HLT4LL: Nick Ellis

Nick Ellis is a Professor of Psychology, Professor of Linguistics, and Research Scientist in the English Language Institute, at the University of Michigan. His research interests include language acquisition, cognition, emergentism, corpus linguistics, cognitive linguistics, and psycholinguistics. His research in second language acquisition concerns (1) explicit and implicit language learning and their interface, (2) usage-based acquisition and the probabilistic tuning of the system, (3) vocabulary and phraseology, and (4) learned attention and language transfer. His emergentist research concerns include language as a complex adaptive system, networks analysis of language, scale-free linguistic distributions and robust learning, and computational modeling. He serves as General Editor of Language Learning. His homepage is here.

Second Language Cognition
This paper describes and illustrates 8 principles of Second Language Cognition that have emerged over the last 20 years of research in the hope that these might usefully prompt consideration of the role of Human Language Technology for Language Learning.

1 The fundamental premise is that language is intrinsically symbolic, constituted by a structured inventory of constructions as conventionalized form-meaning pairings used for communicative purposes. Language is a Complex Adaptive System: it emerges from usage and is acquired from usage.

2 Constructions vary in abstraction from concrete, particular items (words and idioms) to schematic classes (as in word classes or abstract syntactic constructions). Constructions may be simultaneously represented and stored in multiple forms, at various levels of abstraction (e.g., concrete item: table+s = tables and [Noun] + (morpheme +s) = plural things).

3 Abstract constructions are meaningful linguistic symbols in their own right, existing independently of particular lexical items. Nevertheless, constructions and the particular lexical exemplars that occupy them attract each other. Grammar, lexis, and meaning are inseparable. They resonate.

4 Construction learning, like other aspects of cognition, involves processes of perception, attention, categorization, schematization, and memory.

5 Competence and performance emerge as a frequency-tuned conspiracy of memorized exemplars of use of these constructions. Competence is the integrated sum of prior usage, performance is its dynamic contextualized activation.

6 Development is gradual, moving from an initial heavy reliance on concrete items to more abstract linguistic schema. This process is crucially dependent on the type and token frequencies with which particular constructions appear in the input. Storage of wholes depends on token frequency, development of abstract linguistic schema depends on type frequency. Zipfian distributions of language emerge from language usage, and, in turn, make language learnable and robust.

7 Language acquisition is affected by attention. Salient forms are better attended and better acquired. Cues learned early in learning block the acquisition of later-experienced forms. L1 learned-attention and entrenchment thus limit the endstate of usage-based SLA.

8 These limitations can be overcome by recruiting learner consciousness, putting them into a dialectic tension between the conflicting forces of their current stable states of interlanguage and the evidence of explicit form-focused feedback, either linguistic, pragmatic, or metalinguistic, that allows socially scaffolded development.