1. PROTOTYPES AND CATEGORIES
1.1. Colours, squares, birds and cups: early empirical research into lexical categories.
Focal or prototypical colours used as points of orientation
Boundaries: where something begins, where it ends => often boundaries are vague
Vagueness: but for boundaries provided by reality (e.g a kneecap cannot be included in the thigh)
Temperature and colour: continuum; hot – warm - cold (classification - a mental process)
mental process of classification = categorization à cognitive categories (e.g. colour: e.g. red)
Berlin & Kay: FOCAL COLOURS – reference points of orientation (on the continuum); universal
Munsell colour chips (colour samples) – colours – different hues (shades) – speakers of the 20 languages tested results basic colour terms consistent in different languages
THE PSYCHOLOGICAL BACKGROUND OF FOCAL COLOURS
Categorization of natural phenomena:
1. Selection of stimuli à sensors à stimuli attracting attention
2. Identification and classification à comparing selected stimuli to relevant knowledge stored in memory
3. Naming à most cognitive categories are given names
Eleanor Rosch – explored psychological background of focal colours; psychologist; aim: to find out whether focal colours are rooted in language or pre-linguistic condition; in her experiments she worked with children and the Dani (from Papua New Guinea)
perceptually salient (=most noticeable or important) = more attention-attracting; remembered more accurately in short-term memory + more easily retained in long-term memory à perceptual‑cognitive salience (probably independent from language)
Foci (focal points) à PROTOTYPES à Rosch’s further study à shapes, organism, objects
PROTOTYPICAL SHAPES
‘good forms’: squares, circles, equilateral triangles = perceptually salient figures = reference points
PROTOTYPICAL ORGANISMS AND OBJECTS
· ‘goodness’ – typicality of category members (see: table 1.3, page 17 for good and bad examples)
GOOD EXAMPLES, BAD EXAMPLES AND CATEGORY BOUNDARIES
· Category membership – not a yes/no distinction; involves different degrees of typicality
· Prototypes = cognitive reference points (categories are formed around them)
TWO TYPES OF BOUNDARIES AND TRANSITION ZONES
some concrete entities do not have clear‑cut boundaries in reality (body parts, weather phenomena, landscape)
fog knee valley
the boundaries of entities are vague
VAGUENESS
BUT they may coincide (e.g. mountains: not delimited, no clear boundaries in reality)
entities have clear boundaries
BUT the boundaries of cognitive categories are vague
chairs cups
fuzzy category boundaries
FUZZINESS
fuzzy nature of category boundaries
William Labov:
· Experiments: cups + cup-like containers; fuzzy category boundaries = fringe areas between adjacent categories (check: Figure 1.4, page 21) CUP –> BOWL
· First experiment: cups drawings but no background info = “neutral context”
· Later on: different contexts (food context, scene: with flowers inside; various materials)
CONTEXT-DEPENDENCE
· Categories do not represent arbitrary divisions of the phenomena of the world, but should be seen as based on the cognitive capacities of the human mind
· Cognitive categories of colours, shapes, but also organisms and concrete objects, are anchored in conceptually salient prototypes, which play a crucial part in the formation of categories
· The boundaries of cognitive categories are fuzzy = no rigid boundaries; merging
· Between prototypes and boundaries, cognitive categories contain members which can be rated on a typicality scale ranging from good to bad examples
BUT there is no one-to-one relation between categories (or concepts) and words; if one word denotes several categories = polysemous
Cognitive linguistics: exploration of individual categories + exploration of a relationship between the categories signalled by one and the same word.
1.2 The internal structure of categories: prototypes, attributes, family resemblances and gestalt (with birds, lots of them)
- Cognitive categories are labelled by words and it is natural to look for information about them in dictionaries. The book presents 3 dictionary definitions of birds: robin, parrot and ostrich. From them we can acquire two types of information:
1. The name of the category which they belong to (BIRD). The category name suggests properties shared by most birds: they have feathers, two legs, two legs and a beak, and they lay eggs.
2. Properties specific to the particular item (e.g. robin is characterised by small size, brownish colour and red breast-feathers), that distinguish the item from other members of the category.
- However, there are some problems with these definitions - lexicographers may skip properties that are to be taken for granted or modify definitions using words such as ‘usually’ (robins have ‘usually’ X and Y).
- Properties can be called in a more technical term = attributes
- Aristotle contrasted ‘essence’ and ‘accidence’. The notion of essence gave rise to a categorical (or classical) view, according to which a category is defined by a limited set of necessary and sufficient conditions. These conditions are called essential features.
- Back to the birds, in the case of the category BIRD this means that a creature is only a bird if it has feathers, two legs, two legs and a beak, and it lays eggs.
Attributes, good birds and bad birds
- Here’s the attribute list for the category BIRD, based on dictionary definitions of robin and bird:
1. ‘lays eggs’ 2. ‘has two wings and two legs’
3. ‘has a beak’ 4. ‘has feathers’
5. ‘can fly’ 6. ‘has thin, short legs’
7. ‘is small and lightweight’ 8. ‘has a short tail’
9. ‘chirps/sings’ 10. ‘has a red breast’
- Such attribute lists can never be complete but this one is ordered as regards to the most necessary attributes for the category.
- Other types of birds highly representative to the category, such as sparrow, canary and dove share the most aforementioned attributes with robin.
- When you compare the less representative creature, parrot, you can see that it shares just some of the attributes with robin, while ostrich, a poor example of the category, shares only the most fundamental attributes to the category (apart from flying).
- Conclusion: there is a bundle of attributes that represent important aspects of ‘birdiness’. The attributes correlate in nature (they appear together). When a type of bird has most of them, it has a very prominent position in the whole category. As for ‘worse’, not prototypical examples, they differ in two ways: either they deviate to a certain degree in some attributes (like parrot with strong legs) or they simply don’t have them (ostrich can’t fly).
The principle of family resemblances
- Let’s focus on the absence of an attribute. The paradox of ostrich is that we naturally call it a bird, even if it shares few attributes with other birds and has its own, unique, unthinkable for other birds (‘very tall’, ‘runs very fast’). However, it shares some less common attributes with other types (‘long neck’ with flamingos or ‘large soft feathers’ with swans)
- Ludwig Wittgenstein argues that all representatives of the given category are connected by a network of operating similarities. He calls these similarities family resemblances and provides the example of GAME - there are board-games, card-games, ball-games, Olympic Games and so on. They all differ from the others in some aspects but they share some features like ‘competition’, ‘winning and losing’, ‘amusement’, however not to the same degree.
- Rosch and Mervis: ‘each item has at least one, and probably several, elements in common with one or more other items, but no, or few, elements are common to all items’. (non comprende? See the picture on page 29)
- Superordinate categories largely depend on family resemblances.
- The situation changes when we consider concrete categories (like BIRD, CAR, CHAIR, LAMP). Even very bad examples of the category (ostrich, penguin) share some important attributes with all the other category members. The good examples share whole bundles of attributes, so only a few of their attributes rely on family resemblances. (See the twisted picture of this situation on page 30)
Attribute listing and attribute-based typicality ratings
- Prototypical members of cognitive categories have the largest number of attributes in common with other members of the category and the smallest number of attributes which also occur with members of neighbouring categories. This means that in terms of attributes, prototypical members are maximally distinct from the prototypical members of other categories.
- Bad examples (or marginal category members) share only a small number of attributes with other members of their category, but have several attributes which belong to other categories as well, which is just another way of saying that category boundaries are fuzzy.
Attributes and dimensions
- The problem of absence of some features in the category was solved by family resemblances but there are also ‘deviant’ cases, in which category members have attributes not falling into the expected norm. Most of these cases are related to dimensions, such as ‘size’, ‘weight’ etc. This applies to BIRD but also to Labov’s cups and bowls discussed in the previous section. As for dimensions, attributes don’t have to be listed, cause the best examples can be chosen based on generally accepted ‘logical’ properties of the dimension type.
- Labov also distinguishes less common dimensions for cups and bowls like ‘context’ dimension (e.g. neutral, coffee, food, flower contexts), ‘material’ dimension (glass, china), ‘handle’ dimension (presence and absence of handles).
Internal category structure and gestalt
- Goodness-of-example rating might be done based on intuition but this intuition doesn’t derive from our own check-up for attributes. When you see an animal, you don’t begin the process of identification by evaluating specific attributes, unless they are very salient like the stripes of a zebra or the trunk of an elephant. We have a different way - we simply take in an overall picture of the whole and use it for the assessment of its goodness.
- The problem with this way is that it seems impossible to investigate it in controlled test situations. However, it often makes sense to assume that people in tests rate items basing on overall impression of the particular item rather than analysing different attributes and dimensions.
- An experiment was conducted where different three drawings of houses (you can see it on page 35) were presented to informants and they had to categorise it. There was no full agreement between them. When asked to give reasons for their categorising decision, they said that it was based on their general impression of the drawings. Only on second thoughts they could identify certain individual attributes that may influence their decision.
- Categorisation and goodness ratings may involve two stages: the holistic perception (the perception of an object as a whole) and a kind of decomposition of the whole into individual attributes.
- The idea of ‘perceived whole’ comes close to the notion of gestalt. Gestalt psychologists distinguish certain ‘gestalt principles’. The most important are:
· ‘principle of proximity’ - individual elements with a small distance between them will be perceived as being somehow related to each other
· ‘principle of similarity’ - individual elements that are similar tend to be perceived as one common segment
· ‘principle of closure’ - perceptual organisation tends to be anchored in closed figures
· ‘principle of continuation’ - elements will be perceived as wholes if they only have few interruptions
- The more a configuration of individual elements adheres to these principles, the more it will tend towards a unified organisation (Prägnanz) and gestalt perception. Examples showing a high degree of Prägnanz are called ‘good gestalts’ or ‘good forms’.
- Shape is just one aspect of gestalt but it is so important that sometimes is ...
carrie77