<p>This thesis looked at the creation and validation of two tests that measure how efficiently English words are learned. Previous studies have created and validated a number of tests that measure the size (how many words are known) and the depth (how well a word is known) of vocabulary knowledge; however, existing vocabulary tests do not indicate how learners can become proficient in vocabulary learning. This research was one of the first attempts to create such tests. A guessing-from-context test (GCT) and a word part test (WPT) were created, because the skill of guessing from context and word part knowledge are teachable and are the most frequently used strategies for dealing with unknown words. The GCT consisted of the following three sections: identifying the part of speech of an unknown word, finding the contextual clue that helps guess its meaning, and deriving the unknown word’s meaning. Each of these three sections was designed to measure each of the important steps in guessing from context that was identified by previous studies. The test was validated using Rasch analysis through data from 428 Japanese learners of English. The results indicated that the GCT is a highly valid and reliable measure of the skill of guessing from context in terms of eight aspects of construct validity (content, substantial, structural, generalizability, external, consequential, responsiveness, and interpretability). Based on the results, two new equivalent forms were created in order to allow a pre- and post-test design where researchers and teachers can investigate learners’ development of the skill of guessing from context. The WPT measured 118 word parts that were selected based on frequency data in the British National Corpus. It consisted of the following three sections: form (recognition of written word parts), meaning (knowledge of their meanings), and use (knowledge of their syntactic properties). These three sections were designed to measure the important aspects of word part knowledge that were identified by previous studies. The WPT was validated using Rasch analysis through data from 440 Japanese learners of English and 1,348 people with various native languages. The results indicated that the WPT is a highly valid and reliable measure of word part knowledge in terms of the eight aspects of construct validity mentioned above. As with the GCT, two new equivalent forms were created in order to allow a pre- and post-test design. For more practical use of the test, the Word Part Levels Test (WPLT) was created by classifying the 118 word parts into three different levels of difficulty. This may allow teachers to quickly examine whether their students need to work on easy or difficult word parts and which aspects of word part knowledge need to be learned. Taken as a whole, the GCT and the WPT are useful measures both for research and practical purposes.</p>
Does high talker variability improve the learning of non-native phoneme contrasts? A replication
