File Information
File: 05-lr/acl_arc_1_sum/cleansed_text/xml_by_section/intro/90/h90-1071_intro.xml
Size: 14,494 bytes
Last Modified: 2025-10-06 14:04:53
<?xml version="1.0" standalone="yes"?> <Paper uid="H90-1071"> <Title>MacWhinney, B. Competition and Lexical Categoriza-</Title> <Section position="3" start_page="0" end_page="367" type="intro"> <SectionTitle> 2.2. MIDAS </SectionTitle> <Paragraph position="0"> Previously, we have succeeded in doing some automatic lexical acquisition by exploiting conventio,al metaphors as motivations for linguistic forms. In particular, Martin (1988) implemented the MIDAS system which both uses metaphoric word senses to help with language under- null standing, and to extend the lexicon when a new metaphoric use of a word is encountered. For example, the sentence &quot;John have Mary a cold.&quot; is presumed to make recourse to a &quot;a cold is a possession&quot; metaphor. We call such a conventionalized metaphor a core metaphor, since it seems to serve as the basis for related metaphoric uses. Thus, the sentence &quot;John gave Mary a cold&quot; is presumed to involve the &quot;infecting with a cold is giving the cold&quot; metaphor, which entails the previous &quot;cold is possession&quot; metaphor.</Paragraph> <Paragraph position="1"> Suppose the system encounters an utterance like &quot;John got the flu from Mary&quot;, but is not familiar with this use of the verb &quot;get&quot;, nor with the notion of a flu being treated as a possession. Then both the available nonmetaphoric sense of &quot;'get&quot;, along with the metaphors involving diseases and possession, arc brought to bear to hypothesize the word sense that might be in play.</Paragraph> <Paragraph position="2"> Hypotheses are generated by two kinds of lexical extension processes: core extension and similarity extension. Understanding &quot;get a cold&quot; given an understanding of &quot;give a cold&quot; involves core extension, as the core metaphor &quot;cold is possession&quot; is extended to the &quot;getting&quot; concept; understanding &quot;get the flu&quot; given an understanding of &quot;get a cold&quot; involves similarity extension, as the generalization about a role in the metaphoric su'ucture must be extended from colds to diseases in general. Understanding &quot;get the flu&quot; given an understanding of &quot;give a cold&quot; involves both kinds of extension. null The MIDAS system has been used in conjunction with UC to extend metaphoric word senses in the computer domain. The following is an example of MIDAS learning a new sense of the word &quot;kill&quot;, given that it knows some metaphoric extensions of this sense outside the computer domain.</Paragraph> <Paragraph position="3"> You can kill a computer process by typing &quot;c to the shell.</Paragraph> <Paragraph position="4"> Here MIDAS first retrieves a number of metaphors related to the input; of these, &quot;Kill.Conversation&quot; is chosen as most applicable. A simple similarity extension is attempted, resulting in a proposed &quot;Terminate-Computer-Process&quot; metaphor for interlxetation of the input. The interpretation thus provided is passed along to UC, which can answer this question. Meanwhile, the metaphor is incorporated into UC's knowledge base, which allows UC's language generator to use the same terminology in encoding the answer.</Paragraph> <Paragraph position="5"> MIDAS is discussed in detail in Martin (1988).</Paragraph> <Paragraph position="6"> 3. Why MIDAS Works # How can I kill a process? No valid interpretations. Attempting to extend existing metaphor.</Paragraph> <Paragraph position="7"> Searching for related known metaphors.</Paragraph> <Paragraph position="8"> We believe that MIDAS works because it is exploiting metaphoric subregularity by a form of analogical reasoning. That is, it finds a metaphorical usage that is closest to the given case according to some conceptual metric; it then exploits the structure of the prior metaphor usage to construct an analogous one for the case at hand, and proposes this new sl~'ucture as a hypothetical word sense. Note that according to this explanation, metaphor does not play a crucial role in the extension process. Rather, it is the fact that the metaphor is a subregularity rather than the fact that it is a metaphor that makes it amenable to analogical exploitation.</Paragraph> <Paragraph position="9"> Analogy, of course, has played a prominent role in traditional linguistics. Indeed, rather influential linguists (for example, Paul (1891) and Bloomfield (1933) seemed to attribute all novel language use to analogy. However, today, analogy seems almost entirely relegated to diachronic Ixocessses. A notable exception to this trend is the work of Skousen (in press), who appears to advocate a view quite similar to our own, although the primary focus of his work is morphological.</Paragraph> <Paragraph position="10"> Analogy has also been widely studied in artificial intelligence and cognitive psychology. The work of Carbonell (1982) and Burstein (1983) is most relevant to our enterprise, as it explores the role of analogy in knowledge acquisition. Similarly, Alterman's (1985, 1988) approach to planning shares some of the same concerns.</Paragraph> <Paragraph position="11"> However, many of the details of Carbonell's and Alterman's proposals are specific to problem solving, and Burstein's work is focused on formnla:ing constraints on the relations to be considered for analogical mapping. Thus, their work does not appear to have an obvious application to our problem. Many of the differences between analogical reasoning for problem solving and language knowledge acquisition are discussed at length in Martin (1988).</Paragraph> <Paragraph position="12"> Another line of related work is the connectionist approach initiated by Rumelhart and McClelland (1987), and explicitly considered as an aiterative to acquisition by analogy by MacWhinney et al. (1989). However, there are numerous reasons we believe an explicitly analogical framework to be superior. The Rumelhart-McClelland model maintains no memory of specific cases, but only a statistical summary of them. Also, the analogy-based model can use its knowledge more flexibly, for example, to infer that a word encountered is the past tense of a known word, a task that an associationist networks could not easily be made to perform. In addition, we interpret as evidence supportive of a position like ours psycholinguistic results such as those of Cutler (1983) and Buuerworth (1983), which suggest that words are represented in their full &quot;undecomposed&quot; form, along with some sorts of relations between related words.</Paragraph> <Paragraph position="13"> 3.1. Other Kinds of Lexical Subregularities If MIDAS works by applying analogical reasoning to exploit metaphoric subregularities, then the question arises as what other kinds of lexical subregularities there might be. One set of candidates is approached in the work of Brugman (1981, 1984) and Norvig and l~koff (1987). In particular, Norvig and Lakoff (1987) offer six types of links between word senses in what they call lexical network theory. However, their theory is concerned only with senses of one word. Also, there appear to be many more links than these. Indeed, we have no reason to believe that the number of such subregularities is bounded in principle.</Paragraph> <Paragraph position="14"> We present a partial list of some of the subregularities we have encountered. The list below uses a rather informal rule format, and gives a couple of examples of words to which the rule is applicable. It is hoped that explicating a few examples below will let the reader infer the meanings of some of the others: (1) function-object-noun -> primary-activity&quot;determinerless&quot;-noun null (&quot;the bed&quot; -> &quot;in bed, go to bed&quot;; &quot;a school -> at school&quot;; &quot;my lunch -> at lunch'~ &quot;the conference -> in conference&quot;) (2) noun -> resembling-in-appearance-noun (&quot;tee&quot; -> &quot;(rose) tree&quot;; &quot;tree&quot; -> &quot;(shoe) tree&quot;); &quot;tiger&quot; -> &quot;(stuffed) tiger&quot;, &quot;pencil&quot;-> &quot;pencil (of light)&quot;) (3) noun -> having-the-same-function-noun (&quot;bed&quot; -> &quot;bed (of leaves)&quot;) (4) noun -> &quot;involve-concretion&quot;-verb (&quot;a tree&quot; -> &quot;to tree (a ca0&quot;; &quot;a knife&quot; -> &quot;to knife (someone)&quot;) (5) verb -> verb-w-role-splitting (&quot;take a book&quot; -> &quot;take a book to Mary&quot;, &quot;John shaved&quot; -> &quot;John shaved Bill&quot;) (6) verb -> profiled-component-verb (&quot;take a book&quot; -> &quot;take a book to the Cape&quot;) (7) verb-> frame-imposition-verb (&quot;take a book&quot; -> &quot;take someone to dinner&quot;, &quot;go&quot; -> &quot;go dancing&quot;) (8) activity-verb-t -> concretion-activity.verb-i (&quot;eat an apple&quot; -> &quot;eat \[a meal\]&quot;, &quot;'drink a coke&quot; -> &quot;drink \[alcohol\]&quot;, &quot;feed the dog&quot; -> &quot;the dog feeds&quot;) (9) activity-verb-t -> dobj-subj-middle-voice-verb-i (&quot;drive a car&quot; -> &quot;the car drives well' ') (10) activity-verb.i -> activity-verb+primary-category (&quot;John dreamed&quot; -> &quot;John dreamed a dream&quot;; &quot;John slept&quot; -> &quot;John slept the sleep of the innocent&quot;) (II) activity-verb-i -> do-cause-activity-verb-t(patient as subjecO (&quot;John slept&quot; -> &quot;The bed sleeps five&quot;) (12) activity-verb -> activity-of-noun (&quot;m cry&quot; -> &quot;a cry (in the wilderness)&quot;; &quot;w punch&quot; -> &quot;a punch (in the mouth)&quot;) (13) activity-verb <-> product-of-activity-noun (&quot;copy the paper&quot; <-> &quot;a copy of the paper&quot;; xerox, telegram, telegraph) (14) functional-noun -> use-function-verb (&quot;the telephone&quot; -> &quot;telephone John&quot;; machine, motorcycle, telegraph) (15) objecbnoun -> central-component-of-object (&quot;a bed&quot; -> &quot;bought a bed \[=frame with not mattress\]; &quot;an apple&quot; -> &quot;eat an apple \[=without the core\]&quot;)) Consider the first rule. This rule states that, for some noun whose core meaning is a functional object, there is another sense, also a noun, that occurs without determination, and means the tximary activity associated with the first sense. For example, the word &quot;bed&quot; has as a core meaning a functional object used for sleeping.</Paragraph> <Paragraph position="15"> However, the word can also be used in uueraw.es like &quot;go to bed&quot; and &quot;before bed&quot; (but not, say, &quot;*during bed&quot;). In these cases, the noun is detenninedess, and means something akin to sleeping. Other examples include &quot;jail&quot;, &quot;conference&quot;, &quot;school&quot; and virtually all the meal terms, e.g., &quot;lunch&quot;, &quot;tea&quot;, &quot;dinner&quot;. British English allows &quot;in hospital&quot;, while American English does not.</Paragraph> <Paragraph position="16"> The dialect difference underscores the point that this is truly a subregularity: concepts that might be expressed this way are not necessarily expressed this way. Also, we chose this example not because it in itself is a particularly important generalization about English, but precisely because it is not. That is, there appear to be many such facts of limited scope, and each of them may be useful for learning analogous cases.</Paragraph> <Paragraph position="17"> Consider also rule 4, which relates function nouns to verbs. Examples of this are &quot;tree&quot; as in &quot;The dog treed the cat&quot; and &quot;knife&quot; as in &quot;The murderer knifed his victim&quot;. The applicable rule states that the verb means some specific activity involving the core meaning of the noun. I.e., the verbs are ueated as a sort of conventionalized denominalization. Note that the activity is presumed to be specific, and that the way in which it must be &quot;concreted&quot; is assumed to be pragmatically determined. Thus, the rule can only tell us that &quot;treeing&quot; involves a Ire.e, but only our world knowledge might suggest to us that it involves cornering; similarly, the rule can tell us that &quot;knifing&quot; involves the use of a knife, but cannot tell us that it means stabbing a person, and not say, just cutting.</Paragraph> <Paragraph position="18"> As a final illuswation, consider rule 5, so-called &quot;role splitting&quot; (this is the same as Norvig and Lakoffs semantic role differentiation rink). This rule suggests that, given a verb in which two thematic roles are realized by a single complement may have another sense in which these two complements are realized separately.</Paragraph> <Paragraph position="19"> For example, in &quot;John took a book from Mary&quot;, John is both the recipient and the agent. However, in &quot;John took a book to Mary&quot;, John is only the agent, and Mary is the recipient. Thus, the sense of &quot;'take&quot; involved in the first sentence, which we suggest corresponds to a core meaning, is the basis for the sense used in the second, in which the roles coinciding in the first are realized separately. A similar prediction might be made from an imransitive verb like &quot;shave&quot;, in which agent and patient coincide, to the existence of a Iransitive verb &quot;shave&quot; in which the patient is rfsdiTe~d separately as the direct object. (Of course, the tendency of patients to get realized as direct objects in English should also help motivate this fact, and can presumably also be exploited analogically.)</Paragraph> </Section> class="xml-element"></Paper>