Exploring The L1-L2 Relationship In The L2 Acquisition Of Prosody

Total Page:16

File Type:pdf, Size:1020Kb

Exploring The L1-L2 Relationship In The L2 Acquisition Of Prosody

Exploring the L1-L2 Relationship in the L2 Acquisition of Prosody*

Laurent Rasier1,2 & Philippe Hiligsmann2

Université catholique de Louvain1 & F.R.S – FNRS2

Abstract

This paper explores the L1-L2 relationship in the L2 acquisition of prosody. The first part of the paper outlines a method to investigate transfer effects in L2 prosody. Then, we report on an experiment in which we examined native and non-native speakers’ use of (de)accentuation to prosodically signal information status in Dutch and French. With the aid of a picture description task in which the information value of the discourse referents was varied experimentally, we elicited both L1 and L2 data from 20 French-speaking learners of Dutch and 20 Dutch-speaking learners of French. The main research question was whether there would be any relationship between the typological distance between the learners’ L1 and L2

(defined in terms of the “markedness relationships” between the two languages) and the occurrence of transfer in their use of (de)accentuation. The results reveal considerable differences between native and non-native speakers on the one hand and between the two groups of L2 learners on the other hand. More specifically, the results support the view that markedness is an important factor in predicting and explaining learning difficulties as well as the cases of prosodic transfer.

* We gratefully acknowledge the financial support of the Belgian National Fund for Scientific Research (FRS- F.N.R.S.). We also thank the participants in the L1-L2-conference (First and second languages: exploring the relationship in pedagogy-related contexts, Oxford, 27-29 March 2009) as well as an anonymous reviewer for their helpful comments on an earlier version of this paper. 1. Introduction

The relationship between the mother tongue (L1) and the second language (L2) has been the object of numerous studies in the field of second language acquisition (SLA) research. Much of the discussion has focussed on so-called transfer phenomena that are generally defined as

“the incorporation of features of the L1 into the knowledge system of the L2 which the learner is trying to build” (Ellis 1994: 28). After a period in the 1950s-1960s when transfer was seen as the main cause of learning difficulties and a period in the 1970s when it was denied any place in the L2 acquisition process, we have now reached a point in the transfer debate where most authors acknowledge that “despite its sometimes irritatingly elusive character, transfer is one of the major factors shaping the learner’s interlanguage competence and performance”

(Kohn 1986: 21). Besides, it is also becoming increasingly clear that transfer phenomena need not be restricted to the learners’ incorporation of L1 elements into their L2 production. As suggested by Odlin (1989: 27), transfer can also result from “similarities and differences between the target language and any other language that has been previously (and perhaps imperfectly) acquired”. In other words, transfer is not limited to the influence of L1 on L2, but can also involve the influence of L2 on L3. Moreover, there are well-documented examples in the literature in which it is the learners’ L1 that gets influenced by some characteristics of subsequently acquired languages. Taken together, these findings suggest that transfer should no longer be seen as a unidirectional phenomenon (as was e.g. the case in Lado’s Contrastive

Analysis Hypothesis; Lado 1957) but as a multidirectional one (see e.g. Pavlenko & Jarvis

2002 on “bidirectional transfer” and Jarvis & Pavlenko 2008 on “forward transfer”, “reverse transfer”, and “lateral transfer”).

Following Benson (2002: 68), it can be concluded from the above that “transfer does occur, but is a far more complex phenomenon than hitherto believed”. In order to take this complexity into account, transfer will be defined here in very general terms, as the cross-lin- guistic influence, within an individual’s linguistic system, of one (or more) language(s) over another (see also Gilquin 2008: 4-5)1. Despite intensive research efforts, many questions re- garding the exact nature of transfer, the circumstances in which it occurs, and the psychological processes it relies on are still, to a large extent, left unsolved. To give but one example, instances of L1 influence on L2 performance are well-documented in the area of phonology where a so-called “foreign accent” is probably the clearest manifestation of the learner’s L1 in L2 speech. Yet, most research on the acquisition of phonological skills in an

L2 has hitherto concentrated on segmental issues, thereby disregarding suprasegmentals/pro- sody (e.g. intonation, stress, accent, rhythm; De Bot 1986, Leather & James 1991, Trouvain &

Gute 2007).

This paper investigates transfer effects in the L2 acquisition of prosody. It is organized as follows. First, we review earlier methodological frameworks in transfer research. It is argued that they actually left out an important aspect of transfer, i.e. its relation to typology and more precisely the markedness relationships between the learners’ L1 and L2. Then, we present an alternative approach in which typological constraints on transfer are taken into account. The method we propose is illustrated by means of a study of prosodic transfer.

2. Assessing (prosodic) transfer in SLA

The notion of transfer in SLA research is closely related to contrastive analysis (CA). After a period of disregard, we see nowadays a clear resurgence of CA in SLA research which is motivated by the view that “you cannot account for SLA just by describing L1, L2 and/or IL.

To account for SLA, you must move to second order application, and compare these” (James

1994: 180). Indeed, as Selinker (1989: 287) observes, “CA is the best place to begin language transfer studies since structural congruence (or at least, partial structural similarity) is most

1 Grosjean (1998, 2008) suggests that a further distinction should be made between transfer and interference. In his view, transfer is basically static as it reflects permanent traces of one language in the other (as in the case of a foreign accent), whereas interference is dynamic as it refers to elements of another language that incidentally slip into the output of the language being spoken/written (as in the case of occasional pronunciation errors). probably necessary, though not sufficient, for most types of language transfer to occur”. Not surprisingly, then, most contemporary approaches seek to combine CA with an in-depth study of samples of learner language, i.e. interlanguage (IL) (see Rasier & Hiligsmann to appear for a discussion of recent developments in CA in relationship with SLA research).

In this section, we outline our approach to transfer. The method shares a number of features with contemporary approaches to transfer (e.g. James 1994, Granger 1996, Jarvis

2000, Gilquin 2008) for it also relies on a combination of contrastive analysis and contrastive interlanguage analysis. However, there are also a number of differences as our model’s CA component takes the typological distance between the learners’ L1 and L2 into account, and its CIA section examines the L2 performance of learners with different target languages.

More specifically, we compare learners in a given learning situation (e.g. French-speaking learners of Dutch) with learners in the reverse learning situation (e.g. Dutch-speaking learners of French), and then relate the IL data to data from other categories of learners. To guarantee the validity of the comparison, care must be taken that the learners all have comparable profiles.

The first step in our approach consists in a contrastive analysis of comparable data in the learners’ L1 and L2. To guarantee the comparability of the data, these data have to be gathered in the same experimental conditions. Then, the L1 and L2 data are related to comparable data for other typologically (un)related languages so that the typological distance between the languages under investigation can be determined. This comparison results in a contrastive-typological description of native speech that has both a predictive and explanatory value for IL facts.

After contrastively analyzing native speech, we set out to describe the interlanguage data. Using the data from the native speakers as a point of reference (i.e. control data), we first describe the IL varieties under investigation to determine what the L2 learners are doing right or wrong when using the L2. Then, the IL data are compared with each other as well as with data from other IL varieties (provided these are available), leading to a “contrastive- typological description of IL”, or, in James’ terminology, “a quantificational contrastive typology of a number of interlanguages” (James 1994: 187).

Finally, we relate the contrastive-typological descriptions of native and interlanguage speech to each other in order to get insight into the various forms of transfer in the IL data

(e.g. positive/negative transfer, overuse/underrepresentation).

The following figure summarizes our integrated approach to CA, CIA, and typology.

Mother tongue Other natural Target language (L1) languages (L2)

Contrastive-typological description of NS

TRANSFER

Contrastive-typological description of IL

Interlanguage 1 Other interlanguage Interlanguage 2 (IL1) varieties (IL2)

Figure 1. Integrated approach to CA, CIA, and typology (Rasier & Hiligsmann 2007: 46)

3. Case study: prosodic marking of information status in L2 Dutch and L2 French

This section presents a case study using our integrated approach to transfer. Although our example deals with a phonological variable in L2 Dutch and L2 French, it is possible to apply our approach to other languages, other types of transfer (e.g. from L2 to L3), and other phe- nomena (e.g. syntactic, pragmatic, etc). And while our example comes from spoken language, the approach could also be used in a study of transfer in written language, provided the neces- sary material is available.

3.1. Background and experimental setup

The prosodic phenomenon we investigate in this case study is (pitch) accent. Contrastive research on prosody indicates that there are (quite) large differences between languages as far as their patterns of utterance-level accentuation are concerned. This is especially true for

Germanic and Romance languages that, according to Vallduvi (1991), fall into two groups:

1. Plastic languages. These are languages where contextually important information al-

ways gets accented, whereas contextually less important or contextually known

information does not, e.g. English, Dutch, German.

2. Non-plastic languages. These are languages with a fixed accent distribution that does

not reflect the news value of the sentence elements, and in which one has to use syn-

tactic cues instead in order to signal information status, e.g. French, Spanish, Italian.

We wanted to know if such typological contrasts have an influence on the L2 learners’ use of

(de)accentuation2 to signal information status. We investigated this issue in the L2 speech of

20 (Belgian) French-speaking learners of Dutch and 20 Dutch-speaking learners of French (10 from the Netherlands and 10 from Flanders). All of them were 2nd-year university students of economics (age: 19-20) who had been studying the L2 in a formal setting for eight years.

Our material consists of noun phrases of the type “indefinite article + adjective + noun” (Dutch data) and “indefinite article + noun + adjective” (French data). The data were elicited with a picture description task in which the participants were asked to describe geo- metrical figures (a circle, a triangle, a star, a square) appearing on a computer screen in dif- ferent colours (red, yellow, blue, green). All the participants took the test in their L1 and L2 so that we had comparable data in both varieties at our disposal. Using situational contrasts

2 Deaccentuation refers to the fact that “a word that we might expect to be accented fails to be accented in a context where it has recently been used or where the entity to which it refers has recently been mentioned” (Ladd 1996: 175). (see also Swerts et.al. 2002), the information value of the words was systematically varied, re- sulting in four conditions: New/New (NN), Given/Contrastive (GC), Contrastive/Given (CG),

Contrastive/Contrastive (CC). A property is defined as “new” when it has not yet been used in the discourse. It is called “given” when it has already been mentioned in the preceding con- text. A property that differs from the immediately preceding utterance is called “contrastive”.

We used a perceptive approach in order to assess the presence/absence of accent on a particular piece of information. The data were first transcribed by the two authors independently. Then, the transcriptions were compared with each other. In cases of disagreement, the data were discussed until agreement was reached. We did not cross-check our transcriptions with the aid of acoustic measurements (see Demers 1994 who advocates such a ‘mixed/combined’ approach) because the segmental material in the stimuli was insufficiently controlled for in order to carry out reliable acoustic analyses on the data. All noun phrases were used for a distributional analysis involving the different kinds of comparisons discussed in section 2 (see figure 1).

3.2. Step 1: contrastive L1-L2 analysis

The first step in our approach involves a comparison of data in the learners’ L1 and L2. It is predicted that the L1 French data will be characterized by fixed accentuation and no deaccen- tuation of contextually given information, whereas the L1 Dutch data will display a close rela- tionship between the presence/absence of accent on a discourse referent and its contextual news value as well as a high rate of deaccentuation of contextually given information.

Figure 2 summarizes the results for the L1 Dutch and L1 French data. As expected, there is a closer association between accentuation and news value in L1 Dutch (Cramer’s V =

0,878; X2 = 319,107; df = 1; p < 0,001) than in L1 French (Cramer’s V = 0,615; X2 =

204,185; df = 1; p < 0,001). This difference can be attributed to the different treatment of con- textually given information in the two languages. As is apparent from figure 8, it is most of the time un-/deaccented in L1 Dutch (= 89,45%), but much less so in L1 French (= 66,79%).

Accentuation Language [-accent] Dutch French [+accent] 97.99% 100.0% 92.69% 89.45%

80.0%

66.79%

t 60.0% n e c r e P

40.0% 33.21%

20.0% 10.55% 7.31% 2.01%

0.0% Given Important Given Important News value News value Figure 2. Relationship between news value and (de)accentuation in L1 Dutch and L1 French

When we examine the accent patterns that were produced by the native speakers of Dutch, we see that the most frequent accent pattern in each experimental condition reflects the contextual news value of the words (figure 3). The only exception can be found in the GC-condition where a fairly high number of cases with a prenuclear accent on the contextually given piece of information can be found as well. Interestingly, the reverse pattern (i.e. postnuclear accent on the G-element) hardly ever occurs in the CG-condition. NN OC

10,00% 5,00% Accentdistributie een RODE driehoek een rode DRIEHOEK een RODE DRIEHOEK EEN//RODE//DRIEHOEK

Pies show percents

50,00%

90,00% 45,00%

CO CC 2,56% 5,00%

97,44% 95,00% Figure 3. Accent patterns in L1 Dutch in the four experimental contexts

Moving on to the L1 French data (fig. 4), it appears that the French native speakers produce the same accent pattern in the four experimental contexts. Traditionally, French is described as a language with a fixed noun phrase-final accent, but this pattern is largely underrepresen- ted in our data where the dominant accent pattern in each condition involves a double accent on the adjective and the noun, i.e. arc accentual (“bridge accent”) instead of accent final

(“final accent”). The data also contain a few instances of arc accentuel étendu (“extended bridge accent”), i.e. a pattern in which the first accent is realized on the indefinite article and the second one on the noun, as well instances of accent de focalisation (“narrow focus accent”) on the contextually contrastive entity in the GC- and CG-conditions. It should be emphasized, though, that such cases of deaccented G-element are less frequent in L1 French than in L1 Dutch (GC-condition: 5% vs. 50%; CG-condition: 30% vs. 98%). NN OC Accentdistributie 10,00% 5,00% 5,00% un cercle ROUGE un CERCLE rouge 30,00% un CERCLE ROUGE UN cercle ROUGE 30,00% UN//CERCLE//ROUGE

Pies show percents

60,00%

60,00%

CO CC 1,67% 1,67% 6,67% 6,67% 3,33%

30,00%

66,67%

83,33% Figure 4. Accent patterns in L1 French in the four experimental contexts

Taken together, our L1 data confirm the view that Dutch can be considered as a language with plastic accentuation3, whereas French appears to have a (quasi) non-plastic accent system.

Indeed, contrary to the claims made in traditional treatments of French accentuation, deaccen- tuation does occur in French (thanks to the accent de focalisation, i.e. “narrow focus accent”), be it to a lesser extent than in Dutch but to a higher degree than in Italian where Swerts et.al.

(2002) did not find any single instance of deaccented given information.

3.2. Step 2: contrastive-typological description of native speech

The results of the contrastive L1-L2 analysis in the previous section suggest that structural principles outweigh pragmatic information in French, while it is the opposite in Dutch. If we compare Dutch and French with other languages, it appears that many languages have structural constraints on accentuation, whereas pragmatic factors need not be at play. More precisely, although accent placement is mainly governed structurally in non-plastic languages,

3 This by no means implies that Dutch does not have any syntactic strategy to express focus meaning. In fact, there are quite a lot of cases in spoken cases where prosodic and syntactic means are used to foreground a con- textually important piece of information (Rasier 2006). many prosodic systems rely on both structural and pragmatic information in their accent placement strategies, albeit in a different order of preference (Rasier 2008). But, as can be seen in figure 5, there does not to be any language where structural constraints are totally absent and thus where there is a perfect correlation between news value and (de)accentuation.

Structural rules Structural & Pragmatic & Pragmatic rules only pragmatic rules structural rules only

Portuguese, French, English, Dutch, Italian, Spanish Romanian German ?

Figure 5. Typology of accent systems (Rasier 2008: 55)

The data in figure 5 give us an idea of the typological distance between the Dutch and French accent system. If we combine them with e.g. Eckman’s Markedness Differential Hypothesis

(MDH; Eckman 1987), it is possible to formulate a number of testable predictions/hypotheses about the L2 learners’ accentuation strategies.

Eckman’s MDH states that a structure A is more marked than B if the existence of A in a language necessitates the existence of B but B does not need A to exist in a language. In prosodic terms, this means that pragmatic accentuation rules are more marked than structural ones. Eckman (1987) also predicts that the more marked a rule is, the more difficult it is to learn. Pragmatic accentuation rules – and accent systems where such rules have a dominant position – should therefore be more difficult for L2 learners to acquire than structural ones. In other words, the Dutch accentuation principles are in Eckman’s terms more marked than the

French ones, and should therefore cause more learning difficulties.

As figure 6 shows, the same kind of reasoning can be applied to individual accent patterns. In this respect, it is still assumed that accent patterns resulting from the application of structural rules are less marked than accent distributions that are motivated pragmatically. Unmarked Marked

Structural Struct.-prag. Prag.-struct. Pragmatic

Dutch: Dutch: Dutch: French: Dutch: - rhythmic acc. - / - broad focus - focus - narrow French: French: French: accent focus - final accent - bridge accent - extended - - bridge accent deaccentua deaccentua tion tion Figure 6. Typology of accent patterns in Dutch and French (Rasier 2008: 62)

As far as ease of learning is concerned, Eckman (1987) predicts that marked patterns (e.g. deaccentuation) will be more difficult to learn than unmarked ones (e.g. final accent, bridge accent). He also states that marked L2 patterns that are less marked than in L1 should not be difficult. In such a case, Dutch-speaking learners of French should not find it hard to deaccent contextually given pieces of information in French, whereas French-speaking learners of

Dutch should have much difficulty with that pattern. Finally, Eckman predicts that unmarked patterns are more likely to be transferred to L2 than marked ones. So, for example, French- speaking learners of Dutch should transfer the French final accent and bridge accent to their

Dutch interlanguage but not the narrow focus accent (accent de focalisation) of their L1.

3.3. Step 3: contrastive interlanguage analysis

The third step in our approach consists in testing the hypotheses that were formulated on the basis of the contrastive-typological description of native speech against the IL facts. As pointed out earlier, we first compare the IL data with the corresponding control data in order to detect non-native features in L2 performance.

3.3.1. L2 Dutch vs. L1 Dutch

Comparing the accentuation strategies of French-speaking learners of Dutch with the strate- gies of native speakers, it appears first of all that there is a lower association between news value and accentuation in L2 Dutch than in L1 Dutch (Cramer’s V = 0,600; X2 = 151,302; df = 1; p < 0,001 in L2 Dutch vs. Cramer’s V = 0,878 in L1 Dutch). As can be seen in figure 7, the main difference between the two groups of speakers has to do with their treatment of con- textually given information. Most of the time, it gets deaccented in native speech but it is less often the case in the L2 performance of the French-speaking learners of Dutch.

Language background Accentuation L1 Dutch L2 Dutch [-accent] 98.50% [+accent] 100.0% 97.99%

89.45%

80.0%

56.82%

t 60.0% n e c r

e 43.18% P

40.0%

20.0% 10.55%

2.01% 1.50% 0.0% Given Important Given Important News value News value Figure 7. Relationship between news value and (de)accentuation in L1 and L2 Dutch

The difference between the native and non-native speakers of Dutch is even more striking when we look at the results for the GC- and CG-conditions in figure 8, i.e. the two experi- mental contexts where deaccentuation could occur. In comparison with L1 Dutch, the distri- butions in which the G-element has no accent are clearly underrepresented in the L2 data, which points to the relative difficulty of deaccenting contextually given information for the

French-speaking learners of Dutch. Indeed, the G-element is deaccented in only 5% of the cases (vs. 50% in L1 Dutch; X2 = 16,184; df = 1; p < 0,001) in the GC-condition, whereas it is deaccented in 35% of the cases (vs. 95% in L1 Dutch; X2 = 28,057; df = 1; p < 0,001) in the

CG-condition. Interestingly, deaccentuation is a marked accentuation rule, which according to

Eckman (1987) explains why it causes so much difficulty. Another interesting characteristic of the French-speaking learners’ L2 speech is the fact that they produce the same default pattern in all experimental conditions. This pattern involves the accentuation of the adjective and the noun, and is reminiscent of the French “bridge accent”, i.e. an unmarked pattern.

NN GC Accent patterns een RODE driehoek een rode DRIEHOEK 10,00% 5,00% 25,00% 5,00% 30,00% 5,00% 5,00% een RODE DRIEHOEK EEN//RODE//DRIEHOEK

Pies show percents

45,00% 90,00% 60,00% 50,00% 70,00%

CG CC

15,00% 2,56% 35,00% 5,00% 17,50%

50,00% 95,00% 97,44% 82,50%

L1 Dutch L2 Dutch L1 Dutch L2 Dutch Language background Language background Figure 8. Accent patterns in L1 Dutch and L2 Dutch in the four experimental contexts

3.3.2. L2 French vs. L1 French

We now turn to the French interlanguage of the Dutch-speaking informants. Looking at the relationship between news value and accentuation, it appears that this IL variety comes quite close to the French target (Cramer’s V = 0,632; X2 = 205,159; df = 1; p < 0,001 in L2 French vs. Cramer’s V = 0,615 in L1 French) (fig. 9). This result is in line with the prediction that it may be easier for speakers of a plastic language (such as Dutch) to shift to a non-plastic language (such as French) then the other way round. Language background Accentuation L1 French L2 French [- accent] [+ accent]

100,0% 92,69% 94,33%

80,0%

67,14% 66,92%

60,0% t n e c r e P

40,0% 32,86% 33,08%

20,0%

7,31% 5,67%

0,0% Given Important Given Important News value News value Figure 9. Relationship between news value and (de)accentuation in L1 and L2 French

If we compare the accent patterns that were produced by the native and non-native speakers of

French (fig. 10), we see first of all that the Dutch L2 speakers of French produce the same default pattern as the native speakers in three out of the four experimental contexts. In this pattern, the adjective and the noun are both accented, which recalls the French “bridge ac- cent” and seems to confirm the idea that the French “bridge accent” should not be difficult to acquire as it is an unmarked accent pattern. Yet, it should be noted that the structural similarity between the French (unmarked) bridge accent and the Dutch double accent in the

NN-, GC-, and CC-conditions may also have played a facilitating role here. Another crucial aspect is that the Dutch-speaking learners of French do not find it hard to deaccent contextually given information in French, and actually overuse this strategy in their L2 speech

(GC-condition: 5% in L1 French vs. 18,42% in L2 French; CG-condition: 30% in L1 French vs. 56,14% in L2 French). NN GC Accent patterns un cercle ROUGE un CERCLE rouge 30,00% 10,00% 10,53% 5,00% 5,00% 18,42% 18,42% un CERCLE ROUGE 15,79% UN cercle ROUGE UN//CERCLE//ROUGE 30,00% 5,26% Pies show percents

60,00% 73,68% 60,00% 57,89%

CG CC

5,26% 10,53% 1,67% 1,67% 6,67% 6,67% 30,00% 3,33% 14,04%

38,60% 66,67% 56,14% 83,33% 75,44%

L1 French L2 French L1 French L2 French Language background Language background Figure 10. Accent patterns in L1 French and L2 French in the four experimental conditions

In the next stage, we summarize the main results of the contrastive IL analysis, and relate them to comparable data for other IL varieties (provided such data are available).

3.4. Step 4: contrastive-typological description of non-native speech

Our interlanguage analysis reveals that it is easier for speakers of a plastic language (i.e.

Dutch) to acquire the accentuation principles of a non-plastic language (i.e. French) than the other way round (47% contextually adequate accent patterns in L2 Dutch vs. 78% in L2

French; X2 = 35,954; df = 2; p < 0,001; fig. 11). More specifically, our French-speaking lear- ners of Dutch experience much difficulty in matching the accent distribution to the news va- lue of the words, and tend to produce the same “default” accent pattern in every experimental context, which leads to accent distributions that are not always contextually adequate. The overuse of the “double accent”-pattern also results in the underrepresentation of cases of deac- cented G-element in the GC- and CG-conditions. Conversely, the Dutch-speaking learners of

French do not find it hard to produce a “bridge accent” in all experimental conditions, al- though a slight overuse of patterns with a deaccented G-element can be observed in the GC- and CG-conditions.

78.36% Interlanguage variety 80.0% L2 Dutch L2 French

60.0%

47.14% t n e c r 40.0% e 34.29% P

18.57% 20.0% 11.70% 9.94%

0.0% Cont. inadequate Acceptable Cont. adequate Correction level Figure 11. Percentage of contextually inadequate, acceptable, and adequate accent patterns in L2 Dutch and L2 French Interestingly, the tendency not to match the accentuation to the news value of the words has also been reported in other IL varieties, e.g. Spanish English (Ramirez Verdugo 2002), Hun- garian and Polish English (Archibald 1997), Dutch English (Willems 1982). Counterevidence is provided by Barlow (1998) who reports that his Spanish, Italian, and Chinese learners of

English at different levels of L2 proficiency did not make any mistake when assigning pitch accent in L2 English. And this holds true for both normal/non-contrastive contexts with a nuclear accent in default right-most position and contrastive contexts in which the L2 learners had to move the nuclear accent from its right-most position onto a non-final word. Besides,

Archibald (1997) shows that it is the L1’s main accentuation rule that tends to be transferred and overused in L2 speech, and two longitudinal studies of German English by Grosser

(1993) and Wieden (1997) report that L2 learners first need to master the English rule for nor- mal/non-contrastive accentuation, which is essentially motivated structurally, before acquiring the rule for contrastive accentuation and deaccentuation of contextually given information. They also observe that their L2 learners tend to overuse the two types of rules before using them in a contextually adequate way.

3.5. Step 5: transfer analysis

The final step in our approach relates the contrastive-typological description of native and non-native speech to each other in order to detect and explain the cases of transfer in L2. In this respect, it was predicted, following Eckman (1987)’s Markedness Differential Hypo- thesis, that only the marked L1 patterns will be transferred from L1 to L2.

The analysis of the IL data of the French-speaking learners of Dutch reveals that, contrary to expectation, they do not produce a (unmarked) phrase-final accent in their L24.

Rather, they produce a “double accent” in all experimental conditions5, which recalls the L1

French strategy, i.e. using a “bridge accent” whatever the information value of the words may be. This leads to accent patterns that are not always contextually adequate and to the underre- presentation of cases with a deaccented G-element in the GC- and CG-conditions (i.e. nega- tive transfer). French also has a deaccentuation rule, but it is marked. And this explains why this pattern does not get transferred to the learners’ interlanguage. Interestingly, this pattern can also be observed in other interlanguage varieties6, which reinforces the idea that marked pragmatic accentuation rules are more difficult to acquire than their unmarked structurally- motivated counterparts. Conversely, the analysis of the L2 French data from the Dutch-spea- king learners shows that they do not find it difficult to apply the marked deaccentuation rule of their L2, and even tend to overuse it in comparison with native speech. This can be explained by the fact that the French deaccentuation rule, despite its marked status, is less marked than in the learners’ mother tongue.

4 Nor do they in their L1. This may support Di Cristo (1999, 2000)’s contention that the bridge accent is gaining ground in French. An alternative explanation is that the pattern we found in our L1 French data is actually an idiosyncrasy of Belgian French. However, this does not call into question the validity of the IL-L1 comparison. 5 Notice that the structural similarity/congruence between the L1 Dutch double accent in the NN-,GC-, and CC- conditions and the L1 French bridge accent may have had a facilitating effect on the production of contextually adequate patterns in these conditions (i.e. positive transfer). 6 The results reported in this paper also replicate earlier observations by Hiligsmann & Rasier (2002) and Rasier (2003) in which use was made of a reading task and spontaneous dialogues. Taken together, these findings confirm the idea that there are qualitative differences in terms of learning process between unmarked structurally-motivated accentuation rules and their marked pragmatically-motivated counterparts. There is also supporting evidence for the claim that such qualitative differences are reflected in the extent to which they are subject to transfer to and/or over-/underuse in the learners’ L2 speech.

4. Conclusions

This paper investigated the L1-L2 relationship in the L2 acquisition of prosody. We first re- viewed earlier approaches to transfer in SLA research, and argued that these approaches actu- ally did not pay (enough) attention to typological constraints on transfer. Then, we proposed an alternative method in which such constraints are taken into account, and exemplified it in a study of prosodic transfer in L2 Dutch and L2 French. As it stands, our approach does not give insight into the communicative consequences of the learners’ (L1-induced) accentuation strategies. Using perception tests, future work will concentrate on the impact of the learners’ accentuation strategies on the overall acceptability and comprehensibility of their speech.

5. References

Archibald, J. (1997), ‘The acquisition of second language phrasal stress: a pilot study’ in: S.J. Hannahs & M. Young-Scholten (eds), Focus on phonological acquisition, John Benja- mins Publishing Company, 263-289. Barlow, J.S. (1998), Intonation and Second Language Acquisition. A Study of the Acquisition of English Intonation by Speakers of other Languages, PhD dissertation, University of Hull. Benson, C. (2002), ‘Transfer/cross-linguistic influence’, ELT Journal, 56-1, 68-70. De Bot, K. (1986), ‘The transfer of intonation and the missing database’, in: E. Kellerman & M.S. Smith (eds), Crosslinguistic Influence in Second Language Acquisition, Pergamon: New York, 110-119. Di Cristo, A. (1999), ‘Vers une modélisation de l’accentuation en français (première partie)’, Journal of French language studies, 9-2, 143-163. Di Cristo, A. (2000), ‘Vers une modélisation de l’accentuation en français (seconde partie), Journal of French language studies, 10-1, 27-45. Demers, M. (1994), ‘A propos de l’analyse intonative. L’oreille ou la machine?’, Langues et linguistique, 20, 1-27. Eckman, F. (1987), ‘Markedness and the contrastive analysis hypothesis’, in: G. Ioup & S.H. Weinberger (eds), Interlanguage Phonology. The Acquisition of a Second Language Sound System, Cambridge: Newbury House Publishers, 55-69. Eckman, F. (2008), ‘Typological markedness and second language phonology’, in: J.G. Han- sen-Edwards & M.L. Zampini (eds), Phonology and Second Language Acquisition, John Benjamins Publishing Company: Amsterdam/Philadelphia, 95-115. Ellis, R. (1994), The Study of Second Language Acquisition, Oxford: Oxford University Press. Gilquin, G. (2008), ‘Combining contrastive and interlanguage analysis to apprehend transfer: detection, explanation, evaluation’, in: G. Gilquin, S. Papp & M. Belén Diez-Bedmar (eds), Linking up contrastive and learner corpus research, Amsterdam & New-York: Ro- dopi, 3-33. Granger, S. (1996), ‘From CA to CIA and back: an integrated approach to computerized bilingual and learner corpora’, in: K. Aijmer, B. Altenberg & M. Johansson (eds), Lan- guages in Contrast. Papers from a Symposium on Text-based Cross-linguistic Studies. Lund 4-5 March 1994, Lund: Lund University Press, 37-51. Grosjean, F. (1998), ‘Transfer and language mode. Commentary on N. Müller “Transfer in bilingual first language acquisition”, Bilingualism: Language and Cognition, 1, 175-176. Grosjean, F. (2008), Studying Bilinguals, Oxford: Oxford University Press. Grosser, W. (1993), ‘Aspects of intonation L2 acquisition’, in: B. Kettemann & W. Wieden (eds.), Current Issues in European Second Language Acquisition research, Tübingen: Gunter Narr Verlag, 81-94. Hiligsmann, Ph. & L. Rasier (2002), ‘De zinsaccentuering in de tussentaal van Franstalige leerders van het Nederlands’, n/f. Themanummer ‘Huidige tendensen in het NVT-onder- wijs en –onderzoek’, 1, 37-53 James, C. (1994), ‘Don’t shoot my dodo. On the resilience of contrastive and error analysis’, International Review of Applied Linguistics, 32, 179-200. Jarvis, S. (2000), ‘Methodological rigor in the study of transfer: identifying L1 influence in the interlanguage lexicon’, Language Learning, 50-2, 245-309. Jarvis, S. & A. Pavlenko (2008), Crosslinguistic Influence in Language and Cognition, New- York & London: Routledge. Kohn, K. (1986), ‘The analysis of transfer’, in: E. Kellerman & M. Sharwood Smith (eds), Crosslinguistic Influence in Second Language Acquisition, New-York: Pergamon Press, 21-34. Ladd, D.R. (1996), Intonational phonology, Cambridge: Cambridge University Press. Lado, R. (1957), Linguistics across Cultures, Ann Arbor: University of Michigan Press. Leather J. & A. James, (1991), ‘The acquisition of second language speech’, Studies in Se- cond Language Acquisition, 13, 305-341. Odlin, T. (1989), Language Transfer, Cambridge: Cambridge University Press. Pavlenko, A. & S. Jarvis (2002), ‘Bidirectional transfer’, Applied Linguistics, 23-2, 190-214. Ramirez Verdugo, D. (2002), ‘Non-native interlanguage intonation systems: A study based on a computerized corpus of Spanish learners of English’, ICAME Journal, 26, 115-132. Rasier, L. (2003), ‘Le système accentuel de l’interlangue d’apprenants francophones du néer- landais’, in: A. Mettouchi & G. Ferré (eds), Proceedings of the International Conference ‘Prosodic Interfaces 2003’, Université de Nantes, 79-84. Rasier, L. (2006), Prosodie en vreemdetaalverwerving. Accentdistributie in het Frans en Nederlands als vreemde taal, PhD dissertation, Université catholique de Louvain. Rasier, L. (2008), ‘Contrastief prosodieonderzoek Nederlands-Frans. Een contrastief-typolo- gische kijk op de accentuering’, Verslagen en mededelingen van de Koninklijke Academie voor Nederlandse Taal- en Letterkunde, 118-1, 49-67. Rasier, L. & Ph. Hiligsmann (2007), ‘Prosodic transfer. Theoretical and Methodological Is- sues’, Nouveaux cahiers de linguistique Française, 28, 41-66. Rasier, L. & Ph. Hiligsmann (to appear), ‘Contrastive Linguistics, Typology, and Second Language Acquisition. An Integrated Approach to Transfer’, International Journal of Bi- lingualism. Selinker, L. (1989), ‘CA/EA/IL: the earliest experimental record’, International Review of Applied Linguistics in Language Teaching, 27-4, 267-291. Swerts, M., E. Krahmer & C. Avesani (2002), ‘Prosodic marking of information status in Dutch and Italian: a comparative analysis’, Journal of Phonetics, 30-4, 629-654. Trouvain, G. & U. Gute (eds) (2007), Non-native Prosody. Phonetic Description and Tea- ching Practice, Berlin & New-York: Mouton de Gruyter. Vallduvi, E. (1991), The Informational Component, New York: Garland. Wieden, W. (1993), ‘Aspects of acquisitional stages’ in: B. Kettemann & W. Wieden (eds), Current Issues in European Second Language Acquisition research, Tübingen: Gunter Narr Verlag, 125-135. Willems, N. (1982), English Intonation from a Dutch Point of view, Dordrecht: Foris Publica- tions

Recommended publications