Lecture 12: The Mental Lexicon

In a 1940 study Seashore & Erickson estimated that an educated adult knows more than 150,000 words and be able to use 90% ot these. To arrive at this figure they prepared a list of 1,320 words by taking every third word down in the first column of every left hand page of the 1937 edition. They divided the list into four and asked hundreds of college students to define them and use them in illustrative sentences. The proportions correct were applied to the overall number of words in the whole dictionary to produce the overall total of 150,000 words. Some of the problems with this procedure are the definition of a 'word' and the 'big dictionary effect'.

The reading vocabulary of the average American high school graduate has been assessed at 40,000 words. If proper names of people and places, ands idiomatic expressions are included the total rises to 60,000.

This large vocabulary size is in stark contrast to that of the "talking apes". Washoe and Nim Chimpsky, two chimps, learned about 200 signs each, Koko the gorilla had a vocabulary of about 400 signs. None of the apes came close to the 2,000 word vocabularies achieved by most children soon after the age of 2.

The Familiarity Effect

The familarity effect illustrates a clear difference between the mental lexicon and a dictionary. In a dictionary it takes no longer to look up a less commonly used word; in the mental lexicon familiar words are more rapidly accessed. This is measured using a lexical decision task. People are asked to indicate as rapidly as possible whether or not a string of letters spells an English word. The reaction time is the time between the instant that the word appears and the instant that people answer 'yes' or 'no'. The lexical decision task consistently shows faster response times for high-frequency, high-familiarity words. One speculation about the reason for this effect is that frequently used words are easy to find quickly because they are stored in many different places in the brain.

Another more counterintuitive finding that fits with the speculative theory offered above is that people respond faster to homographs than nonhomographs. That is, words that have more than one sense are recognized slightly faster than equally familiar words like neighbor that have only one sense. This implies that homographs are mutilply represented for the variety of meanings.

Bits of Words: Inflectional and Derivational Morphology

An experimental task used to explore the role of morphology in lexical organization is repetition priming. If a word is presented twice, with some gap inbetween, the second lexical decision time will be shorter than the first. Inflected words prime their uninflected forms just as well as the uninflected forms prime themselves. This is also true of derivative forms. In contrast, priming does not occur between morphologically unrelated words whose initial letters coincide; for example, candle does not prime can. The results of such priming experiments support the idea that morphologically related words are stored together in the mental lexicon. In this sense there is some overlap between a print dictionary and the mental lexicon.

If we stop at this characterization of the mental lexicon, we miss one of the ways in which it is most clearly differentiated from print dictionaries. Words are not stored in separate compartments in the mind, they coexist in an elaborate network of associations. When a word is used the activation in the mental lexicon spreads over this network of associations. Words are not only associated with meanings. They are associated with each other.

Rosch's Prototype Theory

66. Consider for example the proceedings that we call "games". I mean board-games, card-games, ball-games, Olympic games, and so on. What is common to them all? - Don't say: "There must be something common, or they would not be called 'games'" - but look and see whether there is anything common to all.-For if you look at them you will not see something that is common to all, but similarities, relationships, and a whole series of them at that. To repeat: don't think, but look! - Look for example at board games, with their multifarious relationships. Now pass to card games; here you may find many correspondences with the first group, but many common features drop out, and others appear. When we pass next to ballgames, much that is common is retained, but much is lost.-Are they all 'amusing'? Compare chess with noughts and crosses. Or is there always winning and losing, or competition between players? Think of patience. In ball games there is winning and losing; but when a child throws his ball at the wall and catches it again, this feature has disappeared. Look at the parts played by skill and luck; and at the difference between skill in chess and skill in tennis. Think now of games like ring-a-ring-a-roses; here is the element of amusement, but how many other characteristic features have disappeared! And we can go through the many, many other groups of games in the same way; can see how similarities crop up and disappear.

And the result of the examination is: we see a complicated network of similarities overlapping and criss-crossing: sometimes overall similarities, sometimes similarities of detail.

67. I can think of no better expression to characterize these similarities than "family resemblances"; for the various resemblances between members of a family: build, features, colour of eyes, gait, temperament, etc. etc. overlap and criss-cross in the same way. -And I shall say: 'games' form a family.

Rosch adopted this idea of family resemblances and explored its significance for representation in the mental lexicon. She observed that humans appear to find some instances of words more basic than others. For example, think of the colour 'red'; I'll wager that you thought of a bright, saturated fire engine red, even though you'd be happy to call other less clear cut cases 'red'. Rosch investigated this phenomenon by asking lots of college students about the representativeness of various instances of a superordinate category term. In other words, Rosch investigated the birdiness of birds. The results were surprisingly consistent. Almost everyone thought that 'robin' was the best example of a bird, pea of a vegetable and chair of furniture. Intriguingly, she also found that shirts, dresses and skirts were considered better examples of clothing than shoes and socks. 'Robin' is a more prototypical bird than 'penguin'. Rosch's idea is that judging category membership against a prototype allows rough matches to suffice.

The question, "Were the students just responding faster to more frequent words?" may be coming to your mind. It turns out that the results are not explained solely on the basis of word frequency. On the furniture list, rare items of furniture such as love seat, davenport, and otoman came out higher than refrigerator. On the vegetable list, pea carrot and cauliflower came out higher than onion, potato and mushroom. On the clothes list, pyjamas came out higher than shoe, tie, hat, and gloves.

Something that sings out from the results of this study is that humans do not often deal with isolated words. Words are represented in network fashion in the lexicon with varying and systematic strengths of connections between the items. This is another way in which the mental lexicon differs radically from a print dictionary.

Models of Word Selection

The simplest model of representation of these two 'sides of the coin' is a stepping stone model: The meaning is represented separately from the sound sequence and there is a marked pathway or connection between them. Several words are activated simultaneously that belong to the same category or 'family'. Hence semantic slips such as 'otter' for 'beaver' are explained by an incorrect selection at this point. Likewise, sound errors are explained in terms of an incorrect choice among all the activated similar sound sequences; for example, 'beaker' for 'beaver'. The difficulty with such a model is its inability to predict or explain combined sound/meaning slips such as 'badger' for 'beaver'. We need to refine the model to incorporate simultaneous activation of the semantic and sound levels.

The waterfall model allows a person still to be thinking about the meaning as they select the sound. All the information activated at the first stage is still available at the next stage, cascading down to the next trough of water on the hillside. This model suggests that word selection is not just a case of following one word through from beginning to end, but is often a case of controlling and narrowing down a cascade of possible words. The error 'badger' for 'beaver' is explained by assuming that the semantic information about small animals was cascading down as the outline phonology was picked. The difficulty of this model is that waterfalls can't flow backwards. Just as selecting the phomology requires semantic evidence, so phonology may be needed to narrow down the semantics. For example, consider the case of searching for members of a particular category. If I say 'think up the names of some woodland animals', you may get stuck after rabbit and squirrel. If I then prompt you with "Beginning with b" you might suddenly produce 'beaver' and 'badger'. Sounds can activate meanings, and meanings can activate sounds. This is a "spreading activation" model in which information can flow both ways. Normally, an initial meaning impetus fans out and activates more and more words as it spreads along the various connections. As the activated links are inspected, those that are relevant get more and more excited, while those that are unwanted fade away: the rich get richer and the poor get poorer. Since the current is flowing to and fro, anything which is particularly strongly activated in the semantics will cause extra activation in the phonology, and vice versa. If the outline phonology fits more than one animal, then both animals will become highly activated. If the speaker is not paying attention the wrong one will get picked.