The Mental Lexicon

Table of contents

1. Introduction

2. Dictionary vs. Mental Lexicon
2.1. The frequency effect

3. Words
3.1. What is a word?
3.2. Do words have a fixed meaning?
3.3. The prototype theory
3.4. Sense relations
3.4.1 Synonymy
3.4.2. Antonymy
3.4.3. Hyponymy
3.4.4. Meronymy
3.4.5. Collocations
3.4.6. Word-webs and atomic globules

4. Lexical Processing
4.1. Serial-autonomous models (Forster's search model 1976)
4.2. Parallel-interactive models
4.2.1. Morton's logogen model (1969)
4.2.2. Marslen-Wilson's cohort model (1978)
4.3. Serial-autonomous vs. parallel-interactive processing models

5. Lexicon acquisition
5.1. Children acquire words
5.1.1. The vocabulary spurt
5.1.2. Over- and underextensions
5.2. Constraints on word meaning
5.2.1. The taxonomic assumption
5.2.2. The whole object assumption
5.2.3. The mutual exclusivity assumption

6. Conclusion

7. References

1. Introduction

Every speaker of a language has a mental lexicon which stores every word the speaker knows. The mental lexicon has huge capacities (translated from Rothweiler 2001:21). Adults usually underestimate the size of their vocabulary. Aitchison (2003:6) refers to a study by Seashore and Eckerson (1940) in which they estimated that an educated adult knows more than 58,000 common 'basic words', 1,700 rare 'basic words' and 96,000 derivates and compounds. This is an overall total of 150,000 words of which 90% are used actively. Aitchison (2003:6-7) argues that "this figure is controversial, because of the problems of defining 'word' and the difficulty of finding a reliable procedure for assessing vocabulary knowledge". Rothweiler (2001:21) refers to Miller (1993) who estimates the vocabulary of an average educated adult at 80,000 words.

Rothweiler (2001:21) argues that the mental lexicon is more than a passive memory. She refers to Levelt (1989:181) that the lexicon is a central station between conceptualization on the one hand and the grammatical and phonological coding of language on the other hand. Both the acquisition of words and the usage of words refer to lexical processes.

The lexicon seems to be well structured because words can be located in a split second (Aitchison 2003:7). Both Rothweiler and Aitchison refer to confirmed experiments that native speakers can recognize a word of their mother tongue in 200 ms or less from its onset and can reject a non-word sound sequence in about half a second. Production of a word is slightly slower: the lexicon is able to generate up to six syllables a second, "making three or more words […] fairly standard" (Aitchison 2003:8). Errors of selection are rare – one in about a thousand words.

This paper is about the mental lexicon. First, I will compare the mental lexicon to a dictionary and come to the conclusion that it differs radically from a dictionary in terms of organization and content. Second, I will focus on 'words', what they are and whether they have a fixed meaning. I will outline Rosch's prototype theory and sense relations. Third, I will come to lexical processing and describe two processing models: the serial-autonomous models and the parallel-interactive models. My last chapter will be an overview on early language acquisition and constraints on word meaning.

2. Dictionary vs. Mental Lexicon

The hypothesis is "that there is […] little similarity between words in our minds and words in book dictionaries" (Aitchison 2003:10).

Language is a communication system employing arbitrary symbols. These symbols, normally words, have to be stored. […] [Different] techniques of storage are available: words can be listed in a reference book, […] or they can be kept in the mind. […] Word-stores that are primarily consulted for the reason of information retrieval are referred to as dictionaries. By contrast, word-stores that constitute a component within a natural language processing system are called lexica. […] The most common types of book dictionaries are encyclopedic, monolingual dictionaries, bilingual dictionaries, or dictionaries for special purposes such as synonym dictionaries, foreign word dictionaries, etc. […] A lexicon, by contrast, is the central module of a natural language processing system […]. It closely interacts with the other components of the language processor and provides detailed information about the words to be produced or comprehended. (Handke 1995:49-50).

Dictionaries and the mental lexicon do not only differ, they also have one thing in common: "They contain a large number of items which are defined linguistically" (Handke 1995: 50). This similarity is, however, the only one. The mental lexicon differs radically from a dictionary. Dictionaries contain words in alphabetical order. Aitchison (2003:10) argues that one could guess that the mental lexicon might be organized in alphabetical order as well, since a person who can read and write spends a lot of time "looking things up alphabetically". But the mental lexicon is not organized in alphabetical order, which can easily be tested by looking at mistakes people make when they are "selecting one word in error for another" (ibid.)

In error for the word 'guitar' one might expect someone to accidentally pick guinea or guipure or guise […] all words which are near neighbors in NODE (The New Oxford Dictionary of English). (Aitchison 2003:11)

But this does not seem to be the case as

mistakes of this type are quite unlikely, as [it] becomes clear when we look at a few 'slips of the tongue'. such as 'He told a funny antidote', with antidote instead of 'anecdote'. (ibid.)

These errors might hint at the fact that the mental lexicon might be "partially organized in terms of initial sounds" (ibid.) but definitely not in alphabetical order. Other hints towards the mental lexicon's phonological organization are sound structures, stress patterns, sound associations, etc. As the mental lexicon does not only concern spoken language, graphology has to be considered as well. Handke (1995:70) refers to McKay's (1970) so-called tip-of-the-tongue phenomena (see also Aitchison 2003:24). One might know the first consonant or the number of syllables of a word one is looking for, but it cannot be retrieved from the lexicon as bits seem to be missing. Another way to organize the mental lexicon are sense relations. If one cannot come up with a word but knows how the item looks like, one could utter that it has to be something similar to… .

Generally speaking entries in a dictionary include access unit (stem or graphological structure), phonological specification, grammatical aspects (morphology and syntax), meaning, further aspects, e.g. history, examples in context, alternative spellings (depending on dictionary type) (see Handke 1995:62). Entries in the lexicon need to be specified for aspects of phonology, graphology, morphology, syntax and semantics (Handke 1995:68).

So far, the organization of both dictionary and mental lexicon has been dealt with, but the content differs as well. A book dictionary contains a fixed countable number of words. "Book dictionaries are therefore inescapably outdated, because language is constantly changing, and vocabulary fastest of all" (Aitchison 2003:11). Book dictionaries are therefore somewhat limited. "They contain a relatively small amount of information about each item" (Aitchison 2003:15). The knowledge per entry in the mental lexicon would let any dictionary burst:

illustration not visible in this excerpt

The mental lexicon offers more knowledge per entry (as seen above) and does not contain a fixed number of words compared to book dictionaries. It is also far more productive, new words can be added instantly, we are able to form rhymes once we heard a word and coming back to the introduction we can reject a non-word sound sequence of our mother tongue within half a second. Looking through a book dictionary for such a non-word would take much longer although with the help of computers access is a lot easier.

2.1. The frequency effect

"It is a well-known observation that high-frequency words are processed faster" (Bonin 2004:xi). "Frequent words are easier to get at than infrequent words" (Handke 1995:267). Searching for a word in a book dictionary which one has never heard before does not take any longer than searching for a word which one has come across many times. The frequency effect argues that high-frequency words are found quicker in the mental lexicon because they are stored in many different areas, i.e. high-frequency words are represented more than once in the brain. A book dictionary organized in such a way that more frequent words appear at the beginning whereas less frequent words have to move towards the end would be rather useless. Because who would define which words to place where? People would have to search the whole dictionary in order to find the word they are looking for.

Last but not least, (translated from Quetz (1998:273) quoting Koll-Stobbe (1994:56)): looking for a word in a dictionary is a conscious activity whereas the usage of the mental lexicon is not. It is something like a system which we activate automatically when we are speaking or reading. Only when we are facing a problem, e.g. we come across a word we do not know or we want to say something but it is on the tip of our tongue, will our attention be drawn towards this aspect of speech processing.

The mental lexicon offers such a vast variety of aspects that it cannot only be regarded as a lexicon but as storage in multiple ways: As a library, a computer memory, an attic, a pigeon loft or even a treasure chest.

3. Words

3.1. What is a word?

What does the mental lexicon contain? Words. What is a word? Intuitively, everybody seems to know what it is, but it has been extremely difficult to provide a comprehensive definition of a word. (Piasecka 2001:7)

So what exactly is a word? People are able to intuitively isolate words from sentences. The same goes for children, even if they are too young to have a concept of metalanguage (translated from Rothweiler 2001:28). Piasecka (2001:7) argues that "there is no single, universal definition of a word". Linguists do not have one definition because a phonetician will give a different definition than a morphologist.

A phonetician is interested in spoken discourse, so for him a word in isolation consists of one or more phonemes and is characterized by stress. When individual words are combined into an utterance, stress of the words carrying meaning is kept, but the so-called grammatical words may lose their stress and become unstressed (ibid.)

A word on the morphological level is an independent unit. Because sentences are made out of words it is necessary that words are isolable, movable and replaceable (translated from Rothweiler 2001:29).

In written discourse a word is defined in terms of orthography […] 'a word is any sequence of letters […] bounded on either side by a space or a punctuation mark (Piasecka 2001:7-8).

We might think of words as separate items as the concept of a word in the Western world is influenced by written language: written words can be easily perceived and counted. But words in a spoken language appear rather as a string of sounds than separate items. Spoken words can be isolated by pausing but if we think of 'sandhi r' in non-rhotic English accents there are also ways of omitting these pauses.

[...]