Retention of Multiword Strings and Meaning Derivation From L2 Reading

Abstract

This research aimed to find out the extent to which multiword strings met while reading for information a text in another language (L2) were retained, and whether such retention was related to unknown vocabulary density, vocabulary knowledge, and text comprehension. A common situation in L2 reading is one where the text contains some unknown words. In this research the unknown vocabulary density was controlled. The research also included a meaning study that focused on guessing the meanings of new words from context and the retention of these meanings. By multiword strings (MWSs) were meant phrases of any kind consisting of more than one word which may or may not be multiword units (MWUs), the fixed, or semi-fixed expressions consisting of more than one word such as collocations, idioms, etc. The control over the unknown vocabulary density was attained by adapting the reading text (a newspaper feature article in English of about 1,000 words long) and by including in the analyses only participants with a certain level of vocabulary knowledge. Except for the target unknown or new words (comprising either 2% or 5% of the English text) that were replaced by pseudo-words, the adapted text was within the first 2,000 words in English. From the 242 Indonesian learners of English who took part in the study, only those showing mastery of at least the 2,000 word level were included in the analyses. The target words were the words with the lowest frequencies in English and each one belonged to a word family occurring only once in the text. The participants read the text for information, returned the text and were immediately tested on one of these three measures: retention of the meanings of new words, guessing from context, or retention of the target multiword strings in the text. Afterwards they were handed back the same text and completed the comprehension test. No significant difference was found on all three measures as well as on text comprehension between the groups reading the 2% text and those who read the 5% text. For both text groups, guessing the meaning of new words from context was not an easy task; on average only 17% of the new words had the meanings worked out from context. On a more sensitive (select-answer) test, around 41% were correctly identified. Retention of the meanings of these new words could only be captured by the more sensitive test; on average 31% of the new words' meanings were retained. Correct responses to around 50% of the comprehension items were recorded. Two kinds of MWSs were under examination: ones composed of familiar words, and ones containing a new word that was replaced by a pseudo-word. Around 10%-11% of the MWSs with familiar words were retained compared to about 29-30% of those containing a new word. The retention of familiar-word MWSs was positively related to vocabulary knowledge, while the retention of MWSs containing a new word was positively related to both vocabulary knowledge and text comprehension. Text comprehension, in turn, was positively related to vocabulary knowledge. Taking this result of MWS retention as an indicator of incidental MWU retention from information-focused L2 reading, it is reassuring that there was incidental MWS retention, and that the more advanced the learners (who presumably have better vocabulary knowledge and better text comprehension), the more MWSs they retained. This suggests that the more advanced the learner, the more L2 MWUs are retained as a by-product of reading for information. However, it was a weak kind of retention in the form of recognition recorded immediately after exposure. The chance of repeated exposure necessary to keep such retention from fading away is unfortunately slim. Corpus studies show that while a large portion of English native speakers' production consists of various multiword units (MWUs), in general individual MWUs have low frequencies. Even the few highly frequent MWUs occur far less frequently than frequent words. These findings partly explain why L2 phraseological competence develops extremely late if at all, and suggest the need for a more direct learning of MWUs. Unlike the case with high and moderate frequency vocabulary items, where learners' knowledge of them can be fine-tuned through repeated exposure to the same items in different contexts, the acquisition of most MWUs cannot rely on repeated exposure. While the sheer number and the complexity of MWUs do not make it feasible for all MWUs to be subjected to direct learning, learners can be trained early on to spot and pay attention to MWUs and their restrictions as well as variations as they appear in natural language.