Semantic diversity: A measure of semantic ambiguity based on variability in the contextual usage of words

Semantic ambiguity is typically measured by summing the number of senses or dictionary definitions that a word has. Such measures are somewhat subjective and may not adequately capture the full extent of variation in word meaning, particularly for polysemous words that can be used in many different ways, with subtle shifts in meaning. Here, we describe an alternative, computationally derived measure of ambiguity based on the proposal that the meanings of words vary continuously as a function of their contexts. On this view, words that appear in a wide range of contexts on diverse topics are more variable in meaning than those that appear in a restricted set of similar contexts. To quantify this variation, we performed latent semantic analysis on a large text corpus to estimate the semantic similarities of different linguistic contexts. From these estimates, we calculated the degree to which the different contexts associated with a given word vary in their meanings. We term this quantity a word’s semantic diversity (SemD). We suggest that this approach provides an objective way of quantifying the subtle, context-dependent variations in word meaning that are often present in language. We demonstrate that SemD is correlated with other measures of ambiguity and contextual variability, as well as with frequency and imageability. We also show that SemD is a strong predictor of performance in semantic judgments in healthy individuals and in patients with semantic deficits, accounting for unique variance beyond that of other predictors. SemD values for over 30,000 English words are provided as supplementary materials.     

Some hypotheses concerning the evolution of polysemous words

The evolution of polysemous words is examined in three studies. Study 1 indicates that older words are more polysemous than recent words, and that frequently used words are more polysemous than infrequently used words. These effects were replicated, controlling for word concreteness (Study 2) and evaluation (Study 3), using two other samples of words. If words evolve additional meanings through metaphor (MacCormac, 1985), then variables (such as concreteness and evaluation) that play a role in metaphor may mediate the evolution of polysemy. The correlation between concreteness and polysemy was negligible (Study 2). As hypothesized, evaluation is a significant predictor of polysemy: Negative words are more polysemous than positive words, and relatively neutral words are more polysemous than evaluatively extreme words. These findings are consistent with the notion that polysemous words evolve, in part, through metaphor.     

Polysemy and the subjective lexicon: Semantic relatedness and the salience of intraword senses

Subjective perceptions of the senses of polysemous English words are collected in questionnaire studies and the effects of variability in semantic distances among these senses are examined in an experiment. In the first of two questionnaire studies, native speakers produce meanings for 175 polysemous words; from their responses, the most frequently produced meaning for each word is identified as its dominant sense. In a second questionnaire, independent subjects rate the semantic relatedness between the dominant meaning and the other senses generated for each word in the first study. Relatedness measures vary, raising the possibility that polysemous words vary in terms of the salience of their different senses in different contexts. This is confirmed in an experiment showing that salience ratings are influenced by the interacting factors of sentential context, extent of relatedness of the senses, and the dominance status of the senses.This research was supported by a grant from the Special Research Fund, University of Western Australia, to Kevin Durkin.     

Not all ambiguous words are created equal: An EEG investigation of homonymy and polysemy

Event-related potentials (ERPs) were used to investigate the time-course of meaning activation of different types of ambiguous words. Unbalanced homonymous ("pen"), balanced homonymous ("panel"), metaphorically polysemous ("lip"), and metonymically polysemous words ("rabbit") were used in a visual single-word priming delayed lexical decision task. The theoretical distinction between homonymy and polysemy was reflected in the N400 component. Homonymous words (balanced and unbalanced) showed effects of dominance/frequency with reduced N400 effects predominantly observed for dominant meanings. Polysemous words (metaphors and metonymies) showed effects of core meaning representation with both dominant and subordinate meanings showing reduced N400 effects. Furthermore, the division within polysemy, into metaphor and metonymy, was supported. Differences emerged in meaning activation patterns with the subordinate meanings of metaphor inducing differentially reduced N400 effects moving from left hemisphere electrode sites to right hemisphere electrode sites, potentially suggesting increased involvement of the right hemisphere in the processing of figurative meaning.     

Modelling the effects of semantic ambiguity in word recognition

Most words in English are ambiguous between different interpretations; words can mean different things in different contexts. We investigate the implications of different types of semantic ambiguity for connectionist models of word recognition. We present a model in which there is competition to activate distributed semantic representations. The model performs well on the task of retrieving the different meanings of ambiguous words, and is able to simulate data reported by Rodd, Gaskell, and Marslen-Wilson [J. Mem. Lang. 46 (2002) 245] on how semantic ambiguity affects lexical decision performance. In particular, the network shows a disadvantage for words with multiple unrelated meanings (e.g., bark) that coexists with a benefit for words with multiple related word senses (e.g., twist). The ambiguity disadvantage arises because of interference between the different meanings, while the sense benefit arises because of differences in the structure of the attractor basins formed during learning. Words with few senses develop deep, narrow attractor basins, while words with many senses develop shallow, broad basins. We conclude that the mental representations of word meanings can be modelled as stable states within a high-dimensional semantic space, and that variations in the meanings of words shape the landscape of this space.     

Probing Lexical Ambiguity: Word Vectors Encode Number and Relatedness of Senses

Lexical ambiguity—the phenomenon of a single word having multiple, distinguishable senses—is pervasive in language. Both the degree of ambiguity of a word (roughly, its number of senses) and the relatedness of those senses have been found to have widespread effects on language acquisition and processing. Recently, distributional approaches to semantics, in which a word's meaning is determined by its contexts, have led to successful research quantifying the degree of ambiguity, but these measures have not distinguished between the ambiguity of words with multiple related senses versus multiple unrelated meanings. In this work, we present the first assessment of whether distributional meaning representations can capture the ambiguity structure of a word, including both the number and relatedness of senses. On a very large sample of English words, we find that some, but not all, distributional semantic representations that we test exhibit detectable differences between sets of monosemes (unambiguous words; N = 964), polysemes (with multiple related senses; N = 4,096), and homonyms (with multiple unrelated senses; N = 355). Our findings begin to answer open questions from earlier work regarding whether distributional semantic representations of words, which successfully capture various semantic relationships, also reflect fine-grained aspects of meaning structure that influence human behavior. Our findings emphasize the importance of measuring whether proposed lexical representations capture such distinctions: In addition to standard benchmarks that test the similarity structure of distributional semantic models, we need to also consider whether they have cognitively plausible ambiguity structure.     

Making sense of word senses: the comprehension of polysemy depends on sense overlap

Studies of polysemy are few in number and are contradictory. Some have found differences between polysemy and homonymy (L. Frazier & K. Rayner, 1990), and others have found similarities (D. K. Klein & G. Murphy, 2001). The authors investigated this issue using the methods of D. K. Klein and G. Murphy (2001), in whose study participants judged whether ambiguous words embedded in word pairs (e.g., tasty chicken) made sense as a function of a cooperating, conflicting, or neutral context. The ambiguous words were independently rated as having low, moderate, or highly overlapping senses to approximate a continuum from homonymy to metonymic polysemy. The effects of meaning dominance were examined. Words with highly overlapping meanings (e.g., metonymy) showed reduced effects of context and dominance compared with words with moderately or low overlapping meanings (e.g., metaphorical polysemy and homonymy). These results suggest that the comprehension of ambiguous words is mediated by the semantic overlap of alternative senses/meanings.     

Judging a book by its cover and its contents: the representation of polysemous and homophonous meanings in four-year-old children

Unlike homophonous meanings, which are semantically unrelated (e.g., the use of bat to refer to a baseball bat and a flying rodent), polysemous meanings are systematically related to one another (e.g., the use of book, CD, and video to refer to physical objects, as in ‘the leather book’, or to the intellectual content they contain, as in ‘the profound book’). But do perceived relations among polysemous meanings reflect the presence of generative lexical or conceptual structures that permit the meanings of these words to shift? If so, these structures may also support children’s early representations of polysemous meanings. In four studies, we demonstrate (1) that four-year-old children can understand both the concrete and abstract meanings of words like book, (2) that when taught a novel label for one of these meanings, children can readily understand an extension of that label to the other meaning, and (3) that extension does not occur between two homophonous meanings, which share a common phonological form but are otherwise unrelated. We conclude that the polysemous meanings of words like book rely on a common representational base early in development, and suggest that this may be the result of foundational, generative properties of the lexicon or conceptual system.     

How Polysemy Affects Concreteness Ratings: The Case of Metaphor

Concreteness ratings are frequently used in a variety of disciplines to operationalize differences between concrete and abstract words and concepts. However, most ratings studies present items in isolation, thereby overlooking the potential polysemy of words. Consequently, ratings for polysemous words may be conflated, causing a threat to the validity of concreteness‐ratings studies. This is particularly relevant to metaphorical words, which typically describe something abstract in terms of something more concrete. To investigate whether perceived concreteness ratings differ for metaphorical versus non‐metaphorical word meanings, we obtained concreteness ratings for 96 English nouns from 230 participants. Results show that nouns are perceived as less concrete when a metaphorical (versus non‐metaphorical) meaning is triggered. We thus recommend taking metaphoricity into account in future concreteness‐ratings studies to further improve the quality and reliability of such studies, as well as the consistency of the empirical studies that rely on these ratings.     

Executive control influences linguistic representations

Although it is known that words acquire their meanings partly from the contexts in which they are used, we proposed that the way in which words are processed can also influence their representation. We further propose that individual differences in the way that words are processed can consequently lead to individual differences in the way that they are represented. Specifically, we showed that executive control influences linguistic representations by influencing the coactivation of competing and reinforcing terms. Consequently, people with poorer executive control perceive the meanings of homonymous terms as being more similar to one another, and those of polysemous terms as being less similar to one another, than do people with better executive control. We also showed that bilinguals with poorer executive control experience greater cross-linguistic interference than do bilinguals with better executive control. These results have implications for theories of linguistic representation and language organization.     

Contextual self-organizing map: software for constructing semantic representations

In this article, we introduce a software package that applies a corpus-based algorithm to derive semantic representations of words. The algorithm relies on analyses of contextual information extracted from a text corpus—specifically, analyses of word co-occurrences in a large-scale electronic database of text. Here, a target word is represented as the combination of the average of all words preceding the target and all words following it in a text corpus. The semantic representation of the target words can be further processed by a self-organizing map (SOM; Kohonen, Self-organizing maps, 2001), an unsupervised neural network model that provides efficient data extraction and representation. Due to its topography-preserving features, the SOM projects the statistical structure of the context onto a 2-D space, such that words with similar meanings cluster together, forming groups that correspond to lexically meaningful categories. Such a representation system has its applications in a variety of contexts, including computational modeling of language acquisition and processing. In this report, we present specific examples from two languages (English and Chinese) to demonstrate how the method is applied to extract the semantic representations of words.     

Word informativity influences acoustic duration: Effects of contextual predictability on lexical representation

Language-users reduce words in predictable contexts. Previous research indicates that reduction may be stored in lexical representation if a word is often reduced. Because representation influences production regardless of context, production should be biased by how often each word has been reduced in the speaker’s prior experience. This study investigates whether speakers have a context-independent bias to reduce low-informativity words, which are usually predictable and therefore usually reduced. Content word durations were extracted from the Buckeye and Switchboard speech corpora, and analyzed for probabilistic reduction effects using a language model based on spontaneous speech in the Fisher corpus. The analysis supported the hypothesis: low-informativity words have shorter durations, even when the effects of local contextual predictability, frequency, speech rate, and several other variables are controlled for. Additional models that compared word types against only other words of the same segmental length further supported this conclusion. Words that usually appear in predictable contexts are reduced in all contexts, even those in which they are unpredictable. The result supports representational models in which reduction is stored, and where sufficiently frequent reduction biases later production. The finding provides new evidence that probabilistic reduction interacts with lexical representation.     

Sensitivity to local sentence context information in lexical ambiguity resolution: evidence from left- and right-hemisphere-damaged individuals

Using a cross-modal semantic priming paradigm, the present study investigated the ability of left-hemisphere-damaged (LHD) nonfluent aphasic, right-hemisphere-damaged (RHD) and non-brain-damaged (NBD) control subjects to use local sentence context information to resolve lexically ambiguous words. Critical sentences were manipulated such that they were either unbiased, or biased toward one of two meanings of sentence-final equibiased ambiguous words. Sentence primes were presented auditorily, followed after a short (0 ms) or long (750 ms) interstimulus interval (ISI) by the presentation of a first- or second-meaning related visual target, on which subjects made a lexical decision. At the short ISI, neither patient group appeared to be influenced by context, in sharp contrast to the performance of the NBD control subjects. LHD nonfluent aphasic subjects activated both meanings of ambiguous words regardless of context, whereas RHD subjects activated only the first meaning in unbiased and second-meaning biased contexts. At the long ISI, LHD nonfluent aphasic subjects failed to show evidence of activation of either meaning, while RHD individuals activated first meanings in unbiased contexts and contextually appropriate meanings in second-meaning biased contexts. These findings suggest that both left (LH) and right hemisphere (RH) damage lead to deficits in using local contextual information to complete the process of ambiguity resolution. LH damage seems to spare initial access to word meanings, but initially impairs the ability to use context and results in a faster than normal decay of lexical activation. RH damage appears to initially disrupt access to context, resulting in an over-reliance on frequency in the activation of ambiguous word meanings.     

INFLUENCES OF LEXICAL PROCESSING ON READING

Reading begins with the very early lexical processing, which is no less complex than the whole reading task. The purpose of the present study is to investigate how early lexical processing (word recognition) could influence reading. The results of this study reveal several facts. First of all, the reader does not necessarily rely on every letter in order to recognize a word; contexts do help the reader determine word patterns. Secondly, when being requested to memorize a passage, the less-proficient reader tends to remember the physical forms of words only, but the proficient reader tends to remember the meanings of the passage in addition to physical words. Thirdly, a reader s speed of word recognition influences the speed of meaning retrieval and comprehension. It was found that less-proficient readers could not finish the task of word recognition within time limits and their accuracy rates were quite low, whereas the proficient readers processed the physical words immediately and translated them into meanings quickly in order to memorize the whole passage.     

汉语多义单字词的识别优势效应

运用词汇判断、简单命名及辨别命名任务,考察汉语多义单字词的识别优势效应。结果显示,在三个任务中均发现了显著的多义词识别优势效应。在词汇判断和简单命名任务中,多义词识别的优势效应表现在低频词中;在辨别命名任务中,多义词识别优势效应同时表现在高、低频词中。实验结果支持了PDP模型,讨论了多义词识别优势效应产生的机制     

Word learning does not end at fast-mapping: evolution of verb meanings through reorganization of an entire semantic domain

This paper explores the process through which children sort out the relations among verbs belonging to the same semantic domain. Using a set of Chinese verbs denoting a range of action events that are labeled by carrying or holding in English as a test case, we looked at how Chinese-speaking 3-, 5-, and 7-year-olds and adults apply 13 different verbs to a range of carrying/holding events. We asked how children learning Chinese originally divide and label the semantic space in this domain, how they discover the boundaries between different words, and how the meanings of verbs in the domain as a whole evolve toward the representations of adults. We also addressed the question of what factors make verb meaning acquisition easy or hard. Results showed that the pattern of children’s verb use is largely different from that of adults and that it takes a long time for children to be able to use all verbs in this domain in the way adults do. We also found that children start to use broad-covering and frequent verbs the earliest, but use of these verbs tends to converge on adult use more slowly because children could not use these verbs as adults did until they had identified boundaries between these verbs and other near-synonyms with more specific meanings. This research highlights the importance of systematic investigation of words that belong to the same domain as a whole, examining how word meanings in a domain develop as parts of a connected system, instead of examining each word on its own: learning the meaning of a verb invites restructuring of the meanings of related, neighboring verbs.     

熟练中-英双语者对多义词早期识别中语义和语法的相互作用

以中、英文的多义词为材料, 采用跨语言隐蔽启动的实验范式, 考察在无语境条件下熟练中–英双语者对多义词的早期识别。结果表明：(1)熟练中–英双语者表现出显著的隐蔽翻译启动效应, 当目标词是主要翻译对等词时, 启动效应更加显著。(2)翻译对等词的性质、多义词的义项联系程度以及是否是词类歧义词都影响熟练中–英双语者对中、英多义词的早期加工。(3)在无语境条件下, 词类信息在意义通达的早期就已经激活, 并且同语义信息之间存在着交互作用。根据研究结果, 扩展和修正了分布式概念表征模型。     

The Heuristic Power Of Oppositionality In An Incidental-Memory Task: In Support Of The Construing Process

From a personal construct view, construing is a top-down process in which wider meanings predicate narrower, targeted meanings. Predicate contexts are invariably oppositional, as Kelly's (1955) theory reflects. Two memory experiments using college subjects are presented. Subjects were asked to focus on a series of 30 target words to determine if they were similar in meaning to a predicating word (e.g., friendly). Ten of these target words were relevant (e.g., congenial), 10 were opposite (e.g., impolite), and 10 were irrelevant (e.g., abstract) in meaning to the predicating word. Subjects were then (unexpectedly) asked to recall as many words as possible. In line with experimental instructions, the majority of these recalled words were relevant to the predicating word. However, as predicted, in both experiments significantly more opposite than irrelevant words were recalled (p < .001). The results are in support of a personal construct view of human cognition.     

The categorization of sentential contexts

A new experimental method, involving the sorting of linguistic contexts, is shown to be effective in discriminating the contexts of polysemous words as well as the contexts of synonyms of those words. These results are interpreted as support for the claim that the method of sorting linguistic contexts is a valid technique for studying the contextual information available to support inferences about word meanings.The research reported in this article was part of the author's work submitted as a doctoral thesis to Princeton University. It was supported in part by Contract MDA 903-86-K-0242 between the Army Research Institute and Princeton University. The views and conclusions in this article are those of the author and should not be interpreted as representing official policies, either expressed or implied, of the Army Research Institute or Princeton University.     

The effect of polysemy on lexical decision time: Now you see it,now you don’t

Gernsbacher (1984) found that number of word meanings (polysemy) did not influence lexical decision time when it was operationalized as number of dictionary definitions. This finding supports her contention that subjects do not store all possible dictionary meanings for words in memory. The present experiments extended Gernsbacher's research by determining whether more psychologically valid measures of polysemy affect lexical decision time. Three metrics were used to represent the meanings that subjects actually access from memory (accessible polysemy): (1) the first meanings subjects think of when asked to define stimulus words, (2) all the meanings subjects generate for words, and (3) the average number of meanings subjects generate. The results showed that the second and third metrics of polysemy influenced lexical decision time, whereas the first metric (representing mostly the access to dominant meanings for words) only approached significance.