Towards variable function-dependent sense ordering in future dictionaries

The present paper deals with the relationship between sense ordering and the primary function of the dictionary, arguing that the former should reflect the latter. I give suggestions as to how this could be achieved in practice, and which sense ordering would best serve the different functions. Given that lexicography appears to be at an important crossroads, I claim that lexicographers should embrace more enthusiastically the potential afforded by new storage media. With regard to sense ordering I thus propose that we move beyond static ordering in electronic polyfunctional dictionaries of the future, to dynamically adjust the ordering to the currently dominant function. Another future possibility that I explore is that Artificial Intelligence systems in electronic lexicographic products could conceivably monitor individual users’ lookup behaviour and thus customize sense ordering on an individual user basis, for example by deprioritizing senses known to have already been acquired by the user, when in the text reception/decoding mode. Finally, I outline a programme of research to provide a more empirical basis for the above. 1. Sense ordering in dictionaries The issue of sense ordering in dictionaries arises in those entries which have more than a single sense. Such entries have often been referred to in the literature as polysemous entries. When talking about polysemy, though, we need to be careful to distinguish between polysemy as a linguistic, lexical-semantic concept referring to hypothetical semantic properties of units of language, and a derived sense referring to an aspect of the dictionary (micro)structure. It needs to be borne in mind that a dictionary description does not necessarily match a lexical-semantic structure of a unit of language postulated by linguists. This is so because of the diverging aims of lexicographers and linguists: A good dictionary should be a maximally helpful tool for the target user in a range of problems and consultation situations for which it is being designed. In contrast, a linguistic description of a language is a model of language (whatever the terms ‘model’ and ‘language’ may mean within a particular linguistic paradigm). True enough, the two may sometimes converge for some dictionaries: those aiming to faithfully model some part of the lexicon; this, of course, would typically reflect a specific theoretical position regarding the nature of the lexicon. Such convergence, however, should be seen as an exception rather than the rule. All in all, given the above problems, it is probably safer to use the more neutral term multiple-sense entries when referring to lexicographic entities, rather than polysemous entries. In multiple-sense dictionary entries and within a linear lexicographic presentation space (see Figure 8 below for an example of the non-linear alternative), a decision must be taken as to the ordering of these multiple senses. The decision as to which sense should be placed first is especially important in view of the available findings from research on dictionary use which indicate that dictionary users tend to stop at the initial sense given, unless they get a very clear signal compelling them to move on (Tono 1984; Lew 2004). In view of this marked preference of users for the first sense(s), it is fair to say that sense-ordering decisions have crucial practical implications for the success of dictionary consultation, and the issue of sense ordering itself deserves more attention from metalexicographers than it has received so far. Lexicographic decisions with regard to sense ordering typically follow several others, notably the grouping of presentation into discrete senses (witness the continuing debate between ‘lumpers’ and ‘splitters’). Still, some decisions as to grouping and ordering may be interdependent. Even so, in