This article examines a research study about lexical proficiency and its implications for teachers. The study begins by introducing some definitions of lexical proficiency. Researches generally agree that the term is not easy to describe and that more research is needed for this area of language learning.
Previously, the use of computational methods has been used to analyze lexical proficiency, but this current study emphasizes human judgments for a better understanding. There is general agreement among researchers about using global models for assessing lexical proficiency. The models use breadth and depth of lexical knowledge. However, there are exceptions when using these two dimensions. For example, lexical properties, word concreteness, among others, does not fit into these dimensions. The research says that in natural language production, there are two categories namely lexical diversity and lexical sophistication. However, arguments still persists as to what words fit into which category. The research goes on to say that more research is needed to refine the lexical features for lexical proficiency.
Previous studies of lexical proficiency
Previous research established three ways to measure lexical proficiency. Longitudinal changes measures lexical proficiency over time. Then there is the difference between proficiency levels which focuses on beginners, intermediate and advanced learners. As for human ratings, it uses a holistic approach that predicts collocation accuracy, lexical diversity, word frequency and specificity. For high levels of proficiency, this is indicated by a wide range of lexical diversity, the use of uncommon words and more general words. This research examines holistic scores and the relationship with analytic scores of individual lexical features as it pertains to breadth, depth and core lexical items.
Two analyses were investigated. Samples were taken from both first and second language learners from a variety of linguistic backgrounds. The samples examined analytic and holistic scores from a corpus of written and speaking samples to determine similarities and differences. Two hundred and forty samples were taken for writing and a similar two hundred and forty samples were taken for the spoken corpus.
The survey instruments were divided into analytic and holistic evaluations. In all, ten items made up the survey instrument. Some examples of these items include basic category words, collocation accuracy, word frequency and lexical diversity. Data collected with these instruments were statistically analyzed with holistic scores as dependent variables and analytic scores as independent variables.
The study revealed that collocation accuracy, lexical diversity and word frequency can predict lexical proficiency for both speaking and writing but word concreteness is a good indicator for speaking. Or it can be said that collocation accuracy indicates depth of knowledge, while lexical diversity and word frequency shows the breadth of knowledge. Word concreteness demonstrates the use of core lexical items.
Further, the ability to use collocation demonstrates a higher level of lexical proficiency and is a major indicator for fluent communication. For lexical diversity, it was found that learners with a wide resource of vocabulary were more lexically proficient. Learners who produced less frequent words, were also judged the same way. Word concreteness, which is in the model for spoken lexical proficiency, suggested that less concrete words showed greater lexical proficiency. Two writing samples of a native speaker and an L2 learner were given to show the differences in proficiency levels by human rating. A qualitative analysis of the two writing samples revealed that collocation accuracy is not as evident in the L2 learner as in the native speaker sample. The use of phrasal verbs and other multi-word units were often used by a more proficient learner. It showed that it is not just how many words are known but it is how the words are organized and to what degree that makes a learner lexically proficient. .
Why is this research important?
The study showed that collocations support perceptions of accuracy. Human judgment is still preferable to automation when measuring collocation accuracy. It is difficult for machines to read how words are used in context using natural language. Teachers can put this knowledge in perspective to understand that lexical proficiency does not come as a result of a series of mechanical operations but rather something that is natural sounding. Statistical analysis has shown that holistic scores are more reliable than analytic scores in this kind of measurement. That is why collocation accuracy is a major feature for determining lexical proficiency.
What can we use the research study for?
Collocation is one of the features that make people think that someone is more lexically proficient. The study confirms this. It is about knowing how to use words and in what manner. To be lexically proficient is to speak and write like a native. If a student can do this, there is more reason to believe that the student can produce fluent language. This understanding can help teachers design a curriculum that focuses on specific features of language chunking. These multi-word units will fulfil a depth of knowledge that a breadth of knowledge lacks.
How can we use this research study for the classroom?
Assessing lexical proficiency implies knowing its features. Since collocation accuracy is a key component of it, lesson plans should incorporate the use of different types of collocation. Students will be better equipped to produce fluent language. That is why the use of multi-word units, like phrasal verbs and other combination of words can help students to be lexically proficient.
What can be said is that human judgment is a necessary part of assessing lexical proficiency and that collocation accuracy is a key feature.
Submitted by Barry Lee, British Council language assessment consultant
Scott. A. Crossley, Tom Salsbury and Danielle S. McNamara. Assessing Lexical Proficiency Using Analytic Ratings: A Case for Collocation Accuracy. Applied Linguistics 2015: 36/5: 570-590.