论文部分内容阅读
摘要:心理词汇也称为内部词汇或者心理字典,指的是储存于大脑中的词汇。桂诗春(2000)认为研究心理词汇应该主要关注词汇是怎样储存在记忆里,又是如何被提取的?为此学者们提出很多不同的模型理论。本文从储存模型和提取模型两个角度着手,总结概括原子球理论,分级网络模型,扩散激活模拟,并行提取模型和窜行提取模型,深入讨论它们的研究现状,存在问题以及进一步的研究建议。总结得出随着这些模型的不断修正,他们当初所强调的假设越来越微弱,它们之间的距离也越来越短。
关键词:心理词汇;储存模型;提取模型
中图分类号:H31文献标识码:A文章编号:1812-2485(2010)03-0039-008
Literature Review on Storage and Retrieval
Models of Mental Lexical
Abstract:Gui Shichun (2000) thinks that the study of the mental lexicon mainly concerns how words are stored in one's memory and how words are retrieved when they are needed in speech or writing. Thus researchers proposed many different model theories. And this paper concludes these model theories, their developments and problems, resulting that together with the improvement of the models, the researchers' differences and conflicts are becoming less and less.
Keywords:mental lexicon, storage models, retrieval models
1. Introduction
One of the most important developments in the word recognition literature in the last four decades was the assumption of a mental lexicon or internal lexicon. Treisman (1960), who proposed the concept of the mental lexicon first, defines the mental lexicon as a repository of all the information a reader or a listener has attained about words of his language. Carroll (2000) defines the mental lexicon as the organization of word knowledge in permanent memory.J.C.Richards et a1. (2000) defines mental lexicon as a mental system that contains all the information a person knows about words. And Aitchison (1987:9-10) compares the mental lexicon to a book dictionary. Mental lexicon is something like a paper dictionary, but not as visible as a paper dictionary with a series of entries. The most influential view is given by Gui Shichun (2000), who thinks the study of the mental lexicon mainly concerns how words are stored in one's memory and how words are retrieved when they are needed in speech or writing. The mental lexicon can be envisaged as composed of a number of grammatical (e.g. nouns, verbs) and semantic (e.g. animate/inanimate, living/still) categories. In other words, a capable speaker-hearer of a language possesses a huge and complicated mental lexicon. However, the biggest problem with the study of the mental lexicon is that we can not enter a human brain and have a look at what is going on there. What we can do is to observe what people do or fail to do in normal speech and from this observation build a model that captures as many features of human brain as possible.
2. Storage models of mental lexicon and their problems
How are words in our mental lexicon storied of organized? Are they arranged in the same level? Currently the main idea regarding the organization of the mental lexicon is set up as a network of interconnected elements. The elements are concepts or nodes, which are connected to one another by virtue of having various relations with one another (Carroll, 2000).
2.1 Atomic globule theory
Atomic globule theory was developed by German philosopher in the seventeenth century and was Leibniz, based on the belief that all human beings have some sort of basic alphabet of human thoughts. These thoughts are new categories and concepts to us, then combined with each other to give in a more modern version, these basic human thoughts have been named “semantic primitives” or “a set of basic atoms of meaning”(Aitchison, 1994) which We combine into new meanings. This also explains how we cope with the overlapping of meanings-it is in fact not whole words that overlap, only the ingredients from which words are combined. An example will illustrate this:
The image that provided by the Atomic globule theory is a good tool for understanding the complexity of the mental lexicon. It should be pointed out though, that no one has at this point in time been able to specify these semantic atoms, and according to Aitchison (1994:81)“The arguments in favor ofthe viewpoint are based mainly on descriptive convenience and wishful thinking.”
2.2 Semantic Network:Hierarchical Network Models
Collins and Quillian (1969) proposed that concepts are stored hierarchically in our mental dictionary of words, or mental lexicon. Property relations are represented within the hierarchy; for example, the concept of BIRD would be stored at a node that is above CANARY and below ANIMAl. Connected to each category node are properties, such as HAS WINGS and CAN FLY for BIRD. Collins and Quillian predicted that the more closely situated on the hierarchy the concepts were, the faster they would activate each other; that is, people should be faster to say that the proposition “Birds have feathers” is true than they would say that “Birds eat” is true. This pattern of reaction times in the verification task was obtained, although for a reason different from that proposed by Collins and Quillian. Conrad (1972) conducted a similar property verification task, but he manipulated the frequency with which the category and its properties were associated.Rather than distance.The frequency of the relationships was responsible for the faster responses. Problems soon emerged with this model, however. The model assumed that all items on a given level of the hierarchy were more or less equal. Canary and ostrich, for example, were both subordinates of bird and one link away from bird, so they should take equal time to verify. In fact, they do not. Smith, Shoben, and Rips (1974) found that “A robin is a bird” takes less time than“An ostrich is a bird.”
2.3 Spreading Activation Models
Collins and Loftus (1975) assume that words are represented in the internal lexicon in a network, but the organization is not strictly hierarchical. In contrast, the organization is closer to a web of interconnecting nodes. The model is built on a complex association network in which specific memories age distributed in conceptual space with related concepts that are linked by associations. More specifically, concepts are represented in memory as nodes and relations between the concepts as associative pathways between the nodes. When part of the memory network is activated, activation spreads along the associative pathways to related areas in memory. This spread of activation serves to make these related areas of the memory network more available for further cognitive processing (Balota and Lorch, 1986). The process of spreading activation has been linked to the effect of dropping a rock into a pool of still water. This model, however, has some limitations. Very little attention is paid to phonological, syntactic, and morphological aspects of words.
A more recent spreading activation model that incorporates lexical as well as conceptual aspects is presented by Bock and Levelt (1994). They assume that our knowledge of words exists at three different levels. The concept level consists of nodes that represent concepts; nodes are connected to other nodes by various connections. A second level is called the lemma level which specifies the syntactic aspects of word knowledge. Finally, there is a lexeme level. A lexeme captures a word's phonological properties. The distinction between these different levels is especially useful in explaining the TOT phenomenon, in which the speaker knows the word’s meaning (concept) and syntactic category (the lemma) but not its phonological features (the lexeme), at least not in their entirety. Despite its considerable appeal, this model still has its own share of problems. It puts much emphasis on sense relations and the referential aspects of word meaning are not dealt with. Nonetheless, spreading activation provides a plausible framework within which to think about the concept of lexical organization and lexical access (Carroll, 2000).
3. Retrieval models of mental lexicon and their problems
Carroll (2000) refers to the process by which we activate meanings as lexical access. It studies the mechanisms involved in the retrieval of words from memory for their recognition and production. Another fascinating issue is how words stored in the mental lexicon are accessed.According to Gui Shichun(2000), there are two main theoretical models-serial search model (also called direct model) and parallel access models (also called indirect model).
3.1 Serial search model
One of the earliest and most influential models is the autonomous search model of Forster (1976, 1979). The serial search model holds that, when encountering a word, we refer to a word list to decide whether the item is a word and then to retrieve the information (for example, meaning) about this word. This model implies that the procedure of searching for the lexical item is going on in a word-by-word order. Garman (1990) made an analogy of the process based on searching a particular book in a library. We do not go to the main shelf of the book straightly, and instead we go to the catalogue, which the particular book belongs to. We know that libraries classify books under different topic areas that are housed on different floors of the building. Searching through the catalogue, we find something that matches what we are looking for: but what we retrieve from this stage of process, not the book itself, but an abstract location marker, telling us where to find the book on the shelves. With this location marker, we carry out the second stage of the process, by using the marker to guide us to the right book on the shelves.
3.2 Parallel access model
Different from the serial search models, different stages of lexical access are mixed and paralleled in this parallel access model. Among those parallel access models, Morton’s logogen model (1969) and Marslen-Wilson’s cohort model (1987) are the most typical ones. The former mainly derives from the work on visual word recognition, and the latter aims at accounting for auditory word recognition.
In Morton’s model, each word (or morpheme)in the lexicon is represented as a logogen, which specifies the word’s various attributes (semantic, orthographic, phonological,and so on). The logogen is activated in either of two ways: by sensory input and by contextual information. Consider first the sensory route. As orthographic or phonological features of the input stimulus arc detected, they are matched to the logogen. The logogen functions as a scoreboard or counter; when the counter rises above a predesignated threshold, the item is recognized.With regard to contextual information, the semantic and syntactic structure of a sentence may influence the activation of the logogen for a given word.In the logogen model,these two routes are assumed to work in parallel;sensory and contextual matches increase the same counter.If both sensory and contextual features are detected, the word is easily recognized.After that, the logogen returns to its initial state only gradually, remaining partially activated for a while. The cohort model was designed specifically to account for auditory word recognition. According to Marslen-Wilson (1987), spoken word recognition occurs in three stages. First, on the basis of an acoustic-phonetic analysis of the input, a set of lexical candidates is activated. This set is referred to as the word initial cohort. Second, one member of the cohort is selected for further analysis.Finally, the selected lexical item is integrated into ongoing semantic and syntactic context.
Compared with serial search model, logogen model and cohort model are more plausible and seem to be more realistic. But it is worth noting that both the two models hold that the recognition of the word is an automatic process instead of a conscious observation, that is, we cannot sense the activating process of searching or storing the word. What we have is nothing but the end of this procedures-recognition of the word and its meaning. The two models attempt to deal with the effect from different orientations, but neither of them can give a completely satisfactory explanation for all the effects. All these semantic networks and lexical access models are introduced in general, not in details. They are still controversial. There are many researchers who are trying to justify and falsify these theories by different psycholinguistic experiments. Even so the theories examined above still provide a theoretical basis for L2 lexical development studies and have much implication for both vocabulary acquisition theory building and vocabulary teaching in TESL.
Conclusions
This paper mainly shows a literature review on the most important aspect of the mental lexicon, that is the storage and retrieval models of mental lexicon. And these models and hypotheses are devised or made to explain how vocabulary is stored and activated in a language in general. However, their research methods are not convincing enough to prove those corresponding models or illustrations of mental lexical as the main research methods are observation: to collect the speech errors in everyday speech, and controlled experiments. And together with the improvement of these models, the researchers’ hypotheses are all becoming weakened and their differentials are becoming shortened.
References:
[1]Carroll, David W.(2000). Psychology of Language. Beijing: Foreign Language and Teaching and Research Press.
[2]Collins, A. M. & Loftus, E. F. (1975). A spreading-activation theory of semantic processing. Psycho-logical Review, 82, 407-428.
[3]Collins, A. M., &Quillian, M. R. (1969). Retrieval time from semantic memory. Journal of Verbal learning and Verbal Behavior, 8, 240-247.
[4]Ellis, R. (1994). The Study of Second Language Acquisition, Oxford: Oxford University Press. Michael Garman. (2002). Psycho-linguistics. Beijing: Peking University Press.
[5]丁瑶.大学英语词汇习得中心理词汇的渐进[R].第二届中国外语教学法国际研讨会论文集,2007(827-836)
[6]桂诗春.新编心理语言学[M].上海:上海外语教育出版社,2000
[7]张淑静.重组二语心里词汇[J].四川外语学院学报,2004
关键词:心理词汇;储存模型;提取模型
中图分类号:H31文献标识码:A文章编号:1812-2485(2010)03-0039-008
Literature Review on Storage and Retrieval
Models of Mental Lexical
Abstract:Gui Shichun (2000) thinks that the study of the mental lexicon mainly concerns how words are stored in one's memory and how words are retrieved when they are needed in speech or writing. Thus researchers proposed many different model theories. And this paper concludes these model theories, their developments and problems, resulting that together with the improvement of the models, the researchers' differences and conflicts are becoming less and less.
Keywords:mental lexicon, storage models, retrieval models
1. Introduction
One of the most important developments in the word recognition literature in the last four decades was the assumption of a mental lexicon or internal lexicon. Treisman (1960), who proposed the concept of the mental lexicon first, defines the mental lexicon as a repository of all the information a reader or a listener has attained about words of his language. Carroll (2000) defines the mental lexicon as the organization of word knowledge in permanent memory.J.C.Richards et a1. (2000) defines mental lexicon as a mental system that contains all the information a person knows about words. And Aitchison (1987:9-10) compares the mental lexicon to a book dictionary. Mental lexicon is something like a paper dictionary, but not as visible as a paper dictionary with a series of entries. The most influential view is given by Gui Shichun (2000), who thinks the study of the mental lexicon mainly concerns how words are stored in one's memory and how words are retrieved when they are needed in speech or writing. The mental lexicon can be envisaged as composed of a number of grammatical (e.g. nouns, verbs) and semantic (e.g. animate/inanimate, living/still) categories. In other words, a capable speaker-hearer of a language possesses a huge and complicated mental lexicon. However, the biggest problem with the study of the mental lexicon is that we can not enter a human brain and have a look at what is going on there. What we can do is to observe what people do or fail to do in normal speech and from this observation build a model that captures as many features of human brain as possible.
2. Storage models of mental lexicon and their problems
How are words in our mental lexicon storied of organized? Are they arranged in the same level? Currently the main idea regarding the organization of the mental lexicon is set up as a network of interconnected elements. The elements are concepts or nodes, which are connected to one another by virtue of having various relations with one another (Carroll, 2000).
2.1 Atomic globule theory
Atomic globule theory was developed by German philosopher in the seventeenth century and was Leibniz, based on the belief that all human beings have some sort of basic alphabet of human thoughts. These thoughts are new categories and concepts to us, then combined with each other to give in a more modern version, these basic human thoughts have been named “semantic primitives” or “a set of basic atoms of meaning”(Aitchison, 1994) which We combine into new meanings. This also explains how we cope with the overlapping of meanings-it is in fact not whole words that overlap, only the ingredients from which words are combined. An example will illustrate this:
The image that provided by the Atomic globule theory is a good tool for understanding the complexity of the mental lexicon. It should be pointed out though, that no one has at this point in time been able to specify these semantic atoms, and according to Aitchison (1994:81)“The arguments in favor ofthe viewpoint are based mainly on descriptive convenience and wishful thinking.”
2.2 Semantic Network:Hierarchical Network Models
Collins and Quillian (1969) proposed that concepts are stored hierarchically in our mental dictionary of words, or mental lexicon. Property relations are represented within the hierarchy; for example, the concept of BIRD would be stored at a node that is above CANARY and below ANIMAl. Connected to each category node are properties, such as HAS WINGS and CAN FLY for BIRD. Collins and Quillian predicted that the more closely situated on the hierarchy the concepts were, the faster they would activate each other; that is, people should be faster to say that the proposition “Birds have feathers” is true than they would say that “Birds eat” is true. This pattern of reaction times in the verification task was obtained, although for a reason different from that proposed by Collins and Quillian. Conrad (1972) conducted a similar property verification task, but he manipulated the frequency with which the category and its properties were associated.Rather than distance.The frequency of the relationships was responsible for the faster responses. Problems soon emerged with this model, however. The model assumed that all items on a given level of the hierarchy were more or less equal. Canary and ostrich, for example, were both subordinates of bird and one link away from bird, so they should take equal time to verify. In fact, they do not. Smith, Shoben, and Rips (1974) found that “A robin is a bird” takes less time than“An ostrich is a bird.”
2.3 Spreading Activation Models
Collins and Loftus (1975) assume that words are represented in the internal lexicon in a network, but the organization is not strictly hierarchical. In contrast, the organization is closer to a web of interconnecting nodes. The model is built on a complex association network in which specific memories age distributed in conceptual space with related concepts that are linked by associations. More specifically, concepts are represented in memory as nodes and relations between the concepts as associative pathways between the nodes. When part of the memory network is activated, activation spreads along the associative pathways to related areas in memory. This spread of activation serves to make these related areas of the memory network more available for further cognitive processing (Balota and Lorch, 1986). The process of spreading activation has been linked to the effect of dropping a rock into a pool of still water. This model, however, has some limitations. Very little attention is paid to phonological, syntactic, and morphological aspects of words.
A more recent spreading activation model that incorporates lexical as well as conceptual aspects is presented by Bock and Levelt (1994). They assume that our knowledge of words exists at three different levels. The concept level consists of nodes that represent concepts; nodes are connected to other nodes by various connections. A second level is called the lemma level which specifies the syntactic aspects of word knowledge. Finally, there is a lexeme level. A lexeme captures a word's phonological properties. The distinction between these different levels is especially useful in explaining the TOT phenomenon, in which the speaker knows the word’s meaning (concept) and syntactic category (the lemma) but not its phonological features (the lexeme), at least not in their entirety. Despite its considerable appeal, this model still has its own share of problems. It puts much emphasis on sense relations and the referential aspects of word meaning are not dealt with. Nonetheless, spreading activation provides a plausible framework within which to think about the concept of lexical organization and lexical access (Carroll, 2000).
3. Retrieval models of mental lexicon and their problems
Carroll (2000) refers to the process by which we activate meanings as lexical access. It studies the mechanisms involved in the retrieval of words from memory for their recognition and production. Another fascinating issue is how words stored in the mental lexicon are accessed.According to Gui Shichun(2000), there are two main theoretical models-serial search model (also called direct model) and parallel access models (also called indirect model).
3.1 Serial search model
One of the earliest and most influential models is the autonomous search model of Forster (1976, 1979). The serial search model holds that, when encountering a word, we refer to a word list to decide whether the item is a word and then to retrieve the information (for example, meaning) about this word. This model implies that the procedure of searching for the lexical item is going on in a word-by-word order. Garman (1990) made an analogy of the process based on searching a particular book in a library. We do not go to the main shelf of the book straightly, and instead we go to the catalogue, which the particular book belongs to. We know that libraries classify books under different topic areas that are housed on different floors of the building. Searching through the catalogue, we find something that matches what we are looking for: but what we retrieve from this stage of process, not the book itself, but an abstract location marker, telling us where to find the book on the shelves. With this location marker, we carry out the second stage of the process, by using the marker to guide us to the right book on the shelves.
3.2 Parallel access model
Different from the serial search models, different stages of lexical access are mixed and paralleled in this parallel access model. Among those parallel access models, Morton’s logogen model (1969) and Marslen-Wilson’s cohort model (1987) are the most typical ones. The former mainly derives from the work on visual word recognition, and the latter aims at accounting for auditory word recognition.
In Morton’s model, each word (or morpheme)in the lexicon is represented as a logogen, which specifies the word’s various attributes (semantic, orthographic, phonological,and so on). The logogen is activated in either of two ways: by sensory input and by contextual information. Consider first the sensory route. As orthographic or phonological features of the input stimulus arc detected, they are matched to the logogen. The logogen functions as a scoreboard or counter; when the counter rises above a predesignated threshold, the item is recognized.With regard to contextual information, the semantic and syntactic structure of a sentence may influence the activation of the logogen for a given word.In the logogen model,these two routes are assumed to work in parallel;sensory and contextual matches increase the same counter.If both sensory and contextual features are detected, the word is easily recognized.After that, the logogen returns to its initial state only gradually, remaining partially activated for a while. The cohort model was designed specifically to account for auditory word recognition. According to Marslen-Wilson (1987), spoken word recognition occurs in three stages. First, on the basis of an acoustic-phonetic analysis of the input, a set of lexical candidates is activated. This set is referred to as the word initial cohort. Second, one member of the cohort is selected for further analysis.Finally, the selected lexical item is integrated into ongoing semantic and syntactic context.
Compared with serial search model, logogen model and cohort model are more plausible and seem to be more realistic. But it is worth noting that both the two models hold that the recognition of the word is an automatic process instead of a conscious observation, that is, we cannot sense the activating process of searching or storing the word. What we have is nothing but the end of this procedures-recognition of the word and its meaning. The two models attempt to deal with the effect from different orientations, but neither of them can give a completely satisfactory explanation for all the effects. All these semantic networks and lexical access models are introduced in general, not in details. They are still controversial. There are many researchers who are trying to justify and falsify these theories by different psycholinguistic experiments. Even so the theories examined above still provide a theoretical basis for L2 lexical development studies and have much implication for both vocabulary acquisition theory building and vocabulary teaching in TESL.
Conclusions
This paper mainly shows a literature review on the most important aspect of the mental lexicon, that is the storage and retrieval models of mental lexicon. And these models and hypotheses are devised or made to explain how vocabulary is stored and activated in a language in general. However, their research methods are not convincing enough to prove those corresponding models or illustrations of mental lexical as the main research methods are observation: to collect the speech errors in everyday speech, and controlled experiments. And together with the improvement of these models, the researchers’ hypotheses are all becoming weakened and their differentials are becoming shortened.
References:
[1]Carroll, David W.(2000). Psychology of Language. Beijing: Foreign Language and Teaching and Research Press.
[2]Collins, A. M. & Loftus, E. F. (1975). A spreading-activation theory of semantic processing. Psycho-logical Review, 82, 407-428.
[3]Collins, A. M., &Quillian, M. R. (1969). Retrieval time from semantic memory. Journal of Verbal learning and Verbal Behavior, 8, 240-247.
[4]Ellis, R. (1994). The Study of Second Language Acquisition, Oxford: Oxford University Press. Michael Garman. (2002). Psycho-linguistics. Beijing: Peking University Press.
[5]丁瑶.大学英语词汇习得中心理词汇的渐进[R].第二届中国外语教学法国际研讨会论文集,2007(827-836)
[6]桂诗春.新编心理语言学[M].上海:上海外语教育出版社,2000
[7]张淑静.重组二语心里词汇[J].四川外语学院学报,2004