Ling 423/640G: Cognitive Linguistics

Ben Bergen

Meetings 3: Categorization

September 2, 2008

We're looking at conceptual categories, the basis for word meaning.

Today: these categories are more or less specific, and not all members of a category are equal.

What we'll find is that once again, conceptual categories depend on embodied human experience.

1. The basic level

What's this?	How about this?

We can categorize entities and events at different levels of specificity.

• plant, tree, palm, fishtail palm
• vertebrate, mammal, cat, Tabby

Convergent evidence suggests that one of these levels is special: the basic level

· Highest level of distinctive and shared actions

• Highest level of perceived overall shape and single mental image
• Learned first
• Shorter name
• Most frequent reference
• Entities more readily named at this level
• Greater cultural significance
• Entities remembered more easily
• Used in neutral contexts

Basic level differences

• Across cultures: New Yorkers versus the Tzeltal
• Within cultures: expertise and jumping between levels (mechanics, biologists, and surfers)

Why basic level categories?

• Shared biological basis for perception and action
• The crucial role of actual experience (as shown by differences in categorization)

2. Prototypes

Many categories have certain members that seem to be better, more canonical, more central, or more important than others, as in chair, dog, home, etc.

How do we test prototypicality?

• Direct rating

Subjects rate more prototypical instances as “better” members of the category.

• Reaction time

Subjects more quickly identify prototypical instances as members of categories.

• Production of examples

Prototypical instances are produced first when subjects are asked to provide examples of a category and appear more frequently across subjects

• Asymmetry in similarity ratings

Non-prototypical instances are rated as more like prototypical instrances than the reverse.

• Asymmetry in generalization

Facts about a prototypical instance are more likely to be extended to less prototypical instances than the reverse.

Ad hoc categories, e.g. 'Things to take on a fishing trip with a small child' also display prototypicality

(Schafer) sunscreen, toys, lots of extra clothes, anti-diarrhea medicine, band-aids, antibacterial waterless handwash, Swiss army knife, life preservers, life savers, tarp

3. Prototypes beyond Rosch

Typical example

husband, tool, bird

reference-point reasoning, categorization

Stereotype

husband, politician, Japanese tourist

judgment, categorization

Ideal case

husband, job, car

evaluation and analogy

Paragon exemplar

husband, baseball player, humanitarian

evaluation and analogy

Salient example

earthquake, flood, court case

generalization, categorization

4. Prototype effects and mental representations

A question remains – what is the relation between prototype effects and category structure?

• Prototype effects could be directly reflective of mental representations of category structure
• Or they could be epiphenomenal (emergent), and result from the accumulation of experience

Whichever model ends up being right, the relationships we have to categories, and the language we use to identify them, shows prototype effects.

Prototypes are not always out there in the world, so this aspect of conceptual and linguistic categorization must be the product of our embodied physical and social interactions with the world.