What are ad hoc categories?
Many of our cognitive categories are stable, others are ad hoc. Crucially, ad hoc categories are context-dependent and people construct them to achieve their communicative goals. For example, constructing the category of “activities one can perform on a sunny Sunday afternoon” can be instrumental to achieving the goal of inviting a friend to spend the Sunday afternoon together. Cognitive psychologists (e.g. Barsalou 1983, 1991, 2003, 2010) have shown that, for ad hoc categories, the category concepts, concept-to-instance associations, and instance-to-concept associations are much less established in memory than for common categories (e.g., “fruit”, “furniture”). In addition to these differences, however, ad hoc categories have been shown to possess graded structures (i.e., typicality gradience) as salient as those structuring common categories. This appears to be the result of a similarity comparison process that imposes graded structure on any category regardless of type (Barsalou 1983).
Stable categories can typically be expressed by fairly short conventional linguistic means (e.g., ‘queen’, ‘eagle’, etc.). Ad hoc categories, instead, do not come with ready-made linguistic labels (words [e.g. furniture, clothing], or small phrases [e.g. grocery stores, vegetarian food]), and are often described by means of complex expressions (e.g., ‘tourist activities to perform in Rome’, ‘clothing to wear while house painting’, etc.). Their identification in discourse nonetheless crucially depends on verbalization, i.e. the linguistic strategies that speakers systematically employ to refer to the process of ad hoc category building.
The linguistic expression of ad-hoc categories
Being the internal structure of ad hoc categories based on typicality gradience, the linguistic strategies expressing them often involve the explicit naming of one or more exemplars, that the addressee processes as pointers for conjuring up higher-level, ad hoc category (Ariel and Mauri 2014).
However, the role of exemplars and the degree of context-dependence may vary in the process of ad hoc category building. For instance, building the category “things I usually do on Sundays” through a non-exhaustive list ‘jogging, going to museums and things like that’ is highly context-dependent and could not be constructed without reference to shared knowledge regarding the speaker (it’s hard to determine what other members belong to the set without knowing me). On the other hand, constructing the category ‘people and situations revolving around Berlusconi’ through a derivational process like ‘Berluscon-ame’ (Italian) does not require access to the specific speech situation, but more to a general cultural knowledge of Italian politics. Furthermore, in the former case the named exemplars act as pointers to a higher-level category, while in the latter, Berlusconi has to be interpreted as both an exemplar and a property that all potential members of the category must share.
We observe some variation also in the discourse functions with which ad hoc categories may be constructed. For instance, there are cases where a ready-made linguistic label is available in the language, but speakers choose not to use it and rather construct the category as ad hoc, in order to stress its context-dependence.
A very shallow cross-linguistic survey shows great variation also at the structural level, where constructions encoding ad hoc categories range from more transparent discourse-level strategies such as English ‘or stuff like that’ (“general extenders” in the literature; Overstreet 1999), to synthetic, less transparent means such as the Japanese non-exhaustive connective -ya in (1), dedicated plurals (so-called associative and similative plurals, see Daniel 2000 and Moravcsik 2003), derivational affixes or special types of reduplication (e.g. Turkish m-reduplication) to convey this function.
There is then a great structural as well as functional variation in the expression of ad hoc categories. However, as argued by Barsalou (2010), although ad hoc categories are ubiquitous in our everyday cognition, they have been subject to relatively little research, especially as far as their linguistic realizations are concerned. No systematic analysis of the linguistic structures, as well as their semantic and functional variability has ever been made.