On Friday, October 22nd (2:20-3:40pm), Dr. Luiz Amaral (UMass Amherst) will give a lecture entitled "How to take full advantage of the Feature Reassembly Hypothesis in SLA? Pay attention to how you describe features."
This event, co-sponsored by CLAS and the Dept. of Spanish and Portuguese is free and open to the public and will be presented via Zoom (https://osu.zoom.us/j/93435947384?pwd=aHYvb3VkVkc2dFBvbWhnZkhmSkltUT09).
SPPO graduate students will meet in Hagerty Hall 042 to participate as part of colloquium.
Talk abstract:
In the generative tradition, features have become a key element to understand the licens ing of linguistic structures and the differences from one language to another (e.g., Chomsky, 1995; Adger, 2003). In second language acquisition (SLA), formal features have also become a core construct to explain how L2 speakers represent language properties in their interlanguage (e.g., Lardiere, 2006; Liceras et al., 2006). In terms of learnability under Minimalism, features dene the properties of functional categories that need to be acquired. Ultimately, the use of features to describe the interlanguage allows for detailed observations of its different properties. However, feature specification is only as helpful as the syntactic theory it is associated with. As Travis (2006) reminds us: "Features allow us to view old processes in a new light, but when mishandled, they can lead to systems with little predicative power" (p.24). Much of the SLA research on feature reassembly has focused on the description of specific functional items and how L2 learners need to rebundle" such features when acquiring a new language (Lardiere, 2017, 2009). Despite the central role features play in L2 acquisition, very little is explicitly said about the internal architecture of feature bundles in SLA papers. In this presentation I discuss the results of two research projects in SLA with colleagues Andrea Faber (Kansas State University) and Candido de Oliveira (CFET-MG), where we discuss two models for feature architecture in syntactic theory and their possible descriptive power for SLA.
The first one is presented by Adger (2006, 2010), who introduces different possibilities for feature descriptions under Minimalism, including some options that are frequently adopted in SLA research, where features are either represented by atomic values or by attribute-value pairs that impose boolean values. According to him, nothing in Minimalism prevents an enriched feature system, where (i) a feature is an ordered paired hAtt: Vali, (ii) Att is drawn from the set of attributes, fA, B, C, . . . g, (iii) and Val is drawn from the set of values f+, {, a, b, . . . g. Such system could generate lexical entries such as: fD, pro:+, case:nom, num:sing, pers:3, gender:mascg1. Based on Adger's work, we discuss the results of an experiment with the acquisition of nominal agreement using nonce words in Spanish as an L2 by L1 speakers of Portuguese and English (Faber and Amaral, in prog). We argue that using complex attribute-value pairs for feature specification allows us to observe feature reassembly as a two step process where positing new attributes for lexical classes and specifying values for individual items are two fundamentally different processes. The second architecture is the one adopted by Head-Driven Phrase Structure Grammar (Pollard and Sag, 1994; Sag et al., 2003), where the authors argue that features used in complex attribute-value matrices to describe linguistic objects can be hierarchically classified using an inheritance system that reflects their properties. Taken as a model of L2 language representation, this system allows us to make predictions about the acquisition (and re-assembly) of different features on different hierarchical levels. We use the intuitions of this architecture to explain data from an experiment with L1 Portuguese speakers acquiring English resultatives (Amaral and Oliveira, in prog). We argue that establishing the subclass of English adjectives that are allowed in resultative constructions is problematic for Portuguese speakers because feature reassembly tends to happen at higher hierarchical levels that are perceived by the L2 learner as natural classes, which in turn results in overgeneralizations as the one found in our data.
Overall, we argue that an analysis of feature architecture (not limited to observations on feature bundles) allow us to propose more detailed descriptions of the interlanguage, strengthening the arguments in favor of feature reassembly approaches in SLA against deficit-based hypotheses, in line with the ideas presented by Lardiere (2017).
