sandwich quand la cloche a sonné
“She … was eating a sandwich when the bell rang”
(2) Il … buvait une bière quand il sortait avec sa femme
“He … drank/used to drink a beer when he went out with his wife”
OUTCOME MEASURES
All measures were identical to those used in McManus and Marsden (Reference McManus and Marsden2017).
Context-Matching Tests (Listening and Reading)
All participants completed two context-matching tests, administered with EPrime (Schneider, Eschman, & Zuccolotto, Reference Schneider, Eschman and Zuccolotto2012): first listening (CMT-Listen), then reading (CMT-Read), each with 24 target and 8 filler trials. Each trial consisted of (a) a written English context: two sentences describing either a habitual or an ongoing activity and (b) a French stimulus: a two-clause French sentence that either matched (k = 12) or mismatched (k = 12) the meaning of the English context. Sample context-stimulus parings are shown in (3).
(3) Context (ongoing): Last Thursday Peter drove to the supermarket to buy some food for tomorrow’s party. On his way there he spotted his mother who was walking to work.
Matched stimulus: Quand Peter conduisait au supermarché, il a vu sa mère
“When Peter was driving to the supermarket, he saw his mother”
Mismatched stimulus: Quand Peter conduisait au supermarché, il voyait sa mère
“When(ever) Peter drove/used to drive to the supermarket, he saw his mother”
The English context appeared on screen for 10 seconds. Then, the French stimulus appeared orally (CMT-Listen) or as text (CMT-Read). Participants rated how good the match was between the meaning of the French stimulus and English context by pressing a number on the keyboard from 1 (“very good”), 2 (“good”), 3 (“neither good nor bad”), 4 (“poor”), and 5 (“very poor”), with a separate option for “I don’t know” (9). The written French stimulus remained on screen until a number was pressed. Answers could not be changed. The task was untimed and took approximately 20–25 minutes.
Self-Paced Reading Test
The SPR test was administered using EPrime after the CMTs and used 16 items from the CMT-Listen, with eight context-stimulus matches and eight mismatches. Half the items were followed by yes/no comprehension questions, which did not focus on the target feature or the context-stimulus match/mismatch, but increased the likelihood that participants focused on meaning (see Keating & Jegerski, Reference Keating and Jegerski2015). For each trial, the English context appeared for 10 seconds before an X appeared in the center of the screen. A spacebar press brought up the first and then each subsequent word of the French stimuli in the center of the screen. After the last word, the next screen displayed “END.” Reaction times were collected from each word.
DATA SCORING AND ANALYSIS
CMT responses were coded as follows: five points per correct response (i.e., one or two for matched, and four or five for mismatched trials); three points per midway responses (three for matched and mismatched trials); and one point per incorrect response (four or five for matched, and one or two for mismatched trials). Cronbach’s alphas were CMT-Listen version A (α = .81), version B (α = .86), CMT-Read version A (α = .74), and version B (α = .79).
In the SPR, a “critical word” in each French stimulus disambiguated ongoing from habitual meaning of the Imparfait and thus determined whether the sentence matched or mismatched the English context. RTs for the critical word were calculated from the onset of the critical word to the onset of the next word. Whole sentence RTs are the time taken to read from the onset of the first word to the onset of the “END” screen. We analyzed the raw RT data, removing critical word RTs less than 150 ms and greater than 2000 ms.
As none of the datasets were normally distributed (according to Shapiro-Wilks tests, all datasets p < .05), we present the results of nonparametric tests.
First, Friedman tests were used to compare Pretest, Posttest, and Delayed Posttest scores. If a statistically significant result was found, within-subject comparisons were made using Wilcoxon Signed-Rank tests with Bonferroni corrected alpha levels between pairs of test results: Pre-Post, Pre-Delayed, and Post-Delayed. Second, we tested for parity (Kruskall-Wallis H tests) on all measures at Pretest between all four groups: the current group (L2+L1prac) and the three groups from McManus and Marsden (Reference McManus and Marsden2017) (L2+L1, L2-only, and Control). Having established baseline parity, we then compared the current study’s results with the patterns of results found by McManus and Marsden (Reference McManus and Marsden2017) calculating between-group effect sizes (ES) and their confidence intervals (CI). Finally, we compared the current group’s (L2+L1prac) performance on matched versus mismatched trials in the SPR, using Wilcoxon Signed-Rank tests. These analyses mirrored that of McManus and Marsden (Reference McManus and Marsden2017).
Following McManus and Marsden (Reference McManus and Marsden2017) and recent discussion on decreasing the probability of Type II errors for low-stakes outcomes and as p-values can be strongly influenced by sample size (Plonsky & Oswald, Reference Plonsky and Oswald2014), the alpha was set at 0.10 for the Kruskall-Wallis H and Friedman tests. This alpha-level choice also reflects our small sample size and the study’s exploratory nature. The Bonferroni adjustment for the post-hoc Wilcoxon Signed-Rank tests revised the alpha to 0.10/3 = .033. For interpreting magnitudes of change, we present Cohen’s d ES and CIs (95%) for d. CIs for d that do not pass through zero can be considered reliable indicators of change. Within-group ES were calculated in relation to the mean and standard deviation of the Pretest as a baseline (and Posttest for ES at Delayed). As within-group ES tend to be larger than between-group, we also present, in the Supplementary Materials, within-group ES corrected for the dependence (correlation) between the means (Morris & DeShon, Reference Morris and DeShon2002, Table S2). Between-group ES adjusted for Pretest differences are additionally provided in the Supplementary Materials. Between-group ES are provided for each of McManus and Marsden’s (Reference McManus and Marsden2017) groups using the mean and standard deviation of the relevant group from McManus and Marsden (Reference McManus and Marsden2017) as the “comparison/control” group. We draw on Plonsky and Oswald’s (Reference Plonsky and Oswald2014) field-specific benchmarks: within subject: d ≥ .60 < 1.00 (small), ≥ 1.00 < 1.40 (medium), and ≥ 1.40 (large); between subject: d ≥ .40 < .70 (small), ≥ .70 < 1.00 (medium), and ≥ 1.00 (large).
RESULTS
HABITUAL TRIALS
Descriptive statistics for accuracy (CMTs) and reaction times (SPR) for the L2+L1prac group are shown in Table 1. All participants scored 100% on the SPR comprehension questions.
Habitual contexts for L2+L1prac group: Accuracy (CMTs) and reaction time (SPR) results
MATCHED TRIALS
Friedman tests showed no statistically significant improvement over time on any measure. All ES were negligible and CIs for d passed through zero.
• CMT-Read: XFootnote 2(2) = 1.80, p = .41 (Pre-Post, d [CI] = .14 [-.50, .78]; Pre-Delayed, d [CI] = .19 [-.45, .82]; Post-Delayed, d [CI] = .03 [-.61, .66])
• CMT-Listen: XFootnote 2(2) = 2.44, p = .29 (Pre-Post, d [CI] = .34 [-.30, .98]; Pre-Delayed, d [CI] = .04 [-.60, .67]; Post-Delayed, d [CI] = -.24 [-.87, .41])
• SPR critical word: XFootnote 2(2) = .74, p = .69 (Pre-Post, d [CI] = -.25 [-.89, .39]; Pre-Delayed, d [CI] = -.13 [-.77, .50]; Post-Delayed, d [CI] = .14 [-.50, .78])
• SPR whole sentence: XFootnote 2(2) = 2.95, p = .23 (Pre-Post, d [CI] = -.35 [-.98, .30]; Pre-Delayed, d [CI] = -.18 [-.82, .46]; Post-Delayed, d [CI] = .25 [-.39, .89])
Comparisons with the groups from McManus and Marsden (Reference McManus and Marsden2017) are presented in Table 2.
Habitual contexts: ES (Cohen’s d including CIs for d) comparisons with Control, L2+L1, and L2-only from McManus and Marsden (Reference McManus and Marsden2017), and ES changes with effects adjusted for baseline differences
Note. Positive ES in SPR indicate that L2+L1prac was slower than the other group. Negative ES in CMTs indicate that L2+L1 was less accurate than the other group.
Baseline parity was found between the four groups on all measures (CMT-Read, X2(3) = .081, p = .99; CMT-Listen, X2(3) = 1.21, p = .75; SPR critical word, X2(3) = .45, p = .93; SPR whole sentence, X2(3) = .29, p = .96) (for between-groups ES and CIs for d at baseline see Supplementary Materials, Table S1. All CIs for d passed through zero).
MISMATCHED TRIALS
Friedman tests showed no statistically significant improvement over time on any measure with only negligible ES. CIs for d passed through zero:
• CMT-Read: XFootnote 2(2) = 2.20, p = .33 (Pre-Post, d [CI] = .26 [-.39, .89]; Pre-Delayed, d [CI] = .31 [-.34, .94]; Post-Delayed, d [CI] = .04 [-.59, .68])
• CMT-Listen: XFootnote 2(2) = .35, p = .84 (Pre-Post, d [CI] = .05 [-.59, .69]; Pre-Delayed, d [CI] = .18 [-.46, .82]; Post-Delayed, d [CI] = .12 [-.52, .75])
• SPR critical word: XFootnote 2(2) = .105, p = .95 (Pre-Post, d [CI] = -.04 [-.67, .60]; Pre-Delayed, d [CI] = -.08 [-.71, .56]; Post-Delayed, d [CI] = -.05 [-.69, .59])
• SPR whole sentence: XFootnote 2(2) = 2.203, p = .33 (Pre-Post, d [CI] = -.34 [-.97, .31]; Pre-Delayed, d [CI] = -.20 [-.83, .45]; Post-Delayed, d [CI] = .02 [-.46, .81])
Comparisons with the groups from McManus and Marsden (Reference McManus and Marsden2017) are presented in Table 2.
Baseline parity was found between the four groups on all measures (CMT-Read, X2(3) = .58, p = .90; CMT-Listen, X2(3) = .82, p = .85; SPR critical word, X2(3) = .63, p = .73; SPR whole sentence, X2(3) = 1.29, p = .52).
RTs IN MATCHED VERSUS MISMATCHED TRIALS
Wilcoxon Signed Rank tests showed no statistically significant between-trial (matched vs. mismatched) differences for either SPR critical word or whole sentence RTs, with negligible ES and CIs for d that passed through zero:
• SPR critical word
o Pre: Z = -.12, p = .90, d [CI] = -.05 [-.69, .59]
o Post: Z = -.97, p = .33, d [CI] = -.29 [-.92, .36]
o Delayed: Z = -.64, p = .52, d [CI] = -.11 [-.74, .53]
• SPR whole sentence
o Pre: Z = -.12, p = .90, d [CI] = .21 [-.43. .85]
o Post: Z = -.40, p = .69, d [CI] = .11 [-.53, .74]
o Delayed: Z = -.64, p = .52, d [CI] = -.25 [-.39, .89]
In contrast, McManus and Marsden’s (Reference McManus and Marsden2017) L2+L1 group had RTs that were significantly slower in mismatched compared to matched trials at both Post and Delayed (medium ES). Between-trial differences for whole sentence processing at Post and Delayed were negligible, suggesting the differences observed at the disambiguating critical word region were specific to that region, rather than due to generally faster processing.
ONGOING TRIALS
Descriptive statistics for accuracy (CMTs) and reaction times (SPR) are shown in Table 3. All participants scored 100% on the SPR comprehension questions.
Ongoing contexts for L2+L1prac group: Accuracy (CMTs) and reaction time (SPR) results
MATCHED TRIALS
There was statistically significant change over time in only CMT-Read, with gains Pre-Post and Pre-Delayed, both with large ES. For all other measures, ES were negligible with CIs for d that passed through zero:
• CMT-Read: XFootnote 2(2) = 26.16, p = .00 (Pre-Post, Z = -3.34, p = .00, d [CI] = 1.65 [.88, 2.35]; Pre-Delayed, Z = -3.83, p = .00, d [CI] = 2.30 [1.44, 3.06]; Post-Delayed, Z = -1.23, p = .22, d [CI] = .23 [-.41, .86])
• CMT-Listen: XFootnote 2(2) = 1.97, p = .37 (Pre-Post, d [CI] = .55 [-.11, 1.18]; Pre-Delayed, d [CI] = .46 [-.20, 1.09]; Post-Delayed, d [CI] = -.06 [-.70, .58])
• SPR critical word: XFootnote 2(2) = 1.37, p = .50 (Pre-Post, d [CI] = -.41 [-1.04, .24]; Pre-Delayed, d [CI] = -.22 [-.85, .42]; Post-Delayed, d [CI] = .15 [-.49, .78])
• SPR whole sentence: XFootnote 2(2) = .32, p = .85 (Pre-Post, d [CI] = -.36 [-1.00, .28]; Pre-Delayed, d [CI] = -.23 [-.86, .42]; Post-Delayed, d [CI] = .07 [-.57, .70])
Table 4 shows comparisons with the groups from McManus and Marsden (Reference McManus and Marsden2017).
Ongoing contexts: ES (Cohen’s d including CIs for d) comparisons with Control, L2+L1prac+EI, and L2-only from McManus and Marsden (Reference McManus and Marsden2017), and ES changes with effects adjusted for baseline differences
Note. Positive ES in SPR indicate that L2+L1prac was slower than the other group. Negative ES in CMTs indicate that L2+L1 was less accurate than the other group.
Baseline parity between all four groups was found on all measures (CMT-Read, X2(3) = .87, p = .33; CMT-Listen, X2(3) = .99, p = .99; SPR critical word, X2(3) = .08, p = .99; SPR whole sentence, X2(3) = 1.02, p = .60) (for between-groups ES and CIs for d at baseline see Supplementary Materials, Table S1. All CIs for d passed through zero).
MISMATCHED TRIALS
Statistically significant change over time was observed only for CMT-Read, due to gains Pre-Post and Pre-Delayed, but with small ES:
• CMT-Read: XFootnote 2(2) = 6.22, p = .045 (Pre-Post, Z = -1.83, p = .07, d [CI] = .68 [.01, 1.32]; Pre-Delayed, Z = -2.22, p = .027, d [CI] = .86 [.18, 1.51]; Post-Delayed, Z = -.63, p = .53, d [CI] = .11 [-.53, .75])
• CMT-Listen: XFootnote 2(2) = 3.25, p = .19 (Pre-Post, d [CI] = .59 [-.08, 1.22]; Pre-Delayed, d [CI] = .43 [-.22, 1.06]; Post-Delayed, d [CI] = -.13 [-.77, .51])
• SPR critical word: XFootnote 2(2) = 1.26, p = .53 (Pre-Post, d [CI] = -.25 [-.88, .39]; Pre-Delayed, d [CI] = -.18 [-.81, .46]; Post-Delayed, d [CI] = .07 [-.57, .71])
• SPR whole sentence: XFootnote 2(2) = 2.95, p = .23 (Pre-Post, d [CI] = -.34 [-.97, .31]; Pre-Delayed, d [CI] = -.32 [-.96, .32]; Post-Delayed, d [CI] = .02 [-.62, .66])
Comparisons with the groups from McManus and Marsden (Reference McManus and Marsden2017) are given in Table 4.
Baseline parity between all groups was found in all measures (CMT-Read, X2(3) = 2.26, p = .52; CMT-Listen, X2(3) = .31, p = .96; SPR critical word, X2(3) = .23, p = .97; SPR whole sentence, X2(3) = 1.11, p = .58).
RTs IN MATCHED VERSUS MISMATCHED TRIALS
Wilcoxon Signed Rank tests showed no between-trial differences for either critical word or whole sentence RTs in the SPR. ES were negligible ES and CIs for d passed through zero:
• SPR critical word
o Pre: Z = -.81, p = .42, d [CI] = -.20 [-.84, .44]
o Post: Z = -.77, p = .44, d [CI] = -.29 [-.92, .36]
o Delayed: Z = -.52, p = .60, d [CI] = -.22 [-.85, .42]
• SPR whole sentence
o Pre: Z = -.12, p = .90, d [CI] = .08 [-.56, .72]
o Post: Z = -.80, p = .94, d [CI] = .08 [-.56, .71]
o Delayed: Z = -.32, p = .75, d [CI] = .12 [-.52, .75]
In contrast, McManus and Marsden (Reference McManus and Marsden2017) found small ES for between-trial differences at Post and Delayed in L2+L1 for critical word and whole sentence processing.
SUMMARY OF FINDINGS
McManus and Marsden (Reference McManus and Marsden2017) found increased accuracy and speed of interpretation of the Imparfait at Delayed following a treatment of L1 EI plus L1 task-essential practice (in addition to L2 EI plus practice). We partially replicated that original study to examine the role played by L1 practice by removing the L1 EI but retaining the L1 practice (and the core L2 EI and L2 practice). We used the original study’s design, procedures, and materials.
Our L2+L1prac group’s results patterned very similarly to McManus and Marsden’s (Reference McManus and Marsden2017) L2-only group and tended not to pattern as well with the L2+L1 group (see Tables 2 and 4).
In habitual contexts, change over time was negligible (all measures). In ongoing contexts, change over time in CMT-Listen and SPR was also negligible (all CIs for d passed through zero), but we did find CMT-Read improvement for Pre-Post and Pre-Delayed, matched (large ES) and mismatched (small ES). Comparisons with grammar instruction meta-analyses showed that the gains in ongoing matched contexts (Pre-Post d = 1.65, Pre-Delayed 2.30) fell within the broad ranges of mean ES reported in Shintani (Reference Shintani2015) (d = 1.71 – 2.65) and exceeded those reported in Spada and Tomita (Reference Spada and Tomita2010) (d range = .73 - .88 ).Footnote 2 However, gains in CMT-Read mismatched contexts were much smaller than in Shintani and fell slightly below Spada and Tomita’s ranges at Pre-Post (d = .68), but within at Pre-Delayed (d = .86). These findings pattern with McManus and Marsden’s (Reference McManus and Marsden2017) L2-only group (limited to a few offline measures, specifically in CMT-Read, Pre-Delayed, Ongoing matched, and CMT-Listen, Pre-Post, Habitual, and Ongoing matched).
Our L2+L1prac group showed no online sensitivity to anomalies, as only negligible matched-mismatched RT differences were found, consistent with McManus and Marsden’s (Reference McManus and Marsden2017) L2-only group.
These results suggest that L1 practice was not as beneficial as L1 EI plus L1 practice for learners’ online processing and offline interpretation of the Imparfait. Further, the L1 practice without L1 EI seemed to contribute little extra compared to the L2-only intervention (negligible ES throughout).
DISCUSSION AND CONCLUSIONS
Whereas the benefits of L2 EI and L2 practice are well researched to date, the current study addressed the role of L1 practice in L2 learning. Classroom-based evidence has suggested benefits of L1 EI (González, Reference González2008; Spada et al., Reference Spada, Lightbown, White, Housen and Pierrard2005), but has not examined L1 practice in L2 learning. Although that research had different designs to McManus and Marsden (Reference McManus and Marsden2017) and this replication, our findings broadly align with it, extending it to show L1 EI plus practice benefited L2 offline and online performance more than L1 practice alone. In short, L1 practice without L1 EI provided few learning benefits, suggesting a role for L1 EI for learning the L2 features investigated here that have crosslinguistic differences for L1 English learners. This sheds some light on the extent to which, and circumstances under which, EI may interface at some level with online processing, but it also raises several questions for future research.
The findings support McManus and Marsden’s (Reference McManus and Marsden2017) arguments that the combination of L1 EI plus L1 practice helped establish knowledge about L1 form-meaning mappings, which aided L2 form-meaning re-mapping, and, possibly, their proceduralization. The prepractice EI in McManus and Marsden (Reference McManus and Marsden2017) may have raised awareness about L1 processing routines, including L1-entrenched attention and competing cues (Ellis, Reference Ellis2006). For example, without EI, the habitual function of “-ed” may be difficult to induce from the input due to low saliency, multiple and complex form-meaning relations, and perhaps also prior instruction about the function of “-ed” (in L2 or L1 classes). The EI may have enabled learners to establish more efficiently (with fewer trials) how English “-ed” could map to a subsequent French Imparfait sentence.
We also note that this crosslinguistic complexity may explain, at least partly, why the L2 EI plus task-essential practice did not seem to help learning reliably (only in the CMTs), unlike in previous research using input-based task-essential practice (e.g., DeKeyser & Prieto Botana, 2015; Marsden, Reference Marsden2006; Marsden & Chen, Reference Marsden and Chen2011; Shintani, Reference Shintani2015; VanPatten, Reference VanPatten2015). The benefits of L2 EI plus practice may not have been detectable by more time-sensitive/speeded tests because of the influence of L1-entrenched processing routines.
As noted by McManus and Marsden (Reference McManus and Marsden2017), future research should also examine whether the effectiveness of L1 EI plus L1 task-essential practice depends on the feature having crosslinguistically different form-meaning mappings. We also acknowledge that our outcome measures (English contexts followed by French stimuli) may have activated L1 representations. Future research requires measures that do not intentionally coerce the L1, such as oral production.
Although we isolated the L1 practice, we did not isolate the L1 EI and so do not conclude that L1 EI alone was solely responsible for the benefits observed in McManus and Marsden (Reference McManus and Marsden2017). The L2+L1prac’s limited development suggests that the combination of L1 EI plus L1 practice led to the L2+L1 group’s superior L2 performance in McManus and Marsden (Reference McManus and Marsden2017), but future research should isolate L1 EI to test this.
SUPPLEMENTARY MATERIAL
To view supplementary material for this article, please visit https://doi.org/10.1017/S0272263117000171
