Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

For almost 25 years, generative research on second-language acquisition has examined second-language learners’ understanding of various linguistic properties such as island effects (White 1989), subjacency constraints (Schachter 1989), case and tense (Lardiere 1998), IP (Haznedar 2001) and interface properties (White 2009). Since generative SLA research such as this typically looks at the acquisition of some property of grammar that is not explicitly taught in the classroom, little work has been done that investigates how classroom language teachers might accelerate acquisition of these structures in an instructed L2 setting.

In an attempt to address this gap between theory and practice, the current study examines the acquisition of UG-constrained properties related to the syntax/pragmatics and discourse/pragmatics interface. In particular, we examine patterns of topic-comment knowledge among two groups of learners: L1 English speakers and L1 Spanish speakers, each learning the other’s language. Since topic-comment structures are primarily found in spoken language rather than written, they are less commonly taught in L2 classrooms, with the exception of Spanish clitics and clitic placement which feature regularly in SSL lessons.

By taking a balanced look at the difficulties these two groups of learners face in topicalising object nouns in their respective target languages, we hope to show that those aspects of topicalisation which were contextualised and taught explicitly – clitics and clitic placement – were acquired and understood at significantly higher levels than aspects which were not explicitly taught. Thus, this study attempts to create a link between generative theory and the second-language classroom by examining the following two research questions: First, can generative linguistics research inform language teaching? Second, given the complexity of interface properties, what if anything is taught explicitly?

In the next section, we briefly outline key issues surrounding research into interface properties and then discuss phenomena related to topicalisation in particular in Section 6.3. Section 6.4 describes the study and findings, and we conclude with a discussion of implications for the classroom.

2 Interface Properties

At the centre of interface properties is the assumption that languages are complex and that when processing or producing a given construction, there are several domains at work at the same time. For example, the syntax/pragmatics interface is when a construction (syntax) is either acceptable or ruled out given the context (pragmatics/discourse) within which it is uttered. Over the last decade, generative research on second-language (L2) acquisition has given increasing attention to the syntax/pragmatics and/or discourse interface (Montrul 2004; Serratrice et al. 2004; Valenzuela 2005; Lozano 2006; Belletti et al. 2007; Rothman 2009; Slabakova and Ivanov 2011; among many others). Despite this, most generative research into interface properties has, until recently, been theoretical and has not addressed the pedagogical implications of targeting interface properties in the L2 classroom.

Interface properties, such as the syntax/discourse interface, are interesting to generative researchers because they are a subtle property of the grammar that is effortlessly acquired by native speakers of a language despite potentially confusing input. Generative grammar has argued that such language subtleties are acquired with the help of universal grammar which imposes rules and constraints on the input. There has been much literature arguing that explicit ‘form-focused’ instruction is beneficial for the language learner (Doughty and Williams 1998; VanPatten 1996; Benati 2001; among others). However, when dealing with interface properties we have, in a way, form interfacing with meaning interfacing with context resulting in complicated language instruction. Topic-comment constructions, discussed in Section 6.3, provide an interesting diagnostic because the form and meaning aspects of the construction are the subject of explicit instruction in the classroom, but context is not. In the following section, we will discuss topic-comment constructions in English and Spanish and look at the extent to which it can be taught in the classroom.

3 Topic-Comment Structures in Spanish and English

Topicalisation is a property found at the interface between syntax, semantics and discourse-pragmatics. It is a strategy that a speaker uses to reintroduce something previously mentioned in the discourse into the conversation. Thus, a ‘topicalised’ noun, in languages like Spanish and English, is typically moved from canonical object position to the leftmost position in the sentence but is still connected to the position in the clause from which it originated. The topic is the constituent in the left periphery which is connected to the rest of the sentence (Rizzi 1997, 2002). Typically, items in the left periphery are discourse-related, and their occupying the left peripheral position is motivated by the context in which the sentence is uttered. The sentence, then, is the ‘comment’ or what is being said about the topic. For example, in (1) the discourse antecedent is ‘a group of friends’, and one of the group, Juan, is reintroduced into the discourse by moving to left peripheral position in the following topic-comment structure. The topicalised element, Juan, is commented on in the rest of the sentence. Juan is connected to the rest of sentence with the clitic (type of object pronoun) lo which refers back to Juan:

1.

[Context: I have a group of friends that I have known for many years.]

 

A Juan i, loi conocí  en París cuando era estudiante.

 

To Juan, CL I-met in  Paris when was student

 

‘John, I met in Paris when I was a student.’

In this way, the syntax (constituent order; moved constituent to the left periphery) interfaces with the discourse (the context identifies the topic that is being reintroduced and thus topicalised). The topicalised element is set apart from the rest of the sentence by an intonational fall, and there is also a connectivity requirement between the topic and the comment. This topicalisation by means of left dislocation is licensed through discourse properties, thus the syntax-discourse interface. In Spanish, as in other languages with clitics (Italian, Greek, etc.), a clitic in the lower clause connects the topicalised element to the rest of the sentence and links the topic to the comment thus ‘clitic left dislocation’ (CLLD) as in Cinque 1990. Topic-comment structures, a form of topicalisation, in Spanish, are expressed as clitic left-dislocation (CLLD) constructions, as in (1).

In English, on the other hand, there are no clitics with the same object pronoun value as in Spanish. In the case of English, then, the non-clitic topic-comment structure is referred to as ‘contrastive left dislocation’ (CLD) as in (2) (Anagnostopoulou 1997):

2.

[Context: I have a group of friends that I have known for many years.]

 

John, I met in Paris when I was a student.

Interpretation of a left-dislocated topic in Spanish is dependent on the semantic notion of specificity (Liceras et al. 1992; Arregi 2003; Valenzuela 2005, 2006). Namely, the presence or absence of the clitic in the comment can result in a change in the interpretation of the topicalised element itself as being either specific or non-specific. When the clitic is present, as in the CLLD construction in (1), the topic is interpreted as specific, but in the absence of a clitic in the lower sentence, the topic is interpreted as non-specific, as in (3):

3.

[Context: I eat fruit and vegetables in order to stay healthy.]

 

Manzanas,

como todos los días.

 

Apples,

   I-eat all the days

 

‘Apples, I eat everyday.’

Following Uriagereka (1995), we assume that the clitic carries a [+specific] semantic feature which, when connected to a left-dislocated referent, results in a difference in interpretation of the topicalised element. That is, the presence or absence of the clitic renders the topic either specific or non-specific. In this way, Spanish CLLD constructions are interpreted as having a specific topicalised element whereas CLD constructions are interpreted as having a non-specific topicalised element. With respect to specificity and topic-comment structures, the syntactic module of the grammar interfaces with the semantic module of grammar which is found in the semantic feature carried by the clitic. To summarise, CLLD = clitic = specific topic, while CLD = no clitic = non-specific topic.

Topic-comment structures are, therefore, very complex constructions that involve at least three modules of the grammar interacting simultaneously: syntax (clitic placement, left-dislocation construction), discourse/pragmatics (licensing the topic-comment) and semantics (specificity of the clitic). The complexity of topic-comment structures can result in rather confusing input since the appropriateness of CLLD (specific topic) or CLD (non-specific topic) is derived from the interaction of context with the syntax rather than syntax alone. That is, the learner’s task is to determine that while both constructions are ‘correct’, one or the other construction is ruled out based on the context alone. While learners receive form-focused and meaning-focused instruction, it is context-focused instruction that is additionally required for a complete examination in the L2 classroom. Context-focused instruction would allow the teacher to clarify on the appropriateness of these construction based on the context in which they are uttered.

3.1 Learnability and Interface Properties

Complexity results in confusing and potentially ambiguous input which in turn has been shown to be problematic, even at very advanced stages in development, in many different types of language acquisition contexts: simultaneous bilingual acquisition (Müller and Hulk 2000; Montrul 2004, 2010), post-childhood acquisition (Hertel 2003; Lozano 2006; Valenzuela 2005, 2006) and L1 language loss (Tsimpli et al. 2004). As discussed in the previous section, topic-comment structures are particularly complex given the fact that they involve not only the syntax/semantics interface (clitics and clitic placement interacting with specificity) but also the syntax/discourse interface (clitic left dislocation and the licensing of the topic-comment structure). Thus, the question remains, what does a second-language (L2) learner have to acquire in order to successfully use and interpret topic-comment structures in Spanish or English, and how can that acquisition be supported in the L2 classroom? That is, what is the learnability issue? In the present section, we will examine the three modules involved in topic-comment and the extent to which they are addressed in the classroom.

The generative analysis of topic-comment structures sees Spanish topic-comment structures as having essentially three modules of the grammar which interact with one another simultaneously. First, the syntax module involves the left-dislocation construction, clitics (object pronouns) and the correct clitic placement in the sentence. While clitics are interpreted as object pronouns, they do not appear in the same position as object pronouns as in English. Thus, there cannot be positive transfer from English with respect to the syntax (Chap. 2 by Bruhn de Garavito, this volume). Second, the semantic module involves specificity. The [+specific] feature of clitics is a semantic feature which is encoded in the clitic itself, and so the presence or absence of the clitic can derive a specific or non-specific interpretation. When a clitic is coindexed with a topicalised element, that topic is interpreted as ‘specific’. Conversely, when the topic appears without a clitic, the topicalised element is interpreted as non-specific. Third, the discourse module involves licensing the reintroduction of a discourse antecedent (old information) into the discourse.

With respect to English topic-comment structures, there are two modules of the grammar that interact. The syntax module involves left dislocation of the topicalised element, while the discourse module involves licensing the reintroduction of old information into the discourse.

With respect to learnability, an L1 English learner of L2 Spanish will hear both the CLD and the CLLD forms in the input but will have to learn the subtle interpretive differences (specificity distinction) from the semantic feature associated with the clitic. The semantic feature of specificity is encoded in the clitic but the learner will have to sort out the appropriateness of the clitic or non-clitic form from hearing it in context. Moreover, topicalisation is not a high-frequency construction, and so the learner will have to be exposed to extensive input in order to derive the specificity contrast. Transfer from the L1 English will give them the CLD structure and associated syntactic constraints but will not give them the specificity distinction since English does not have clitics. On the other hand, L1 Spanish learners of L2 English have to lose the interpretive distinction associated with the presence or absence of clitic since English does not encode specificity semantically in topic-comment structures. Given these distinctions, classroom instruction that includes highly contextualised examples of the various structures, combined with explicit explanations of why one structure is preferred over another in a given situation, would promote acquisition of the target forms.

To summarise, an L1 English learner of L2 Spanish will have to acquire the syntax (CLLD structure), a syntax-semantics interface property where the presence or absence of the clitic can result in an interpretive difference and a syntax-discourse interface property whereby the context of the utterance will require either the clitic or non-clitic forms. An L1 Spanish learner of L2 English will transfer the CLD structure from their L1 and learn to extend it to contexts in which an L1 English speaker would not use it. In the following section, we will discuss the results from a bidirectional study conducted on the acquisition of topic-comment structures in both English and Spanish that tested CLLD and CLD constructions and their interpretive properties.

4 Methodology

4.1 Research Questions

We have seen that topic-comment structures in both English and Spanish involve interface properties and, therefore, present a level of complexity that can potentially pose difficulties for a language learner and/or acquirer. The data and results presented in this section reflect this complexity and lead us to suggest that overt instruction of interface properties can, in fact, be explicitly taught.

4.2 Participants

For this chapter, we are re-examining a subset of data from a larger study (Valenzuela 2005) which was bidirectional (study 1: L1 English/L2 Spanish; study 2: L1 Spanish/L2 English) and had two participant groups, respectively. Participants for study 1 consisted of 15 L1 English speakers of L2 Spanish who had had their first exposure to Spanish after puberty (i.e. post-childhood) in a foreign language classroom setting. Based on self reports, they learned Spanish via a mix of communicative and form-focused instruction. Participants were from England, Canada and United States and were living in Spain at the time of testing. The L2 participants were end-state, near-native speakers of L2 Spanish. Near-nativeness was established following similar procedure to that implemented in White and Genesee (1996) and Montrul and Slabakova (2003) whereby speech samples from all participants (controls and L2) were extracted from short oral interviews. The speech samples were randomised and both L1 and L2 Spanish speakers were mixed together. The tapes were given to two impartial native Spanish speakers who were asked to listen to the speech sample several times, each time evaluating the ‘nativeness’ of the individual speaker for syntax, morphology, pronunciation, vocabulary and overall fluency on a scale where 1 = least like a native speaker of Spanish and 10 = native speaker of Spanish. Average scores for the speech samples of the L1 Spanish speakers ranged from 8.5 to 10. Based on the native speakers’ score margin, L2 speakers whose average scores were between 8.5 and 10 were deemed near-native. Their score on the near-nativeness interview together with the age of first exposure and number of years living in a Spanish environment was the criteria for inclusion in the study. In addition to the L2 Spanish/L1 English experimental group, 25 monolingual L1 Spanish participants were tested in Spain as a control group. Participants for study 2 underwent the same criteria for inclusion in the experiment. Participants for this study consist of 17 Spanish speakers of L2 English who had had their first exposure to English after childhood in an EFL setting and also received a mix of communicative and form-focused instruction. Participants were from various Spanish-speaking countries and were living in either Canada or Spain at the time of testing. In all cases, their work was conducted in English, and in most cases both work and home life were conducted in English. The L2 participants were near-nativeFootnote 1 speakers of L2 English. In addition to the L2 English/L1 Spanish experimental group, 15 monolingual L1 English participants were tested in Canada as a control group. A summary of the participant information is shown in Table 6.1 above.

Table 6.1 Summary of participant information
Full size table

4.3 Tasks

We report on results from two tasks for studies 1 and 2, respectively, each of which targeted topic-comment constructions in both specific and non-specific contexts. The tasks were an oral sentence selection task (comprehension data) and a Sentence Completion Task (written production data).

4.3.1 Sentence Selection Task

The sentence selection task provides comprehension data for the specific and non-specific interpretation of topic constructions and the appropriateness of a topic-comment structure in a given context.

In this task, participants listened to a context story and were asked to select the most appropriate concluding sentence. Context stories forced either a specific or a non-specific interpretation of the target topic in the concluding sentence. For example:

Spanish:

  • Lola está haciendo los deberes de la universidad pero se acaba de dar cuenta que le faltan unos apuntes importantes. Mira por todas partes en la biblioteca, en su habitación, y en la clase pero…

    a.

    Esos apuntes, no encuentra.

    		 

    b.

    Esos apuntes, no los encuentra.

    desired response

    c.

    Ni a ni b

    		 

    d.

    Ambas a y b

    		 

English:

  • Lola is doing her homework. However, she just noticed that she is missing some important class notes. Lola looks in the library, in her room and in the classroom but…

    a.

    Those class notes, she cannot find.

    desired response

    b.

    Those class notes, she cannot find them.

    		 

    c.

    Neither (a) nor (b).

    		 

    d.

    Both (a) and (b).

    		 

4.3.2 Sentence Completion Task

The Sentence Completion Task provides written production data for the specific and non-specific interpretation of discourse context, left-dislocated topic constructions and correct placement of clitics (in the case of the Spanish tasks).

For this task, participants read a context story and were then presented with a sentence that was begun, and they were asked to complete it. Context stories forced either a specific or a non-specific interpretation of the target topic in the concluding sentence. For example,

Spanish:

  • El Sr Fernández ve mucho la televisión. No ve programas de deportes sino ve programas policíacos y documentales. Le pregunto por qué no ve programas de deportes y me explica:

  • ‘Deportes, porque los encuentro increíblemente aburridos.’

English:

  • Mr. Fernández watches a lot of television. He does not watch sports but rather he watches detective shows and documentary programmes. I ask him why he never watches sports and he explains:

  • ‘Sports, because I find them incredibly boring.’

Given the complex nature of topic-comment constructions, we predict that our tests will have higher levels of accuracy on aspects of the constructions for which explicit instruction is received. Specifically, learners of L2 Spanish will be more accurate with the CLLD structure and clitic placement (syntax) than with the specificity distinction. L2 learners of English will have trouble letting go of the specificity distinction from their L1 Spanish.

4.4 Results

The data from study 1 (L2 Spanish) are presented in sections 6.4.4.1 (Sentence Selection Task) and 6.4.4.2 (Sentence Completion Task). The data from study 2 (L2 English) follow in sections 6.4.4.3 (Sentence Selection Task) and 6.4.4.4 (Sentence Completion Task).

4.4.1 Study 1, L2 Spanish: Sentence Selection Task

In Table 6.2 we see a summary of the results for the Spanish sentence selection task.

Table 6.2 Sentence selection task: % choice of sentence with clitic
Full size table

As can be seen in Table 6.2, both groups chose the sentence with the clitic (CLLD construction) in [+specific] contexts the majority of the time. However, in [−specific] contexts, where the clitic option should be ruled out, the L2 Spanish group chose the clitic option significantly more times than the L1 Spanish control group. A two-way ANOVA shows that there is a between-group significant difference in both the L1 Spanish group’s and the L2 Spanish group’s treatment of specific versus non-specific topics with respect to preference for the ‘clitic’ responses (L1 Spanish (F(1, 88) = 435.044, p < 0.01; L2 Spanish (F(1, 56) = 66.936, p < 0.01))).

4.4.2 Study 1, L2 Spanish: Sentence Completion Task

In Table 6.3 we have a summary of the results for the Spanish Sentence Completion Task.

Table 6.3 Sentence Completion Task: % suppliance of clitic
Full size table

As can be seen in Table 6.3, both groups correctly provided clitics with specific left-dislocated topics in main clause environments. The L2 Spanish group, however, differed from the L1 Spanish group in [−specific] contexts where they provided a clitic significantly more often than the control group. But they are also distinguishing between [±specific] contexts. A single-factor ANOVA showed no significant difference between groups for topics in [+specific] contexts (F(1,38) = 6.441, p = 0.015). However, in [−specific] contexts a single-factor ANOVA showed a significant difference between the groups (F(1,38) = 19.113, p < 0.01). The tendency on the part of the near-natives was to provide a clitic in non-specific contexts which is consistent with their results in the previous tasks and indicates that they are not totally distinguishing specificity.

The tendency in the L2 Spanish group is to prefer the clitic in [−specific] contexts (53 %) whereas the L1 group preferred no clitic (17 %). In both production and comprehension, the L2 Spanish group accepts and produces topic-comment structures as well as provides correct placement of the object clitic. Thus, the syntax appears to have been acquired, but the specificity distinction (interpretive differences) is not target-like and therefore appears to be more problematic for these learners. These results seem to be in line with the notion that explicit instruction facilitates acquisition.

4.4.3 Study 2, L2 English: Sentence Selection Task

In Table 6.4 we see a summary of the results for the English Sentence Selection Task.

Table 6.4 Sentence selection task: % choice of sentence with pronoun
Full size table

As can be seen in Table 6.4, the L1 English group does not distinguish between specific and non-specific contexts for clitics as they treat both contexts the same. The L2 English group, however, prefers pronouns with topic-comment constructions that are in [+specific] contexts. A one-way ANOVA shows that there is a within-group significant difference in the L2 English group’s treatment of specific versus non-specific topics with respect to the preference for ‘pronoun’ responses (specific 30 %; non-specific 7 %) (F(1,30) = 10.054, p < 0.01). This suggests that they are transferring a specificity distinction from their L1 Spanish into their L2 English.

4.4.4 Study 2, L2 English: Sentence Completion Task

In Table 6.5 we see a summary of the results for the English Sentence Completion Task.

Table 6.5 Sentence Completion Task: % suppliance of pronoun
Full size table

In this written production task, we see that the L1 English group is low in suppliance of pronouns for sentences in [+specific] contexts. A within-group ANOVA shows no significant difference in the L1 English group’s treatment of specific and non-specific tokens (F(1,28) = 2.514, p = 0.124). While the L2 English group supplied topic constructions with pronouns in [+specific] contexts (mean 54 %) more often than in [−specific] contexts (mean 36 %), a within-group single-factor ANOVA does not show a statistically significant difference in their suppliance of pronouns on the specific versus non-specific tokens (F(1,32) = 3.295, p = 0.079).

5 Discussion and Implications for the L2 Classroom

The data for both the L2 Spanish and L2 English studies show the following for the three modules of the grammar involved in topic-comment constructions. With respect to the syntax module, the L2 Spanish group is both accepting and producing CLLD constructions, and they have correct word order for clitics. The L2 English group also correctly accepts and produces CLD. This accuracy can be attributed to the traditional form-focused instruction they received in their home countries. A high level of syntactic accuracy on topicalised object NPs is expected from explicit instruction and positive evidence. As for the semantic module, the L2 Spanish group indeed differentiated between [±specific] contexts. However, there was an overall tendency to overproduce/prefer the clitic in non-specific contexts. Oversuppliance of clitics across tasks in contexts where the no-clitic option is preferred by L1 speakers reflects a lack of native-like sensitivity to the semantic feature of specificity. This failure to differentiate between the [±specific] feature may be the result of over instruction of the cliticised option (see Selinker’s 1972 five processes of fossilisation), as instructors may judge this a more ‘difficult’ structure to learn. This is an example of a surface feature (placing of the clitic with the object NP), being singled out for practice, while the more central, UG-based constraints on specificity are overlooked. An informed instructor would realise the importance of not judging a structure as ‘difficult’ simply because it has a seemingly more complex rule (i.e. add clitic to fronted object NP). Instances of interpreting a topicalised element as non-specific ([−specific]) which require that the clitic not be placed with the left-dislocated NP are equally and arguably more difficult to master. Knowing when to leave out an element is just as important as knowing when to include one. Finally, with respect to the discourse module, topic-comment structures were both accepted and appropriately produced. However, in both cases, the notion of topic-comment can be positively transferred from their L1.

Let us return to our research questions. Research question 1 asked how generative acquisition research can inform language teaching. In this chapter, we have summarised results from a study conducted under the generative framework on the L2 acquisition of an interface property. What was shown was that both groups of L2 speakers had native-like performance of certain aspects of the syntax of left dislocation. The SSLers accurately judged and produced clitics, and the ESLers were native-like in the construction of left dislocations. In contrast to this accuracy, both groups diverged from the L1 speakers judgments on the discourse feature. That is, despite grammatical accuracy, they were non-target-like in their understanding of the specificity feature, resulting to overuse of [+specific] judgements. That is, they were grammatically accurate but contextually unacceptable. Identifying this mismatch between syntax and discourse properties is an example of how generative linguists can contribute to increased L2 learning outcomes by drawing classroom teachers’ attention to how L1 transfer of learners’ understanding of specificity can contribute to non-native-like competence in this domain. Once teachers are aware of the semantic notion of specificity, they can then push their learners to notice which way they must move to acquire target-like forms, namely, L2 Spanish learners must acquire the feature, and L2 English learners must lose the feature. Judging from this mismatch, we can see that in addition to receiving explicit instruction on the syntax of left dislocation, learners also need instruction on specificity, rather than simply being left to infer an understanding of this feature from ambiguous and possibly infrequent input regarding the contexts in which CLLD or CLD are appropriate.

Research question 2 was that, given the complexity of interface properties, what if anything can be taught explicitly? Explicit instruction is provided for the syntax in these complex structures, and with that knowledge comes the [+specific] feature of the clitic (in the case of CLLD constructions). While explicit instruction is only provided for the clitic placement, frequency of exposure to the language over time allows the learner to acquire sensitivity to the semantic notions. That is, with increased exposure, the frequency with which the learner will hear a particular pattern in the input (e.g. CLLD construction uttered in a [+specific] context) will increase thereby leading to convergence on TL sooner. It is here that an informed classroom teacher can provide an increased amount of input for learners. By understanding the specific/non-specific distinction himself/herself, the instructor is in a position to provide carefully selected, authentic input that increases learners’ opportunities to process the syntactic forms and increase their connection to meaning and interpretation (VanPatten 1996).

6 Conclusion

In this study, we have attempted to demonstrate that syntax – in this case left dislocation – can be acquired to an advanced level through classroom instruction. However, we have also tried to show that, despite their syntactic accuracy being judged native-like by linguistically naïve native speakers of the given language, these informants exhibit non-target-like judgments in their knowledge of a less obvious linguistic property, in this case the specificity feature. By going into classrooms, working with teachers and collecting real learner data, generative linguists, with their insights into underlying properties of language that typical L2 teachers may not be aware of, have much to offer second-language classroom instructors.