I should preface this second post on the limited usefulness of semantic domains for learning Biblical Hebrew vocabulary by stating that I did not always think this was the case.  In my intermediate Hebrew course, I remember having Landes as the vocabulary text and thinking it was great to see someone finally do something other than a strict word list.  And, going into my dissertation research on Biblical Hebrew vocabulary learning, I believed that I would use this approach heavily, but supplement it with more vocabulary learning strategies.

However, I started my vocabulary research by reading vocabulary texts written by researchers who focus specifically on Second Language Vocabulary Acquisition (particularly the three at the end of this post plus a couple of others), rather than Second Language Acquisition in general.  And, I am glad I did so because I don’t believe these two groups of researchers are saying the same things (Yes, I have read many more general SLA texts as well).  I began to see the use of semantic domains as an overarching strategy as problematic after reading Nation.  He states the following:

Care should be taken in introducing lexical sets because this goes against the criterion of frequent use… Further arguments against the presentation of lexical sets as a way of introducing vocabulary can be found in the research of Higa (1963), Tinkham (1993 and 1997) and Waring (1997b) which show the difficulty caused by learning similar items together (387).

I, like some of the readers of my first post, was not completely convinced by the principle of frequency as there could be ways of tweaking this.  However, after reading the experimental studies referenced by Nation (as well as about five or six others) on “the difficulty caused by learning related items together,” I became much more strongly convinced that introducing vocabulary by lexical relationships was not a beneficial practice.  That is not to say that I think there is no place for lexical sets in vocabulary learning.  I do not have time to elaborate on this here, but word knowledge is multi-faceted, with semantic associations being one of those facets.  So, this should receive attention; however, in my opinion the focus should come later in the learning process when learners are consolidating their knowledge of a word for which form and meaning has already been paired.

At any rate, what is it that these studies report?  The studies report on a problem that Nation refers to as interference (303).  As in the comments of a previous post, interference in this context basically means that learning items like “mouth” and “nose” together does not make it more likely that these items will be remembered because of their close association, but rather makes it more likely that learners will have problems retrieving the meaning of either of these words because the initial learning was not distinct enough.  More simply, learners may be more likely to confuse the words, whether this results in delay or error.

As a basic summary of the experiments in these articles, researchers present groups with two different ways of learning words.  One group learns a set of unrelated words and another group learns a set of semantically related words (sometimes there are variations like pictures being used, etc.).  Then, the groups are each given a task to perform using the vocabulary.  Across the board, in the 10 studies that I have put into a bibliography, the subjects who learn words in related sets perform worse on the task in which they put the vocabulary to use.  These results have been replicated so often, I think, because they are so unexpected.  Throughout the psychological literature we are told that we learn by association.  These studies seem to violate that principle until one realizes that the idea of learning by association is extremely complex.

Against this background though, these results have been replicated so often that linguists and psychologists have developed a reasonably good explanation for them them.  In the comments section of a previous post, I noted that Finkbeiner and Nicol explain these results in terms of spreading activation, a concept that will be familiar to almost anyone who has reviewed psychological literature on language.  Spreading activation basically means (at least in this context) that when a person sees a lexical item all the other items associated with it are activated as well.  This obviously has a lot of beneficial effects in terms of language processing.  It helps us to read quickly because when we see a particular word we can begin to expect which words might come next.

Yet spreading activation also has an inhibitory effect.   This effect is possibly at work in the results of the studies cited here.  When a person learns words in lexical sets, the learning is not distinct enough.  Thus, in some cases, when a learner sees a particular lexical item all of the associated items are activated too strongly for the learner to make a correct or timely decision about the meaning of the item.  This happens time and time again in the literature.

At this point, I suppose a number of objections might be raised.  But, I think there is enough evidence here to suggest that there is at least something to this and that this is a very important issue for vocabulary learning.  There is enough here to suggest that these studies cannot be dismissed off-hand.  These are studies that I feel must be at least grappled with by those who develop the vocabulary component of a course.  On the other end of the spectrum, I, as Finkbeiner and Nicol, have not seen any empirical research that suggests introducing vocabulary in lexical sets can be beneficial.  Most, if not all, evidence proposed in this line comes from monolingual studies in which subjects learn lists of words.  This is not the same thing as pairing form and meaning in vocabulary learning.