Summary of¡@ ¡§Therepresentation of polysemous words¡¨

¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@ ¡@¡@¡@¡@¡@¡@¡@ ¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@¡@ Emile


Main argument: Polysemous word have the separate representationsfor each sense and that any core meaning is minimal.


(1)Five experiments:

¡@¡@¡@¡@¡@¡@¡@ a. Experiment 1showed that sense consistency aided memory for the polysemous words.

¡@¡@¡@¡@¡@¡@¡@ b. Experiment 2extended this result to a timed sensicality judgment task.

¡@¡@¡@¡@¡@¡@¡@ c. Experiment 3demonstrated that the effects for polysemous words were very similar to thosefor homonyms.

¡@¡@¡@¡@¡@¡@¡@ d. Experiment 4 ruled out the possibility of modifier-modifierpriming.

¡@¡@¡@¡@¡@¡@¡@ e. Experiment 5showed that sense consistency facilitates comprehension relative to a neutralbaseline, while sense inconsistency inhibits comprehension.


Polysemy:The phenomenon of words having multiple related senses iscalled polysemy.

Polysemy is the normal, expected presence of related senses in aword, such as an object and the substance making up that object.

Homonym: The words having multiple senses which have no obviousrelation. Homonymy is the unpredictable coincidence of two different wordshaving the same name.


Two view:

(1) Lexical meanings can be argumented or extended in a givencontext, but those extended senses are not permanently stored in the lexicon.

(2)Common senses would have separate entries connected to the samelemma.



1.Clark & Clark, 1979; Sweester, 1990: The uses of the word¡§paper¡¨ are related to one another and clearly arose through a process ofextension of similar meanings rathr than through an arbitrary historicalcoincidence.


2.Lehrer,1990;Nunberg 1979: Certain semantic relations between aword¡¦s senses appear over and over in polysemy, for example, object/substance,object/representation of that object, type/token, and text/object containingthat text.


3.Murphy, 1997: According to above, these forms of polysemy arehighly productive, and they are used quite easily when new words enter thelexicon.


4.Levelt, 1989: The outline of a theory of homonymicrepresentation are fairly clear. The different meanings of bank or calf areconsidered to be different words, so it is generallybelieved that they arerepresented by different lemmas(lexical units).


5.Zgusta ,1971: (a)In lexicology, there also seems to be a beliefthat homonyms are different words, as indicated by separate dictionaryentries.¡@ (b)Each distinguishable senseis separately represented.


6.Ruhl, 1989: There is a single represented sense that accountsfor all these uses of a word.


7. Cruse 1986; Geeraerts 1993: Polysemy is used to refer to a wordhaving related senses.


8. Nunberg, 1979: he argued against the idea that all distinctsenses should be represented in the lexicon. Instead, he proposed thatpragmatic principles could be used to derive word senses from others.(Nunberg¡¦sargument concerned how a linguistic theory of the lexicon should representdifferent sense. it was not presented as a psychological theory orrepresentation and processing). On this view, different sense are not prestoredbut are rather computed from contextual features.


9.Caramazza and Grober (1976): They identified 26 separate butrelated senses for word line. They suggested that these senses are all relatedto a core meaning and that ¡§it is precisely the core meaning that is stored inthe psychological representation for the meaning of line¡¨. They argued againstthe notion that each sense is explicitly stored in the mental lexicon.


20. Ruhl 1989: He argued that there is a single, defining sensefor words (even most homonyms), with distinct senses neither created or stored.


21. Lehrer 1990: She agreed with Bumberg that much polysemy can bepredicted through general principles of meaning extension, but she also notedthat these principles sometimes fail. She argued that the lexicon is simplyunpredicted to some degree and that language users must learn which words canbe extend in which ways, rather relying entirely pragmatic principles.


22. Zgusta 1971: he argued that it is usually impossible to find asingle basic senses can be derived.


23. Williams 1992: he found that contexually irrelevant senes ofpolysemous word are active even over long delays in a lexical decision task. Hecompared this to results from the homonym literature, which show that primingfor the contexuality irrelevant meaning of homonymous word is short-lived.Therefor Williams argued that the senes of polysemous words cannot berepresented independently, as homonym meanings are. One possibility is that thepolysemous sense are connected through a common core.


24. Anderson and Ortony 1975: They argued that understanding ismore than finding the correct lexical entry in a semantic associativenetwork.¡@ They seem to be suggestingthat the lexical network contains core information, and other informationnecessary to understand the exact sense of the word is suppiled by context.


25.Murphy 1997: he showed that novel extensions of¡@ a word that were closely related topreviously known senses were more acceptable than were more distantly relatedextensions. This suggests that polysemy could develop by the construction of achain of extensions , each building on its predecessots.


¡@¡@¡@¡@¡@¡@¡@ However , althoughnovel uses of a word may be comprehensible, this does not mean that commonlyencountered senses are not stored. Murphy used novel words and novelextensions, but it is possible that many senses of actual words are representedin memory. It is the goal of the authors to investigate this question. The mainquestion being investigated in the present experiments is the degree to whichdifferent senses of polysemous words use the same or different representation.

¡@¡@¡@¡@¡@¡@¡@ The problem inbeginningan investigation of polysemy is that there are few explicit models ofprocesses of polysemous words. Linguistic approaches virtually never discussprocessing issues.

¡@¡@¡@¡@¡@¡@¡@ Their strategy was toinvestigate the amount of overlap in different sense, using a priming techniquein which one use of a word was followed by a subsequent use that involved thesame or different sense.

¡@¡@¡@¡@¡@¡@¡@ Although our resultswill not be able to narrow the field down to a single model, it is neverthlessimportant to begin to perform empirical work that will elucidate therepresentation of polysemy, because of its implications for pur understandingof lexical processing and representation.



(1)This experiment used memory performance as measure of therepresentation of polysemous senses. It investigated whether people are betterable to recognized a word uses in the same sense of a different sense than itsoriginal presentation.

(2) This experiment was based on a paradigm developed by Light andCarter-Sobell(1970). In their experiment, most of the test words would becounted as homonyms rather than as polysemous words.

(3) Material: Twenty-four polysemous words were used.

¡@¡@¡@¡@ Procedure:Subject viewedthe mateirals on a Macintosh Quadra 630 computer, which was connected to aPsyScope button box. Their dominant hands were designed to the YES response andtheir nondminant hands to the NO response.

(4) Results and Discussion: The results indicate that the way apolysemous word is processed initally affects later memory access. Theseresults lend support to thte htpothesis that are stored separately and areevidence against a single core meaning hypothesis. In their experiment, asingle context word was not sufficient to force subjects to distinguish thesense of the learned words, whereas a setence ontext was. The most importantone is that their subjects read the words as a phrase and therefor presumablyinterpreted them as a coherent concept.



(1) Their main interest in Experiment 1 was not memory for sensesper se, but rather the issue of how polysemous words are represented andprocess. The result suggest that different senses are stored sparately. Thisresult is surprising enough to follow up in a task that involves comprehesionand semantic processing¡@ rather thanmemory. In this experiment, the task was to make a sense/nonsense judgment¡@ on phrase similar to those used in the firstmemory experiment. They again manipulated sense consistency by presentingphrases that involved the same or a different sense of a word and by lookingfor priming of a consistent use of a word compared to the inconsistent use.

(2) The result of Experiment 1 and 3 use very different techniquesto converge on the conclusion that the different senses of polysemous word arefunctionally distinct.


Experiment 3

(1) Experiment 3 served both replicate Experiment 2 and to gaugethe size of the consistency effect. In sum, homonyms provide a way to scale thesize of the consistency effect in polysemous words, since they represent thecase in which different meanings are completely unrelated.

(2) The size of the priming effect was very close to being equalfor hymonyms and polysemous words.


Experiment 4

(1) Experiment 4 was a control experiment that investigatedwhether priming of te modifiers themselves might be responsible for theconsistency effects obtained.

(2) It found strong consistency effects for both word tpyes. Thus,the results show very similar findings for hymonyms and polysemous words.


Ecperiment 5

(1) This experiment was conducted to examine whether the primingof polysemous word senses is due to inhibitory or facilitory process.

(2) The data indicate that multiple¡@ processes combine to give sense-consistent phrases an advantage.First, it appears that the consistent sense is being activated, whicheffectively lowers the threshold for subsequent activation. Second, theactivation of one sense appears to cause a dampening of activation fordifferent senses.


General Discussion

¡@¡@¡@¡@¡@¡@¡@ The sudies foundevidence for sense priming in polysemy. Using a word in a specific sensefacilitated comprehension for a phrase that used the word in the same sense andinhibited comprehension for a phrase used the word in the different sense. Theyfound that the effects of sense consistency in polysemy are similar to theeffects of meaning consistency in homonyms. Finally using a word in the samesense was a good memory cue, but using the word in a different sense was not.

¡@¡@¡@¡@¡@¡@¡@ In conclusion,polysemous senses are semantically related but are not very similar, resultingin the same-sense facilitation and cross-sense inhibition. If polysemous wordsdo have a core meaning, it cannot be a substantial semantic component that iscommon to all the senses of a word.


Critical: In Experiment 1, the authors use two example to show thepolysemous senses of ¡¥paper¡¦---sheets of a material (wrapping paper and shreddepaper) and newspaper (daily paper and liberal paper)--- the modifiers of formerhave a meaning of action, but the later do not have. The different type ofmodifiers may affect the subjects make a prejudgement to the polysemous words.