The linguistic essay below has been submitted to us by a student in order to help you with your studies.

Back to Linguistic Index

Relevance and problems of prototype theory


Prototype theory is an interesting approach towards the research of language acquisition in psycholinguistics. It accounts for the way children learn words. However, it also presents some problematic issues. Although it has been studied by many experts, the general idea is that there should be other methods giving a more relevant explanation of the process of learning words.

Prototype theory

The subject of my paper is prototype theory, that is, how children learn words, which is important in the research of language acquisition. This is an interesting theory in psycholinguistics for the purposes of understanding how the human brain works and what processes children undergo during learning. The aim of my research is the investigation of this theory, summarising the relevant sources and providing a basic understanding of the topic. The problem which I will try to resolve is the relevance of this theory: I will present the arguments for and against it based on the sources as well as my own views. The main point of my argument is that this theory is not satisfying in the explanation of word learning, and there should be some alternative methods. I will try to create this solution based on my ideas. Thus, first I will review the related literature, then present the basic problems and possible approaches to this theory, trying to reach some agreement. Finally, I will state the possible solutions I would offer.

Review of literature

The basics of the theory: categorisation

The basic method children apply when learning the words of their native language is categorisation. This is a practice people do basically all the time: they constantly categorise the world around them to understand and manage it easier. Possibly the ability to categorise in childhood develops into this kind of categorisation that people use in adulthood.

Lakoff (1987) defines this method: "things are categorised together on the basis of what they have in common." (p. 5) He claims that categorisation becomes automatised, so when we categorise things, we do it unconsciously and without planning. Probably children do need some planning and conscious thinking when they learn to speak and thus to categorise, but it is usually entirely automatic, for example, selecting similar coloured objects does not require much consciousness, as children can judge them by visual features. Indeed, the first criterion children often base their decisions on is the visual appearance of objects, as Slobin (1971) states. He presents the findings of Bruner (as cited in Slobin, 1971) whose studies showed that children at an early age categorise objects based on their colour or other similarities which can be actually seen. Older children can recognise more abstract features and form semantic categories. The requirements of membership of a certain category is explained by prototype theory.

The history of the theory

The first to examine this phenomenon was Rosch, whose experiment (1975) tried to answer to the question of which items do we consider typical and which less typical a member of a certain category. Rosch stated: "If categories are defined only by properties that all members share, then no members should be better examples of the category than any other members." (as cited in Lakoff, 1987, p. 7) But since all members cannot possibly share all features (e.g., penguins cannot fly, but they are still birds), there must be typical features which define the categories. In the case of birds, it can be feathers, flying, wings and beaks, and some of these features occur in all members of the category bird. Since children judge more likely by visual features than by abstract ones, they probably see a penguin or an ostrich as a less typical bird, as these do not share the features which others do. Also, because these birds cannot fly, they are less typical than other ones. Later, when children can understand biological factors, they will be capable of putting these animals in the bird category. According to Rosch, the best examples are called prototypes; these are the most common elements of a category on which children base their judgments. Basically the prototypes are created by examining the appearance of objects, or the way they occur (e.g., vegetables seen at the greengrocer's can be identified as members of the same category). Thus children tend to make over or underextensions, until they can judge upon adult standards. In the case of overextensions, they might apply the word dog to dogs and cats as well, since they are both relatively small, four-legged furry animals. Underextensions might happen if they do not accept something as a member of a larger category, for example, they say that a penguin is not a bird, it is just a penguin. This way we can instantly see a problem with prototypes; they are more likely to account for the mistakes children make (which is a good aspect; through the mistakes researchers can trace back the method itself), than being a good model of word learning in all aspects. It is not the theory that is problematic, it is the practice itself, but since children possibly cannot understand exactly why a bird is a bird, it is easier for them to create their own categories even if they are sometimes wrong.

Category boundaries

Lakoff (1987) refers to this phenomenon as basic-level categorisation, while what later occurs is superordinate categorisation. He also claims that the boundaries of the categories are not solid; there can be many factors influencing the membership of a certain category. He uses the term inherent degrees of membership regarding for example, red, which is a category with fuzzy boundaries, with members which more certainly belong than others. In the case of the above-mentioned problem of birds, bird is a category with clear boundaries (feathers, flying, beaks, wings, etc. define it), with graded prototype effects, the prototypical members of the category. In his view, the characteristics defining the boundaries of a category do not need to be shared by all members (in that case, we could say that a penguin is not a bird, because it lacks the ability to fly), but some of the features occur in every member. (Penguins have beaks, wings and feathers; bats have wings and can fly - but they are still not birds, we need additional world knowledge to judge this). Lakoff calls these features family remembrances. He also presents an idea of Wittgenstein's, who basically said the same about boundaries, calling them clear and extendable boundaries (as cited in Lakoff, 1987). An example of the latter are games, many words can be put into this category, including words for sports, board games, etc. The boundaries can be extended, it is not defined precisely what features do members have to share to be games.

Other aspects defining boundaries

Slobin (1971) introduces another method called componential analysis. He talks about shaded boundaries, which occur in categories such as chair. In my previously presented examples which form some actual category, for example, bird being a biological term as well, the categories can be narrowed down and the members can be viewed as categories themselves. Thus a chair, being a piece of furniture, is a category in itself; according to Slobin, we can differentiate it from bench or stool. This distinction is based on dimensional features, whereas in case of componential analysis, a certain component makes the difference between categories. This study, carried out by Wallace and Atkins (as cited in Slobin, 1971) examined the different terms used for family members. Slobin explains that the differences in this case are based on gender and generation features, that is, on certain components. For example, father and son are both male relatives, but they belong to different generations, whereas father and mother are both parents, belonging to different genders. The relevance of this theory is that it provides an alternative method of categorisation; it might be applied to prototypes as well, regarding the features defining a category as such components (e.g., duck and tiger are both animals, but the lack of feathers excludes tiger from the category bird). As opposed to this, Slobin's example of chair is a category of not clear boundaries, we cannot differentiate such components, but the different degrees help us to decide which object belongs to which category (in Slobin's example, if a seat of the chair is wider, it is a bench). Basically we can apply this degree-based distinction to smaller categories and the componential one to larger categories, such as family members or animals (there might be many differences between a cat and a dog in terms of components - ears, shape of their head, etc. -, but what is a difference between two dogs? We need elements of degree to make differences between the types of dogs, such as their colour, size of the body, etc.). As we have seen, different elements can be used for defining different category boundaries, but they still cannot define a universal prototype, at least, it is not a good solution. The related works offer many alternative views to protoype theory, which might demonstrate the subject better, as I just presented.


Limitations of prototypes

Several problematic aspects might be considered about this theory. First, as I have already mentioned, children tend to make over or underextensions when applying this method of learning. A good example of this is what Vygotsky called a heap concept. According to Vygotsky, when overextending, that is, generalising a concept of a word, children use that particular word for more, semantically unconnected objects. In his example, a child used the word 'qua' (quack) to refer to a duck in a pond, but because of the water, also used it for a glass of milk, based on the remembrance of milk and water. So in Vygotsky's view, children can recognise some similarities on which they base their judgements, but it often results in such mistakes (as cited in McDonough, 1992). This is an aspect of prototype theory as well, children consider something they see as a prototype of a category and use the word for anything which resembles that object in any way. It might be that while categorisation is a perfectly good explanation of how children learn words, prototype theory is not; so we need to establish another method which gives us a better insight.

Categorisation and word classes

Another problem might arise in connection with word classes. So far, I only mentioned nouns and their categories. Similarly, adjectives can be learnt, dividing them into different categories, for example, colours, words describing appearance (tall, short, fat), or words describing emotions (happy, sad). The boundaries are also important in these cases, for example, children have to learn the differences between happy, delighted, joyful, etc. In case of verbs, some categorisation can also occur, depending on the meaning. This categorisation is easy regarding content words, but the case of function words is more problematic. Since they do not carry any meaning in themselves, they cannot be given a proper definition, thus cannot be sorted into categories the same way as content words. Steinberg (1999) states this problem, saying that children usually associate function words with those words (and their word classes) with which they can occur together. This way they make a certain categorisation, but it is rather a grammar-based than a conceptual one. Still, it is a problem because these categories can be misleading if some irregularity occurs. Functions words are rather learnt according to certain rules than categories based on meaning.

Something similar is phrased by Slobin (1971) in connection with word classes. Learning word classes is also a kind of categorisation. According to Slobin, children learn in what position can word classes occur in the sentence or how can they be combined together. "A child should learn that words like boy and dog are members of the same class, because they can follow the." (p. 61) But, as he says, it is problematic, because sentences are built upon rules, children cannot learn sentence formation based on these connections. This example shows that we can make prototypes in other aspects of speech as well, and that it is basically problematic to follow this method. Johnson (1965) also investigates the learning process of function words, claiming that though these do not carry any meaning, we attach some meaning to them in connection with the syntax and grammar of a sentence where they occur. These must be done unconsciously by children, similarly to categorisation, since they cannot understand syntactic or grammatical rules yet. Johnson bases this on Chomsky's theory of language behaviour. In this case, the problem is that prototype theory cannot be used effectively in the learning of function words, so basically the theory is only relevant in the case of content words.


In the case of the first problem, that is, over and underextensions my idea is that prototypes should not be used in all cases. In my view, instead of prototypes, the emphasis should be on the features defining the category. Basically, children might categorise a bat as a bird as well, based on wings and flying, but probably there are some features among that of the categories' which could be applied to all members. Thus, for the category bird, having wings and feathers should be obligatory, while the ability to fly is optional. This way, children can exclude bat from the category bird, but include e.g. penguin or duck. In my view, instead of making prototypes, the features themselves should be considered. Every category should have a set of characteristics defining it, some of which are obligatory (which apply to all or at least most members) and some optional (which apply to certain members, but not all). This way we could make clear distinctions between different categories and could avoid those mistakes often made by children.

In the case of the problem of function words, I cannot offer the kind of solution as previously. Since no direct categorisation occurs, we cannot select features which could describe function words according to categories. Though the use of prototype theory would be interesting in this case as well, and maybe also relevant to some extent, it seems from the examples that it arises many problems. In my opinion, function worlds should be learned through context, and later through rules of grammar. Possibly the theory could be used in an early stage when children are trying to understand the basic concept and usage of function words, but the applied categorisation should not be made into a standard. Children should understand later how grammatical rules work and use this words accordingly. In my view, prototype theory can provide a basic understanding even in this case.


As we have seen, the basic problem with prototype theory is that it does not reach far enough to account for all aspects of word learning. We can mainly use it to explain how children learn content words (mainly nouns), but it is also slightly problematic. Instead of using prototypes, my solution would be to make more specific characteristics of categories, based on which we could decide the membership of words. As for function words, we have seen that this theory is less relevant; in those cases children should make connections between function and content words and learn the former through context. Similarly, word categories and grammatical rules cannot be learnt through categorisation, since irregularities occur everywhere. Prototype theory is good to understand the basics of language learning, but its use can be quite complicated in a wider sense. Thus, we might need different methods to examine all aspects and different theories to account for the various problems which might arise during the research of word learning.


  • Johnson, N. F. (1965). Linguistic models and functional units of language behavior. In S. Rosenberg (Ed.), Directions in psycholinguistics (pp. 29-66). New York, NY: The Macmillan Company.
  • Lakoff, G. (1987). Women, fire and dangerous things. What categories reveal about the mind. Chicago, IL: The University of Chicago Press.
  • McDonough, S. H. (1992). Psychology in foreign language teaching. London, England: Routledge.
  • Slobin, D. I. (1971). Psycholinguistics. Glenview, IL: Scott, Foresman and Company.
  • Steinberg, D. D. (1999). An introduction to psycholinguistics. New York, NY: Longman.