In recent acquisition research (L1, L2 and bilingual acquisition), there has been considerable emphasis on interfaces between the linguistic system and grammar external components (e.g. syntax/discourse), or between different modules of grammar (syntax/morphology; morphology/phonology; syntax/semantics; etc.). This has led to revisions in how researchers account for difficulties and delays in L1 acquisition and ‘success’ or ‘failure’ in bilingual and L2 acquisition. With some exceptions, there has been a tendency to consider at least some interface phenomena as inherently problematic for language acquirers. In this paper, I will provide an overview of this research, drawing comparisons across various acquisition domains. I will suggest that we must be wary of assuming that all interfaces are equally problematic/unproblematic or that different linguistic phenomena relating to the same interface will necessarily behave alike.
I present data from a collaborative project with Rodrigo Romero Méndez(Universidad Nacional Autónoma de México) in which we examine the expression of ‘path functions’ in L2-Spanish descriptions of motion events by native speakers of Yucatec Maya. Following the framework of Jackendoff 1983, path functions are representations of the beginning or endpoint of motion or some route traveled in between, defined with respect to some reference point or ‘ground’ (in the terminology of Talmy 2000). Bohnemeyer (in press) proposesthat path functions may not be encoded at Conceptual Structure (CS) in Yucatec, but only in the image-schematic Spatial Structure system.Jackendoff’s Thematic Relations Hypothesis designates path functionsas a central ingredient of CS; Bohnemeyerclaims that this core component may be language specific. The present study explores second language acquisition as a possible testing ground for this hypothesis.
Bohnemeyer (2007, in press) shows that path functions are not lexicalized in Yucatec and motion is systematically framed as change of location instead. The primary evidence for this claim consists in the same descriptions used to describe motion of a given ‘figure’ with respect to a particular ground also being applicable to scenarios in which the ground moves instead of the figure or in which the spatial configuration changes due to figure or ground emerging or disappearing. This analysis was pioneered by Kita 1999 for Japanese hairu ‘enter’ (or rather, ‘become inside’) and deru‘exit’ (‘become outside’). In Yucatec, it applies to a larger set of verbs. Some verbs are less compatible with non-figure-motion scenarios than others; a pragmatic account of this phenomenonis offered.
The absence of expressions of path functions in Yucatec does not entail that path functions are not encoded in the CS of Yucatec speakers. However, three sources of evidence support the latter hypothesis: lack of ‘fictive motion’ metaphors (Talmy 1996) and temporal connectives with meanings such as ‘after’ and ‘before’ (Bohnemeyer 1998) - both types of expressions are assumed to depend on path functions as their source domain - and the widespread absence of path encoding in the L2 Spanish of Yucatec L1 speakers. The apparent transfer of Yucatec state change semantics to L2 Spanish is demonstrated in Bohnemeyer (in press) solely on the basis of anecdotal evidence. The present study is a systematic investigation of this phenomenon. We compare Spanish descriptions of a set of 60 short animated video clips by six adult native speakers of Yucatec to Spanish descriptions of the same stimuli collected from two control groups: L1 speakers and L2 speakers whose first language is English. Preliminary analysis of the data confirms that path encoding is at most marginal in the descriptions of the Yucatec speakers. We address the potential, limits, and pitfalls of using L2 data as evidence in studies of the language-cognition interface.
A growing body of work in second language acquisition has targeted crosslinguistic differences in the mapping between syntax and semantics. In one domain, L2 research has examined how learners acquire the semantics of aspectual forms such as the imperfective or the progressive (Gabriele, 2005, in press; Montrul and Slabakova, 2002, 2003; Slabakova, 2003). These aspectual forms are of interest in that they show variation across languages, both with respect to the restrictions that they impose on the verb phrases with which they combine as well as the range of meanings that the forms themselves encode. Although the results of this work suggest that L2 learners are generally successful in acquiring the semantics of these forms, several studies have highlighted contexts in which even advanced learners are not as successful, namely those contexts which target coercion (Gabriele, in press; Slabakova, 2002; Slabakova and Montrul, 2007). Aspectual coercion has been defined as an operation which resolves a mismatch between the semantics of the verb phrase and the constraints of another element in the sentential context (De Swart, 1998, 360). One interesting finding is that difficulty with coercion has been found even in cases in which the L1 and L2 exhibit similar properties (Gabriele, Maekawa and Alemán Bañón, 2008; Montrul and Slabakova, 2002; Slabakova and Montrul, 2007). In this talk I will review the L2 research on coercion and consider the significance of these findings for both current models of transfer and ultimate attainment as well as for recent proposals regarding difficulty at interface levels. I will also discuss recent work on the processing of coercion by native speakers and suggest how we might extend this line of research into the domain of L2 development and processing (cf. Brennan and Pylkkänen, 2008; Pickering et al., 2006; Piñango et al., 1999; Pylkkänen and McElree, 2006).
A great deal of psycholinguistic research has recently focused on the interaction between quantifiers and negation. In particular, many studies have attempted to scrutinize the validity of one generalization presented in the child language literature. This is the Observation of Isomorphism, the claim that children’s semantic scope coincides with surface syntactic scope (see Musolino (1998)).
The present talk reviews recent experimental results with children and adults which cast doubts on the conclusion that surface scope plays a crucial role in adults’ interpretation of sentences containing negation and a quantificational noun phrase, but rather suggest that ambiguity resolution is influenced by the context in a transparent fashion (see Hulsey et al. 2004). In the second part of the talk, I consider whether the account proposed by Hulsey et al. (2004), which focuses on the role of contextual factors in ambiguity resolution, applies to adults in the same way it applies to children. I illustrate the results of an experiment carried out in collaboration with Luisa Meroni showing that adults select a different interpretation of the ambiguous sentence in (1), depending on whether that sentence is presented as answer to the question in (2) or to the question in (3).
(1) All the pizzas were
Taken together, the findings documented in the literature and the ones of on-going work suggest that there is no reason to assume a default preference for surface scope interpretations, in children or adults.
I will then illustrate the findings of an experiment conducted with Luisa Meroni and Arjen Zondervan which was designed to investigate whether the role of context as modeled by Hulsey et al. (2004), also extends to a different linguistic phenomenon, namely the computation of scalar implicatures in adults. The results show that adult speakers of English are more likely to calculate the scalar implicature associated with some for (4) more often when that sentence is presented as an answer to (5) than when it is presented as an answer to (6).
(4) Some pizzas were delivered
On the basis of these results, I will argue that the computation of scalar implicatures is contextdependent, in that it tends to take place when it is needed to address the relevant Question under Discussion. Finally, I will use the data from these experiments with adults to propose a new outlook on children’s well-documented non-adult behavior with scalar implicatures.
In recent years, explanations and predictions of developmental delays or inability to reach native-like attainment in specific areas of grammatical knowledge have been linked to the architecture of the language faculty. Under generative theorizing, the language faculty consists of a series of discrete modules (syntax, semantics, phonology), each with their own structural and hierarchical organization, as well as connections between modules or “interfaces” (Jackendoff, 2002). A recurrent recent claim is that linguistic properties at interfaces are inherently more “complex” than linguistic properties internal to a specific domain (syntax, phonology, semantics) due to the integration of different levels of linguistic knowledge/analysis. But even within interfaces, not all interfaces are created equal; some have been claimed to be even more vulnerable or more problematic than others. Sorace & Tsimpli (2006) and Sorace & Serratrice (in press) have argued that the syntax-semantics interface is not problematic and is eventually acquired at the near-native level. By contrast, the syntax-discourse interface presents prolonged difficulty in both L2 acquisition and L1 attrition. Slabakova (2008) agrees with Sorace that the syntax-semantics interface is largely unproblematic but claims that functional morphology is the real bottleneck of L2 acquisition.
After reviewing the rationale and intended scope of the so called "Missing (Surface) Inflection Hypothesis," this paper goes on to compare two recent studies on the acquisition of gender and number agreement in nonnative Spanish (McCarthy, 2007, 2008; Van Espen, 2007). McCarthy's experimental L2 study finds that errors of gender agreement not only occur in both production and comprehension but do so at essentially the same rates for her lower level L2ers. This parallelism in mode leads her to reject an account in terms of the "Missing (Surface) Inflection Hypothesis." Rather, she argues that these findings are best explained in representational terms, by a theory of morphological competence (Lardiere 2005), i.e. not syntactic competence, and she borrows from the Harley & Ritter (2002) feature geometric analysis of pronouns to establish [masculine] as the underspecified (default) gender feature value. This captures the fact that [masculine] is overused in both production and comprehension. Van Espen's experimental study of nonnative Spanish also tests the production and comprehension of gender and number concord. Her results likewise reveal gender agreement to be more problematic than number agreement, again with [masculine] manifesting as the default value. However, contrary to McCarthy's data, Van Espen's lower level L2 data show an asymmetric pattern, with errors of production far greater than errors of comprehension. Scrutiny of the two studies side by side leads to the proposal that the results of McCarthy's study actually point to the comprehension analogue of the "Missing (Surface) Inflection Hypothesis."
We examine the complex interplay between sentence prosody, syntax, and meaning in English and the Romance languages, in particular Spanish and Italian, and explores its implication for second language acquisition (SLA). As is well-known, English has more prosodic plasticity than Spanish and Italian regarding sentence prominence patterns. The latter uses syntactic means where English uses prosody to encode meaning distinctions, such as categorical vs. thetic statements (depending on the presence or absence of a semantic subject-predicate relation) and the focus/presupposition divide (depending on what part of the sentence is under assertion). Nevertheless, we argue that not all prosodic differences between the two language types are of the same nature and that this has important implications for SLA.
More specifically, we argue that two types of prosodic differences must be distinguished: Type 1 difference emerges from the fact that the English Nuclear Stress Rule (NSR) generates prosodic patterns that are a superset from those generated by the Romance NSR, due to deeprooted phonotactic differences between the two types of languages. Type 2 difference emerges from superficial differences between English and Spanish/Italian in the usage of low-level pitch accent deletion associated with previously mentioned or presupposed material --a process with the side effect of triggering Nuclear Stress Shift. While type 1 determines the manner in which a language encodes the thetic-categorical distinction (Sasse 1987), type 2 determines the manner in which a language encodes the focus/presupposition divide (Zubizarreta 1987).
Type 1 difference:
Romance NSR systematically places Nuclear Stress (NS) at the right-edge of the sentence, a prosodic pattern also available in English. Yet English has alternative options in certain structures, for example it can place NS on the subject in SV intransitives. Nava & Zubizarreta (N&Z) in press argue that this difference arises from the fact that in English, but not in Spanish, functional categories may be analyzed as metrically invisible. This property can in turn be linked to the reduced/unstressed nature of English functional words, in particular to those associated with the Infl category (copulas, auxiliaries and modals). In English intransitives, the thetic interpretation is encoded by placing NS on the subject (SV) and the categorical interpretation by placing NS on the verb (SV). Spanish marks categorical statements in the same way as English does. On the other hand, Spanish is distinctly different from English in the manner of marking thetic statements. Spanish marks theticity by resorting to a structure with a post-verbal subject. Although the thetic/categorical distinction does not lead to differences in truth-conditional meaning, unaccusative verbs (due to their semantics) tend to give rise to a thetic-type statement. While unergatives have a variable behavior (depending on pragmatic considerations such as “predictability” and “noteworthiness”), they tend to express a categoricaltype statement more readily than unaccusatives. This gives rise to the generalization that for wide focus contexts, SV structures in English are more common with unergatives than with unaccusatives, while preverbal subjects are more common in Spanish with unergatives than with unaccusatives.
Implications for SLA. The acquisition of prosodic marking of the thetic/categorical distinction by L1Spanish/L2 English learners involves moving from a subset to a superset grammar based on positive evidence (on the Subset Principle in L1, see Manzini & Wexler 1987 and for an application to L2, see Slabakova 2006). N&Z op.cit. have shown that this is achievable by L2ers, but that it is contingent on learners having acquired the relevant phonotactic properties of English, in particular the systematic vowel reduction of functional words (although age-ofexposure might very well be an important factor here).
On the other hand, the acquisition of syntactic marking of the thetic/categorical distinction by L1 English/L2 Spanish learners entails something more than pre-empting the English-specific part of the NSR (i.e. moving from a superset to a subset prosodic grammar). Namely, it requires the acquisition of a word order unavailable in the L1 (i.e. moving from a subset to a superset syntactic grammar). L2 research has shown that L2ers can acquire L2 word order that are distinct from the L1, and Spanish L2 is not an exception. The question is whether L2ers use word order to express the thetic/categorical distinction. While there is no comprehensive L2 study that has investigated the entire spectrum of relevant phenomena, based on currently available investigations, it seems that English natives acquiring Spanish as an L2 do come around to making the distinction in a native-like way. In particular, studies by Hertel 2003 and Lozano 2006 show that advanced learners make a native-like distinction between the two intransitive verb classes with respect to word order, a distinction that reflects the categorical/thetic distinction. A pilot study by Nava 2006 furthermore shows that advanced L2 learners that have a native-like VS usage with unaccusatives never produce English-type NS patterns, suggesting that these L2ers have drop the analysis of Infl as metrically invisible.
Type 2 difference:
While Standard English readily allows for “deaccenting & NS-shift” in wide and narrow focus cases, Standard Spanish does not (Ladd 1996, Cruttenden 1997, Zubizarreta 1998). Thus, in English, a sentence internal narrow-focused constituent that is not in a position for receiving NS by the NSR can nonetheless be made prosodically salient by deaccenting the presupposed material that contains the NS, a process that triggers NS shift onto the sister (focused) constituent. Spanish, on the other hand, resorts to its flexible word order to align the narrow focused constituent with the NS position, which, as mentioned earlier, is located at the right-edge of the sentence. Yet, this difference is not linked or contingent on any other grammatical property of the language.
Implications for SLA. Spanish L1/English L2 learners have no difficulty acquiring “deaccenting & NS-shift” both in wide and narrow focused contexts (N&Z op.cit.). On the other hand, investigation on the acquisition of Spanish by English natives show that these L2 learners never completely abandon their L1 mechanism for expressing narrow focus, in particular narrowfocused subjects (Lozano 2006, Belletti, Bennati & Sorace 2007). We suggest that this preference reflects something more that L1 transfer, namely the fact that such modified L2 grammar is compatible with all other properties of the target grammar, precisely because “deaccenting & NS-shift” is an isolated or superficial parameter. Consequently, this is an area where we expect language change or attrition to easily emerge in a contact situation. We submit the above explanation as an alternative to the one put forth by Sorace 1999, 2000, who proposes that grammatical forms that engage discourse-based (or “interface”) notions are without exception difficult to acquire by second language learners and vulnerable to loss in the case of bilinguals.
Mind-Context Divide Workshop
The workshop is made possible by generous funding from the International Programs Major Projects fund, The College of Liberal Arts and Sciences’ Bond Funding, the Vice President of Research, as well as the Co-Sponsorships of FLARE/SLA Ph.D program, the Iowa Center for Developmental and Learning Sciences and the Departments of Linguistics and Spanish and Portuguese.