How Multiple Regression Models Can Be Written to Better Reflect the Complexity of First and Second Language Acquisition Research: An Attempt to Limit Type VI Error

Multiple Linear Regression Viewpoints, 2014, Vol. 40(2) 41 How Multiple Regression Models Can Be Written to Better Reflect the Complexity of First and Second Language Acquisition Research: An Attempt to Limit Type VI Error Kyle Perkins Isadore Newman Florida International University The purpose of this paper is to show how multiple regression models can be written to investigate and to better reflect the complexity of four major research themes in first and second language acquisition research in order to limit Type VI error. The four major themes were selected because of their importance to their respective fields, their multidimensional nature, and the interdisciplinary research influence on them. Each theoretical perspective examined in this paper brings with it its own research methodology, and; thus, the need for a paper that demonstrates how multiple regression models can be written to investigate four of the major foci of first and second language acquisition research to better reflect the complexity of language acquisition research. econd language acquisition is an interdisciplinary, multidimensional field whose principal foci include the study of how second language learners create a new language with scant exposure to the target language and the description of what is learned of the second/target language and what is not learned (Gass, 2013). Researchers from a broad spectrum of disciplines (e.g., education, linguistics, psychology, sociology, psycholinguistics, sociolinguistics, discourse analysis, neurolinguistics, and conversational analysis) conduct second language acquisition research. The first language acquisition topics shown below are also multidimensional and have been investigated with interdisciplinary approaches. First and second language acquisition data were chosen as examples for which multiple regression models could be constructed and shown in this paper because first and second language acquisition involve non-linear, multidimensional, and multiplicative phenomena. The following examples are offered to illustrate this point. First and second language differences can predict the amount of avoidance that learners show in using particular linguistic structures. First and second language similarities can result in differential learning rates, different learning paths, an overproduction of first language-influenced forms, and predictability/selectivity. Semantic and conceptual facts about the first language can affect a learner’s choice of forms in the second language. Some individuals are more successful in learning a second language than others. Some factors that may be responsible for these differences include age (i.e., the Critical Period Hypothesis), aptitude, motivation (which may change over time), attitude, affect, anxiety, social distance (perceived or real), learning strategies, extroversion and introversion. English Language Learners develop hypotheses (consciously or unconsciously) regarding the rules of the second language and create rules that may vary according to the context of use. Linguistic principles reflect the infinite possibilities of human language creation and the limits of human language variation. English Language Learners create their own language systems (i.e., an Interlanguage), which is composed of elements from the learner’s first language, the second/target language, and elements that do not originate from either the target language or the native language, and a learner’s Interlanguage is in constant flux. There is individual variation in learner data and individual data can be obscured with/within group data. The paths of acquisition are not always identical for learners from different language groups/language families. The second language research topics listed above are discussed in Gass (2013) where more information on each of these topics can be found.