How can we Predict Incidental L2 Vocabulary Learning? A Meta-Analytic Examination of the Involvement Load Hypothesis
Abstract
This dissertation investigated Laufer and Hulstijn’s (2001) Involvement Load Hypothesis (ILH). The ILH claims that the retention of unknown words is conditional on one motivational factor (need) and two cognitive factors (search and evaluation) and predicts the relative effectiveness of activities on incidental vocabulary learning. While research tends to provide general support for the ILH, several studies revealed that the ILH prediction is not always accurate. Aiming to provide a summative evaluation of the ILH and enhance its predictive ability, the present thesis conducted a series of three meta-analytic studies to examine research that tested the ILH.
Chapter 1 outlines the thesis and provides background literature and the rationales for the three studies. Chapter 2 (Study 1) meta-analyzed studies testing the prediction of the ILH to investigate (a) the overall predictive ability of the ILH, (b) the relative effects of different components of the ILH, and (e) the influence of potential factors moderating learning. The results showed that the ILH significantly predicted learning gains. However, each ILH component contributed to learning differently and other factors were found to influence learning, suggesting potential for the ILH to be enhanced.
Chapter 3 (Study 2) aimed to update the ILH to enhance its accuracy in predicting learning. The results of the ILH studies were examined with the information-theoretic approach to determine the optimal statistical model that best predicts learning gains. The results showed that the prediction of the ILH improved by adopting the best operationalization of ILH components and optimal test format grouping and including other empirically motivated variables.
Chapter 4 (Study 3) systematically analyzed incidental vocabulary learning conditions that have been examined in studies of the ILH and calculated the estimated learning gains occurring across different activity types. The results revealed that the estimated mean learning gains were highest for composition-level varied use activities (e.g., composition-writing), followed by sentence-level varied use (e.g., sentence-writing), evaluation (e.g., fill-in-the-blanks), meaning-focused input (MFI; reading and listening) with need for comprehension of target words, and MFI in that order.
Lastly, Chapter 5 provides a final discussion of the thesis, followed by the limitations and potential future directions.