CONTENTS Paper Formal Semantics, Pragmatics, and Situated Meaning

January 1988

Vol. 2, No. 3

The monthly newsletter of the Center for Research in Language, University of California, San Diego, LaJolla CA 92093.(619) 534-2536;electronic mail:crl@amos.ling.ucsd.edu

CONTENTS

Paper:Formal Semantics, Pragmatics, and Situated Meaning

byAaron V.Cicourel,Department of Sociology,UCSD

EDITOR’S NOTE

This newsletter is produced and distributed by theCENTER FORRESEARCH INLANGUAGE,aresearch center at the University of California, San Diego, which unites the efforts of researchersin such disciplines as Linguistics, Psychology,Computer Science, Artificial Intelligence, Commu-nication, Sociology,and Philosophy, all of whom share an interest in language.We regularly fea-ture papers related to language and cognition (1 - 10 pages, sent via email) and welcome responsefrom friends and colleagues at UCSD as well as other institutions.Please forward correspondenceto

Teenie Matlock, Editor

Center for Research in Language, C-008University of California, San Diego 92093Telephone: (619) 534-2536Email:crl@amos.ling.ucsd.edu

SUBSCRIPTION INFORMATION

If you are currently receiving a hardcopyofthe newsletter only and have anemail address to which thenewsletter can be sent, please forward that information to CRL.

Can we send you the email version only to save printing and mailing costs?If you require a hardcopyinaddition, please request it and we will be happytosend you one.

If you require the unformatted nroffversion of the newsletter,please request it from CRL after you havereceivedthe current regular formatted version.

If you knowofothers who would be interested in receiving the newsletter,please forward the email orpostal mailing address to CRL.Thank you.

CRL NewsletterJanuary 1988Vol. 2, No. 3

BACK ISSUES

Back issues of this newsletter are available fromCRLin hard copyaswell as soft copyform.Papers featured in previous issues include the following:

The Cognitive PerspectiveRonald W.Langacker

Department of Linguistics, UCSDvol. 1, no. 3, February 1987

TowardConnectionist SemanticsGarrison W.Cottrell

Institute for Cognitive Science, UCSDvol. 1, no. 4, May 1987Dimensions of AmbiguityPeter Norvig

Computer Science, UC Berkeleyvol. 1, no. 6, July 1987

WhereisChomsky’sBottleneck?S.-Y.Kuroda

Department of Linguistics, UCSDvol. 1, no. 7, September 1987

(2nd printing of paper in no. 5, vol. 1)Tr ansitivity and the LexiconSally Rice

Department of Linguistics, UCSDvol. 2, no. 2, December 1987

__________________________________________

RECENT DISSERTATIONS

Sally Rice

Department of Linguistics, UCSD

Title:Towards a Cognitive Model of Transitivity

Recent Colloquia:William Marslen-Wilson (APU-MRC/Cambridge) presented a talkon 1/7 entitledSequential processes in the recognition of spoken words.

2

CRL NewsletterJanuary 1988Vol. 2, No. 3

Formal Semantics, Pragmatics, and Situated Meaning*

Aaron V.Cicourel

Department of Sociology,U.C.S.D.

Draft - November,1987; Not for publication nor quotation

Introduction

Recent work in linguistics seeks to link concepts derivedfrom an autonomous theory of meaning (compositionalsemantics) to notions likeintuition, psychological realities or mental models, and the speaker-listener’severyday experiences.Compositional semantics can be viewed as the product of a collective memory that represents normatively abstracted experi-ences with imagined and actual language.Amajor focus of the paper is to contrast ethnographically situated on-line dis-course in sociolinguistics with the notion of an autonomous-likecompositional semantics as an idealized yet simultaneouslyfunctional knowledge resource.The idealized aspects of formal semantics derive from a self-conscious decomposition of lex-ical, sentential/clausal meaning that permits formal scrutinyofthe compositional properties of utterances.An ethnographi-cally oriented sociolinguistic basis for compositional semantics relies on the intuitive reconstruction and practical use of thisidealized knowledge resource in locally instantiated and managed daily life situations as guided and constrained by resource-limited conditions of information processing.Notions likebackground or procedural knowledge must be viewed as empiri-cally contingent resources by speakers, listeners, and research analysts when compositional properties of utterances or dis-course are used and/or assigned meaning.In this paper,Iseek to clarify the linguist’snecessary use of intuitive socioculturalfolk models while engaged in formal semantic analysis.The folk models instantiate the linguist’stacit ethnographic knowl-edge as part of the analysis of individual utterances.

Three linguistic problems motivate the present paper.First, howdoweassess the relevance of a theory of speech actsthat examines only half of what takes place in discourse by using single utterances?Second, under what theoretical andmethodological conditions do linguists justify decisions about evidence that force them to embrace an \"archeological linguis-tic\" perspective which formally excludes data that are contingent on the circumstances of language use and the constraints oflimited capacity processing?It would be difficult to imagine archeologists confining themselves to a cave filled with artifactsif theyalso have access to nativestalking about and producing the same artifacts outside of the cave.Finally,asconsumers offormalisms, or at least of a formal notation system, sociolinguists need to knowsomething about the consequences of adopt-ing some version of available formal linguistic theories.

Formal studies of semantics in western speech areas are invariably produced by native speaking linguists or linguistsfamiliar with the language.Linguists using French, German, Polish, or Russian examples do not record native informants intheir local habitats in the sense of cross-linguistic acquisition or anthropological linguistic research with exotic languages.Nor are these formal studies likely to survey representative samples of native informants in everyday settings in order toexplore the social distribution of lexical knowledge in highly stratified societies.

Formal semantics presupposes standardized meanings that are assumed to be accessible intuitively or directly,butwhose clarification need not go beyond normative descriptions of hypothetical environments. Thereis, however, a necessaryreliance on the linguist’sand informant’snormative and intuitive knowledge and experiences of sociocultural conditions andlanguage structure and use.The linguist’sexpertise with language structure and use, therefore, depends heavily on a mixtureof dictionary definitions and tacit native (or native informant) understanding of hypothetical examples of language use.Thepragmatics of language use in mundane settings and the way newlexical items or frozen expressions emerge and are negoti-ated in emergent, locally managed interaction are not treated as problems in traditional theories of compositional semantics.Afew lexical items often used in formal semantics will be the point of departure for exploring sociocultural conditionsthat can alter the normative meanings that students of compositional semantics address with logical categories and rules ofinference. Theidea is to reviewaspects of the standardized normative meanings that enable linguists to assume that theirstudy of lexical items, phrases, and sentence-likeutterances contain elements of meaning that possess a kind of autonomousstatus in the structure of a givenlanguage.

*Delivered at the 1987 International Pragmatics Conference held at Antwerp, Belgium, 17-22 August 1987.Iamgrateful toGilles Fauconnier,Edwin Hutchins, Yuki Kuroda, and George Lakofffor their helpful critical remarks and suggestions on earlierdrafts of this paper.

3

CRL NewsletterJanuary 1988Vol. 2, No. 3

Tacit or explicit awareness of an idealized compositional semantics is analogous to speaker-listeners’ normative knowl-edge of abstract notions of civil rights, status and role expectations that are associated with institutional, organizational, andinterpersonal relations in daily life.In the pages that follow, I discuss progressively less formal approaches to semantics byreference to notions likehyponymy,mental spaces or schemata, and psychological realities.The use of these notions callsattention to culturally constituted discourse strategies that are always presupposed in work on formal semantics and recentattempts to link semantics to cognitive processing issues.Some aspects of hyponymy

In unpublished lecture notes on hyponymy and semantic opposition written in 1982 and revised in 1985, S-Y.Kurodamakes several brief but useful remarks on the role of intuition and psychological realities when linguists attempt to createappropriate definitions in formal semantics.Iwill use some of Kuroda’ssuggestions as a point of departure to exploresemantic issues associated with intuition and psychological realities by reexamining the normative and practiced aspects ofculture presupposed in the notion of hyponymy.

Borrowing from Kuroda, the notion ofhyponymycan be defined as the relation that holds between a specific or subor-dinate and general or superordinate concept.Forexample:

Def. 1 - If a conceptccontains another,C,ccan be said to be a hyponym of C.

If \"boy\" contains the concept expressed by \"male,\" then \"boy\" is a hyponym of \"male.\"\"Girl\" is not a hyponym of\"male,\" but is a hyponym of \"female.\"

If \" \" denotes words and the brackets [ ] denotes concepts, then \"boy\" is a \"symbol\" that denotes the \"concept\"expressed by it.If [\"boy\"] denotes this concept, we could eliminate the \" \" and simply write [boy].

In more general terms, [ ] can be used to represent concepts not easily expressible by an English word. Hereare threeexamples.

Ex. 1. [male] is not a hyponym of [human], nor is [female] because nonhuman animals can be male or female.Both[cat] and [dog] contain [animal] and are both hyponyms of [animal].

If an \"animal\" is viewed as a four-footed \"beast\member of the animal kingdom as contrasted with \"plant,\" then [human] is a hyponym of [animal].

Ex. 2. [father] and [mother] are hyponyms of [parent], and [brother] and [sister] are hyponyms of [sibling].

Ex. 3.Kuroda uses the terms \"boy\" and \"man\" and \"girl\" and \"woman\" to suggest a common aspect of meaningbetween \"boy\" and \"girl\" that is lacking in \"man\" and \"woman,\" yet this common aspect of meaning is not captured entirelyby \"child,\" \"young,\" \"immature,\" etc.The term [JUVENILE] seems more adequate.Hence, [boy] and [girl] are hyponymsof [JUVENILE], while [man] and [woman] are not.

All of the terms used thus far to illustrate the notion of hyponymy presume unstated aspects of a common culture andan unequivocal meaning for each term.The terms chosen to illustrate formal semantic relationships are seldom if everprob-lematic to writer and reader.Ifweare to link formal semantics to the pragmatics of language use, normative understandingsof lexical items often do not suffice. Pragmaticaspects of semantic relationships presume perceptual and interaction condi-tions associated with prior experiences or the activation of such conditions as part of the sociocultural knowledge base that isconstitutive ofthe conceptual organization of semantic memory.

When students of language engage in field research in western cultures, theyuse informants to makejudgments on howlanguage use in local settings influences and is influenced by the types of participants present, their relationship to each other,their perception and assessment of each other’sactivities, and the kind of social competence each attributes to the other(s).The standardized normative meanings assumed to be available may prove tobeinaccessible to some participants and/or per-ceivedasinappropriate for locally defined social interaction.The perception of local and complexstatus and role relation-ships can consciously and implicitly affect howthe speech event unfolds, its content, and anyconsequences that may follow.The concept [JUVENILE] is not easy to define because the notions of \"child,\" \"young,\" \"immature,\" etc., are not read-ily specifiable vis-a-vis everyday much less academic use in a compositional semantics sense.Afew minor complicationscan emerge if we try to specify a \"child’s\" age.Forexample, should we begin at age 2?Howfar should we continue aboveand belowthe age of twofor the concept \"child\" to coverwhat we hope will be its normative features? Whatshould we sayabout the upper end?Should we stop at age 10, 11, or 12?Should we reservethe age of 13 for something we can

4

CRL NewsletterJanuary 1988Vol. 2, No. 3

normatively call \"adolescence?\"

A[JUVENILE] can be someone who is 13-18 years of age for legalpurposes, but as empirical research has shown,things can become ambigous if a person is 15 years of age and is charged with a serious felonysuch as murder.Apersonaged 13 or older who is accused of committing a serious felonyposes difficulties for legalauthorities because the judgmentsthat have tobemade require semantic constructions and interpretations of such folk notions as \"community reaction,\" pastoffenses, school history,the person’sphysical appearance, their \"demeanor,\" howserious the adult authorities deem theinfraction to be, how\"remorseful\" the accused appears to adult authorities, the family’scommunity influence, the kind oflegalrepresentation the family can afford, and the way the case is presented in court.All of these latter conditions can con-ceivably apply to \"older children,\" ages 8 or 9 through 12.

The definitions of hyponymy employed by Kuroda rely on a formal use of common sense terms that help us understandintuitive aspects of compositional semantics.The linguist creates formal knowledge structures by an abstraction process inwhich her or his intuitive knowledge and sociocultural experiences in formal and local situations become implicit resourcesfor perceiving lexical items as possessing standard meanings that can be decomposed using a consistent notation system.Another example can help to illustrate some of the contingencies of meaning construction and use.Kuroda notes thatthe concept of legmay be linked to the concept of \"walking\" or that walking contains the concept of \"leg\" because of thelocomotion involved when we refer to the use of leg. Butthis viewisrejected because \"walking\" is said to contain the con-cept of \"moving,\" but not the concept of \"leg.\"

Ex. 4. [walk] contains the concept of moving; it is a hyponym of [move].

In Ex. 4, the formal term \"hyponym\" is nowassociated with what appears to be a more self-evident expression, \"con-tain.\" Theterm \"contain\" can help us to understand the idea of \"walk.\" Thegeneral point is that, with the help of examples,we assume that we understand what is meant by \"contain\" in this context. Wemust rely on our intuitive understanding ofordinary words like\"contain\" in order to construct a semantic theory based on intuition and psychological realities.Aparal-lel problem exists when the linguist uses intuitive sociocultural knowledge to parse or segment textual and discourse materi-als by reference to rules and linguistic structures.These activities do not include a self-conscious attempt on the part of theanalyst to address the interaction of intuitive and compositional knowledge employed in constructing a formal analysis.Inthe present context, we seek to clarify the intuition of the linguist about the sociocultural folk models that motivate the analy-sis produced.The folk models instantiate the linguist’sethnographic knowledge as part of the analysis of individual utter-ances.

It seems normatively acceptable to associate walking with moving, and not with leg(one can walk on their hands).Anearlier compositional viewsuch as generative semantics could allowfor the idea that the term \"walk\" can imply a substruc-ture such as moving by the use of one’slegsand feet.

What is of importance cognitively is the extent to which a reference to \"walking\" activates mental images pertaining tothe movement of legs and feet.The notion of spreading schema activation (McClelland and Rumelhart, 1981) suggests thatlexical items trigger conceptual and empirical elements not explicitly specified but presumably implied by the linguist’srefer-ence to intuition.If linguists explicitly acknowledge the necessity of integrating intuitive understanding about psychologicalreality and the formal conceptual clarification of terms like\"walking,\" \"moving,\" and \"leg,\" then sociocultural notions arepresupposed and should be clarified in order to understand the research analyst’sreasoning strategies and tacit and explicituse of cultural knowledge. Aformal strategy of analysis enables the analyst to build normative,objective structures by usingher or his intuitive knowledge. Thereader is unable, however, toexamine the claim that objective structures exist and linkthem to the analyst’staken for granted native intuition. Anabstract compositional strategy can also obscure the empiricalstudy of the processes by which humans construct and attribute meaning to language structures and use.

The union of twoconcepts can be clarified by recalling that a first concept is a hyponym of a second concept if the firstis a more specific and the second is a more general concept and if theyare of the same semantic type.Providing for the gen-eral condition of only comparing twothings of the same type, however, isnot always clear.For example, the three conceptsof [yellowbutterfly], [yellow], and [butterfly], notes Kuroda, could be viewed as being of the same semantic type, where \"yel-lowbutterfly\" and \"butterfly\" are nouns, yet \"yellow\" is an adjective and can be used as a color name or noun, though not as anoun in reference to an individual entity.Onthe other hand, the term \"male\" can also be an adjective (\"male banker\") and anoun, as in \"a male.\"

Parts of speech can guide the way decisions are made about what concepts are of the same semantic type, but are notalways helpful in creating significant semantic generalizations, notes Kuroda. Thekeytosemantic types seems to be the

5

CRL NewsletterJanuary 1988Vol. 2, No. 3

concepts expressed by typical common nouns and adjectivessaid to be attributes which can be predicated of individual enti-ties. Sothe idea of being able to say of \"butterfly\" that it can truthfully or falsely be predicated of an entity,that is, attributingacertain property to it, is the heart of the semantic type notion.Hyponymy,therefore, is usually associated with a relationthat is said to hold among particular attributive predicates.

The viewofhyponymy presented above frames the notions of \"contain\" and \"concept\" as primitive terms in a formalsense that can be part of an axiomatic system.Amore informal linguistic point of viewsees hyponymy as a psychologicaltheory about the psychological reality of a particular semantic relation that needs to be explained. Theterms \"concept\" and\"contain\" are nowused by reference to their assumed standard meaning.The psychological viewcan be seen as a \"model\" or\"standard\" interpretation of the formal theory of hyponymy.Further,within a formal perspective,statements that are basedon the theory of hyponymy could be or are taken to act likeaxioms that often become empirical claims about the psychologi-cal reality of hyponymy.

In the preceding pages, I have outlined common ground and theoretical differences within linguistics by using thenotion of hyponymy as a convenient topic for hinting at the need for a broader conceptual framework for the study of seman-tics. Mygoal has been to clarify a number of related theoretical and methodological issues without endorsing existing formalapproaches to semantics.To bepsychologically relevant, hyponymy presumes knowledge about the researcher’sand/or thenative speaker-listener’spsychological reality.When the researcher also views herself as a native speaker-listener,the empiri-cal aspects of inquiry become filtered because of the interaction between the linguist’sexpertise and use of context-freeknowledge, and the linguist’snative intuition and ability to generate and assess a variety of lexical items and utterances,including their intended use in hypothetical contexts. Whenthe \"native\"speaker-hearer is a lay member of a group, or thereare several members and several groups, as well as different settings, the empirical status of \"common understanding\" or the\"standard meaning\" of terms can become problematic.

The concern with meaning in formal linguistic semantics favors a consistent notation system, and assumes a correspon-dence with an intuitive (native speaker-listener’s) understanding of the limited notion of natural language usage as found inthe single utterances used for explicating semantic theory.The utterances used as data are always produced as part of hypo-thetical circumstances.If the native speaker-listener’sunderstanding of utterances that are produced in these hypothetical cir-cumstances are different from the understandings that arise when utterances are embedded in the larger organizational condi-tions of everyday language use, then the judgments that the linguist is looking at are not what he is taking them to be.Theutterances become the product of unusual and unexamined hypothetical circumstances.

In the above examples, a standardized, normative sense of each term is assumed to be available to the reader.Theemphasis is on the logical relations among terms presumed to be well-defined.The intuitive and psychological realities of thelexical items are dependent on the researcher’shypothetically defined world, a world where there are fewsurprises and virtu-ally free of empirical contingencies.The reader is assumed to be able to recognize the lexical items used as self-evidentrather than an interaction between everyday cognition or reasoning and prior and local cultural experiences. Intuitive knowl-edge is employed as an unexamined resource for constructing a compositional, declarative knowledge base, but the processesof construction are ignored.

As long as the concepts employed by the linguist appear to have face validity vis-a-vis a normative viewofcomposi-tional semantics, it is clearly possible to assume uniform, standard meanings exist and to employalogical format that is incorrespondence with the structural properties of target utterances.Logical forms help us clarify manyaspects of human intu-ition and psychological realities but theylack a grounding in contingencies that invariably surround information processingand environmental constraints in the construction of meaning and language use in daily life settings.Iwant to link these intu-itive understandings and hypothetical circumstances to speech events that are empirically emergent, locally managed andlarger organizational conditions of everyday,situated language use.Categories and cognitive models

Gilles Fauconnier’s(l985) notion of mental spaces is a useful modification of semantic theory because of its close affin-ity with cognitive issues associated with respresentational and comprehension issues often ignored by traditional semanticviews. Theidea of mental spaces or mental structures in Fauconnier’swork proposes that mental spaces are not part of lan-guage or its grammars, and hence do not refer to hidden levels of linguistic representation.Mental spaces help to supportmeaning but do not constitute nor represent meaning.Such spaces would presumably develop evenwhen we cannot identifyaknown language as might arise for a deaf child in a hearing home who has not been exposed to sign language.

6

CRL NewsletterJanuary 1988Vol. 2, No. 3

Mental spaces are knowledge structures in memory,but such spaces are built up by language.Language not only con-tributes to the building of relations between mental spaces and between elements within such spaces, but language also cre-ates its own constructions as well as being a source of interpretations about worlds, models, contexts, situations, and the like.If particular forms and contents of language are linked to specific occasions of use, with particular participants and eco-logical conditions, then language and the conditions in which its use is embedded become intimately attached to our ability toactivate, link, and access distributed elements of knowledge (Rumelhart, Smolensky, McClelland, and Hinton, Vol 2, Chap.14, 1986).

In his recent book on what categories reveal about the mind, George Lakoff(l987) discusses folk models of categoriza-tion and the way ordinary speakers of English employconflicting theories or cognitive models to use words to refer to thingsin the world. Paul Kay (1979; 1982) and Lakoffdiscuss a folk version of Frege’stheory of reference.

ForKay,words can fit the world either \"loosely\" or \"strictly\" according to Frege’sviewthat the way that words canrefer to objects in the world is due to the way words have intentions that can be tied to actual properties of world objects.Anexample that illustrates the notions of \"loosely\" and \"strictly\" speaking by Kay (1982: 3) can be found in the following hypo-thetical dialogue between an Anthropologist A and Layperson L.

(1) L: Wheredid the first human beings live?

A: Looselyspeaking, the first human beings livedinKenya.

Kay makes the general point that the hedge is probably used to apologize for the lack of \"quality\" in the Gricean senseof telling the truth, expecting a speaker to say no more than necessary (Quantity), and being brief (Manner).He is concernedwith pragmatic hedges that enter folk theories of language like\"loosely speaking\" and inherent meanings that can fit theworld strictly or loosely.Kay views different hedges as simultaneously incorporating world knowledge about language andas part of a speaker’slinguistic competence or knowledge of language.

Kay provides the reader with several reasons for the use of the hedge \"loosely speaking.\"This hedge, notes Kay,canreveal incoherent description while doing reference, the use of a coherent but faulty description in an act of reference, sayingsomething the speaker views as containing a false presupposition that allows an unintended interpretation to occur,and pro-ducing a sentence that lacks truth or Gricean Quality.

Adifferent folk theory of howwords refer to the world is associated with the term ’technically’. HilaryPutnam (and tosome extent Saul Kripke) are identified as the proponents of an expert version of a folk theory in which certain people in asociety,notes Kay,can be called \"experts\" because theyhav ethe authority to specify what words should designate in someexpert domain.The following example is given:

(2) Technically,that’sarodent (order \"Rodentia\").

Professional biologists are presumed to be the experts here.In the case of (2), there can be truth conditions that permitaconvergence beween \"technically\" and \"strictly speaking\" but for different reasons.The term \"Rodentia\" can be traced toscientific biology and a domain of expertise associated with the term \"technically.\" Considerthe sentence (Kay,1982: 8) in(3).

(3) Technically,street lights are health hazards.

Here the reference to street lights as health hazards can lead to questions about the authority behind the utterance.

Recent work by Fillmore and Kay (Fillmore, 1982) helps to clarify the linguist’suse of aspects of natural settings tounderstand and makepredictions about a person’scapacity to comprehend written materials and to use this comprehension tosolveproblems that are assumed to indicate future academic progress as well as general intelligence and levelofachievement.

Fillmore and Kay studied children’sability to understand a Grade Three leveltest of achievement in twoCalifornia ele-mentary schools.One goal of the study by Fillmore and Kay was the identification of the background knowledge and the

7

CRL NewsletterJanuary 1988Vol. 2, No. 3

interpreting and integrating skills needed by the students to comprehend the passages and test questions if theyare to provideanswers expected by the test creators.Subjects were interviewed in order to learn if theypossessed the knowledge and skillspresupposed in attending the passages and questions making up the test.

Fillmore and Kay found that the texts of the standardized tests examined in their study were seriously flawed andrequired considerable patience on the part of the reader because the genre of English used contained unnatural aspects of lexi-cal choice, grammatical structuring, and synonym alterations.

The study by Fillmore and Kay examines the possible ways in which the tests become educationally unexamined obsta-cles to the students and thereby can compromise the goals of the creators and users of them for practical and research pur-poses. Thetests, therefore, can incorporate cognitive and linguistic difficulties that are in conflict with the model’spredictionof howunderstanding is accomplished by the reader.

Fillmore and his colleagues devised a system of annotations that helped them create a corpus which served as selectiveinterviewprobes and a checklist with which to assess the children’swork with the texts and their production of free retellings.Interviews and probes were used to get at a metacognitive lev elofanalysis in order to identify the knowledge and skillsemployed for understanding the texts. Theresearchers sought to transform the children’stacit knowledge, and their ability tocommunicate some of this knowledge reflexively,into a more formal understandng of the knowledge and skills employed bythe children.

Fillmore discusses the use of the annotations to indicate howanideal reader might see connections, create specificexpectations, derive inferences, and ask questions that the annotations are designed to represent.This ideal reader is a con-struction of the researcher who is said to employprinciples from compositional semantics, the building of schemata, thedetection of plans and goals, and making inferences.

The reading experience requires the child to create and understand a coherent imagery of activities that can be congru-ent with the language of the text and the \"world\" suggested therein.The child’spersonal experiences and imagination areseen as levels of \"envisagement.\" Several levels of \"envisagement\" are identified in order to pinpoint the kinds of inferencesderivable from the text’slinguistic material, or the inferences that seem to be based on knowledge the child brings to the textin order to contextualize what is happening within a common framework.

The Fillmore and Kay work goes beyond single sentence semantic analysis and demonstrates the need to expand a nar-row, traditional semantic methodology if we are to understand the child’scomprehension of written texts during problemsolving.

My interest in the pragmatic work of Fauconnier,Fillmore, Kay and Lakoffstems from the clear implication that differ-ent cognitive models (the intuition and psychological realities alluded to by Kuroda and noted earlier) are activated by differ-ent pragmatic hedges.From the point of viewofthe speaker,wecan say that an emergent comprehension of an environmentof objects and social relations is shaped by a constructed instantiation of a cognitive model and the use of pragmatic hedgesto express the sense of the understanding and/or intentions of speakers and listeners.

The cognitive viewofsemantics noted earlier needs to be extended to particular,socially organized settings and locallyemergent speech events in which everyday cognition or reasoning interacts with sociocultural experiences to produce andactively reconstruct tacit and normative knowledge about the world. Thesesocially organized settings typically have associ-ated with them certain types of people with overlapping interests, social relationships, and expectations of which participantsshould be responsive byusing domain-specific and generally understood lexical items and utterances.Expert commentarypresupposes tacit and declared knowledge of specialized information, normatively shared meanings, and local socioculturalconditions. Iwill return to this point belowinabrief dicussion of medical communication and diagnostic reasoning.Yankee kinship terminology

Responding to a paper by Ward Goodenough (1965) on \"Yankee Kinship Terminology: AProblem in ComponentialAnalysis,\" David M.Schneider (1965) reexamined Goodenough’spaper in terms of his own research on American kinshipand arrivedatasomewhat different point of view. Schneider wanted to compare the same aspect of American culture fromthe perspective oftwo approaches. Schneiderobserved that there was no way to depict the domain of kinsmen in a formal,clear,categorically limited way.For example, while it can be said that all second cousins are kinsmen, he noted that it wasnot possible to say that all third cousins are kinsmen.If twopersons talk about their kinsmen, there will be uncertainty aboutthe names, ages, occupations, and places of residence for each relative aseach person calculates howdistant each relative isassumed to be.Schneider then introduced the notion of the \"Famous Relative\"inorder to underscore the way a famous

8

CRL NewsletterJanuary 1988Vol. 2, No. 3

person could be construed as a relative evenwhen the relationship was obscure or distant.The point is that although such afamous relative,when traceable, could be quite distant, closer relativescould be ignored, unknown, or unheard of (Schneider,1965: p. 290).

The general point made by Schneider is that there are variant rules in the kinship system at the boundary rather than atthe core.Modifiers (father-in-law, mother-in-law, daughter-in-law) are applied in a context of variant patterns that result in atransformation of basic terms into derivative sets which appear to be capable of infinite extension. Schneidercalls this the\"fade-out\" principle because of the fuzzy boundary associated with a term like\"distant cousin.\"

Another issue raised by Schneider is what he calls the \"principle of unbalanced, dribbling dyads\" that is associated withtwosets of kinship terms; those that are basic like\"father,\" \"mother,\" \"brother,\" \"sister,\" and those that are derivative like\"father-in-law\" and \"sister-in-law.\"The latter are compounds of a kinship term and a non-kinship term or a basic term plus amodifier.Dyads like\"mother\" and \"father\" and \"brother\" and \"sister\" cannot be found in the term \"cousin.\"The generalpoint here is that the use of the term \"distant\" (as in \"distant cousin\") refers to \"a chain of unbalanced dyads\" and these dyadsare said to be unbalanced for Schneider because they\"dribble\" or fade-out (Schneider,l965; p. 293).The chains of dyadsdiminish in value and thus fade away. The famous relative,notes Schneider,takes on exaggerated importance and stands outin terms of a fading chain of relatives.

Akey aspect of Schneider’spaper is that we often assume a close fit between an analytic domain and a semanticdomain of a particular culture.The analytic framework is necessary for having a standard and controlled way of approachingacultural setting, but it must then be modified in light of the way eliciting techniques are employed. Myconcern with theformal use of hyponyms is that the analytic apparatus is informed by reference to the linguist’suse of normative data thatinvariably is derivedfrom an intuitive understanding of the investigator’sown culture or a use of secondary sources where lit-tle or no attention is paid to eliciting techniques and the way concepts are employed in daily life settings.The concept of [mother]

Consider one of Lakoff’sexamples about kinship.The concept of mother can be made problematic because of struc-tural and practiced changes in \"family life\" in the United States.Lakoffdoes not examine family circumstances that havebeen described in empirical research, yet his useful examples call into question traditional semantic views that assume stan-dardized, normative meanings can be assigned unequivocally to lexical items.In keeping with our earlier discussion, underwhat conditions can we say that [mother] is a hyponym of [parent]?

Lakoff(1987: 74) offers a possible definition of mother from the standard semantic perspective (\"A woman who hasgivenbirth to a child\") and states that the definition is inadequate.He describes several \"models\" for the term \"mother,\"which are taken in the normal case as coinciding.Forexample:

birth model - \"a woman who has givenbirth to a child\"

genetic model - a female who has contributed genetic materialnurturance model - a female who givesnurturance to a childmarital model - a wife of a father

geneological model - the closest female ancestor

The above normative models are part of a \"cluster model\" for Lakoff. Anexample of a deviation from this model would be a\"stepmother.\"

Of interest in Lakoff’sdiscussion of \"mother,\" is the existence of several criteria for \"real\" motherhood.Several, quasi-empirical-likeutterances are provided by Lakoffthat give the reader a convincing sense of the way the term \"mother\" can berendered problematic in everyday discourse as a hyponym of \"parent.\"Forexample:

\"I was adopted and I don’tknowwho my real mother is\"\"I am not a nurturant person so I don\"t think I could everbeareal mother to anychild\"

\"I had a genetic mother who contributed the egg that was plantedin the womb of my real mother,who gav ebirth to me and raisedme\" (Lakoff, 1987: 75).

9

CRL NewsletterJanuary 1988Vol. 2, No. 3

Lakoff’sexamples illustrate the kinds of contingencies that can mitigate or weaken the normative conditions associatedwith a compositional semantics that avoids routine local discourse or exchanges in daily life settings in which structural ele-ments of a kinship term like\"mother\" can in practice express problematic conditions.

Akey issue for Lakoffisthe fact that models for \"mother\" do not pick out a single individual each time.The net resultis a number of compound expressions like\"step-mother,\" \"surrogate mother,\" \"adoptive mother,\" \"foster mother,\" \"biologicalmother,\" and so on.ForLakoff, models convergeinthe ideal case and various models for mother are related to the ideal case.But as Lakoffnotes, mother is a \"cluster model,\" or a cluster of converging cognitive models for \"mother.\"

To takeamore extreme case for the normative viewofparent, would a \"voluntary father\" of a child whose mother is aself-proclaimed lesbian be a hyponym of [parent] when the agreement was that the \"father\" was not to return to the\"mother’s\" house after the pregnancywas confirmed?The use of terms like\"mother\" and \"father\" can become complicated ifthe \"mother\" identifies herself as a lesbian (or a \"father\"identifies himself as homosexual) and seeks welfare assistance forwhich legalcredentials are required, including a statement that claimed the \"father\" (or \"mother\") had abandoned the child orwasunknown. Isthere a compound term for \"father\" here?The \"father\" (or \"mother\") is known, yet his (or her) name maynot be used in naming the child and the notion of \"father\" (or \"mother\") can be relegated to the status of an unknown sexualpartner.

Another interesting challenge to the normative viewisacase recently reported in northern California newspapers. Apreviously married couple obtained a divorce and the mother asked the court not to allowthe \"father\" to visit the childbecause the \"father\" had gone through a sexchange and had become a \"female.\"The court approvedthe mother’srequest.The normative use of terms like\"father,\" \"mother,\" \"brother,\" \"sister,\" and \"cousin,\" can be contingent on a variety ofdaily life circumstances in which there can be explicit challenges to a person’squalifications, entitlements, or the right to becalled \"mother\" or \"father\" or \"brother\" or \"sister\" or \"cousin.\"In everyday settings, there can be considerable variation inthe extent to which participants of exchanges makeuse of kinship terms.Such everyday usage tacitly or pointedly challengessomeone’sright to be called by a normative kinship term.Normative usage, therefore, can be conceptually and empiricallycontingent on practical and enforced social and personal relationships, physical propinquity,and a variation of the famous rel-ative notion. Thereare strong reasons to claim that normative usage plays a role, but such usage is seldom sufficient for prac-tical purposes.There are manynontrivial instances where what emerges locally and is enforced in daily life and legalsettingscan be at variance with normative usage.

Challenges to an idealized compositional semantics can also come from biological offsprings who may decide that a\"stepmother\" is not a \"real\" mother despite not knowing the \"real\" mother.Or, a disgruntled offspring who feels that the\"real\" father is not fit to be called \"father\". Normative terms are convenient but theycan also be misleading.Areliance onthem for understanding the way lexical items emerge in a language and takeondifferent meanings across time and situationscan be seriously deficient if we ignore the sociocultural conditions that give rise to and change or reconstruct normative andlocally emergent practiced and enforced usage.

Acase that creates newcomplications for the semantic analysis of the concept mother revolves around the notion of\"surrogate pregnancy\" for a \"test-tube\" baby.The biological parents’ sperm and egg are used to conceive a test-tube babythat is then implanted in another woman to produce a genetic off-spring of both parents.The case can arise when the geneticmother is fertile but possesses a diseased uterus.The woman who givesbirth to the child is not the child’sbiological norgenetic mother and presumably could be required to release the baby at birth to the biological-genetic mother and father.Thewoman carrying the baby would presumably have tosuppress or adapt to anythoughts and feelings about being a \"mother\"that could result from experiencing the pregnancyand the delivery of the baby.These latter experiences are often consideredto be especially important for developing a self-conception of being a \"mother\" in western culture.The woman who agrees toasurrogate pregnancyisa\"mother\" in a cultural and physiological sense, and may feel likea\"real mother,\" especially if thepregnancycreates emotionally-charged, difficult, uncomfortable, and pleasant experiences.

Tw orecent cases of a surrogate mother in South Africa and France (reported to me by Gilles Fauconnier,personal com-munication) is of interest because of the way theycompounds kinship notions.The French case involves a mother being asurrogate mother for her daughter’sthree test-tube babies (\"bebes-eprouvettes\"). TheFrench press called the surrogatemother \"la premiere grand-mere eprouvette.\" Inthis case, the grandmother is also a mother of sorts to her \"grandchildren.\"An understanding of this complicated case presumes more than standard normative notions about the concept of mother.The concept of [physician]

10

CRL NewsletterJanuary 1988Vol. 2, No. 3

Afinal example of complexhyponymy can be found with the term [physician]. Inthe U.S., the term House Staff(interns, residents, training fellows), and specialist and subspecialist attendings such as physicians who have completed aperiod of residencyand perhaps more training in a particular sub-specialty (neurosurgery,urology,cardiology,orinfectiousdiseases) can all be hyponyms of [physician]. Thesedifferent types of physicians also represent variations in technical exper-tise within a family of experts. Thepronouncements of these experts can takeondifferent meanings depending on who is therecipient of their utterances, and the listener’sknowledge of medical education and the responsibilities and duties of medicalpersonnel within teaching hospitals in the United States.

When a person seeks medical care at a variety of health care delivery settings, the differences between the above cate-gories of physician (intern, residents, training fellows, and different attendings) may not be evident. Legally,all of the above\"physicians\" may be qualified to attend to aspects of the patient’sproblems, but in practice the medical care that is likely tobe offered will be contingent on the bureaucratic and ethnographic conditions that prevail in the medical setting at the timethe patient appears asking for help.Different physical locations can be expected to be organized around the presence of cer-tain types of health care delivery personnel according to the organizational constraints of the medical bureaucracyand thekinds of routinized and improvised activities that might occur.

The rather large number of terms that can be hyponyms for [physician] presupposes highly complicated socioculturalsettings and knowledge about the interaction between normative and improvised emergent encounters between patients andhealth care delivery personnel.When the term \"doctor\" is used in medical and non-medical settings, there can be consider-able variation in what is implied or taken for granted vis-a-vis the particular types of knowledge a researcher attributes tospeakers and listeners and their social distribution across and within settings.These knowledge attributions must be viewedas empirical issues.

Abrief example from the beginning of an actual initial interviewbetween a Training Fellow(TF) and patient (Cicourel,1986) can illustrate and clarify empirical aspects of the notion of background knowledge often attributed to native speakers orparticipants of discourse.

(1) TF: Ummm,who sent you to arthritis?(2) P:Uh, uh, oncology.

(3) TF: Oncology,(unclear) that’sokay,(other voice) now(4) letme just get a piece of paper (7 seconds)(5) (closingdrawers).(6) Howold are you?(7) P:44..

(8) TF: Okay(9 seconds) and (do you?) have any problems?(9) P:Oooooh, the whole body.(10) TF: Wholebody.(11) P:Joints, really bad.(12) TF: Uhuh,yeah, okay.

(13) P:And ummm, breakout in these big red spots, (mumbling)(14) topsand toes.(15) TF: Uhummm

(16) P:But only when I sit in the hot water,theycome out(17) quiteabit, my hands get, likethis, theystiffen(18) up.

The above exchange took place in a rheumatology clinic of a university hospital.The TF introduced herself as \"Dr.X.\"She was wearing a white lab coat.The terms \"arthritis\" and \"oncology\" as used by physician and patient presuppose implicitmetonymical common understanding.Each assumes or takes for granted that the other is familiar with such metanymicusage to refer to physical locations and medical personnel associated with arthritis and oncology clinics.The terms \"oncol-ogy\" and \"arthritis\" presuppose a knowledge of social and medical categories that are not clarified in the discourse but suchknowledge is necessary for appropriate metonymic communication.

In the above exchange, nothing is said by either participant about the qualifications of the physician, the patient’sknowledge of the terms employed, nor the grounds for the patient being sent to \"arthritis\" by \"oncology.\" TheTF did not

11

CRL NewsletterJanuary 1988Vol. 2, No. 3

reveal to the patient that she was being supervised by a rheumatology attending but simply proceeded with the interviewasifshe was engaged in normal behavior.Although patients often wonder about the competence of their physicians, theyare sel-dom informed enough to ask much less comprehend appropriate questions and answers.The TF’sexpertise or lack of it didnot appear to be a problematic aspect of the patient’slocal instantiation of a folk model of medicine.

Anumber of interpersonal and local organizational constraints impinge on the way physicians explain or justify theirproposed and actual activities. For example, the number of patients scheduled for the morning or afternoon, the complexityof the symptoms, and hence the amount of time that will be available for each history and physical examination are condi-tions the patient is not likely to know. The initial part of the interviewproduced a number of problems vis-a-vis the interpre-tation of the patient’ssymptoms and the lack of adequate probes on the part of the TF in order to clarify the reference to\"Joints, really bad,\" \"big red spots,\" and \"hands get, likethis, theystiffen up.\"The complaints about the joints, red spots, andhands that stiffen are of immediate possible significance for a rheumatologist.What is of interest linguistically are the manydeictic and anaphoric functions and referents like\"tops and toes\" that are taken for granted by the physician and patient.Telling the reader that the participants of the exchange share a common knowledge base and language is not very informativein the present case unless we can also showthat the ambiguities that emerge in the discourse create problems that are oftenignored in everyday communication.Ambiguous deictic or anaphoric referents in everyday speech events are routine andapparently essential devices for sustaining what otherwise could turn into a tedious and boring exchange in which everyunclear term would have tobeclarified. Theuse of indexicals in everyday speech events appears to be necessary for creatingand attributing coherencytothe event as a believable event despite violations of idealized Gricean maxims.

The patient’suse of terms like\"whole body,\" \"Joints, really bad,\" \"these big red spots\comeout,\" \"my hands get, likethis, theystiffen up,\" is common in medical settings and all discourse.The TF responded in waysthat are characteristic of daily life speech events in which much is taken for granted and where elements of polite discourseprevail that are locally managed.From the perspective ofanonmedical observer (the present writer) and reader of this paper,the patient’sremarks appear as intelligable or meaningful.From the standpoint of everyday discourse, there is no reason forparticipants to question the utterances produced.In legaland medical practice, scientific research, and a range of economicand political activities, the use of indexicals can lead to requests for clarification and confrontation.In the present case, theattending physician, after examining the transcript of the interviewbetween the TF and patient and listening to the audio tape,stated that the TF should have pursued the patient’sremarks with careful probes about joint pain, the location of the pain, not-ing anytenderness, disfiguring, and swollenness in each area mentioned.Similar probes should have been invokedfor thereference to red spots and the hands.In addition, the attending physician noted that the patient’ssymptoms were not consis-tent with what is formally known about arthritic patients.

The patient was not told that after the interviewthe TF would consult with the attending in a separate room and thenthe twoofthem would visit the patient again. Norwould the patient be told that the TF’sexpertise would be subjected toscrutinyduring the TF-attending discussion of the case in the other room.The patient would also remain uninformed of thefact that both the TF and the attending misdiagnosed the medical problem but for quite different reasons.The patient is alsounlikely to knowthat the TF’sexpertise was subjected to further scrutinyafter the attending accompanied the TF for theirjoint visit to the clinic examining room.The TF’sreturn to the examining room after the second consultation with the attend-ing makes it appear that the TF is the expert in charge of the patient’scare for she will prescribe medications deemed neces-sary,laboratory test or x-rays needed, and indicate if and when the patient is to return for a subsequent visit.

When the brief excerpt from a medical interviewbetween a training fellowinrheumatology and a patient in a hospitalclinic is compared to the semantic analysis described earlier in the paper,anumber of contrasts appear.Inorder to under-stand what appears to be routine talk between a physician and patient, a considerable amount of background informationmust be presented.Normally such background information is imagined by the research analyst by her or his use of personalknowledge of the world. Insome cases, the background knowledge has been abstracted by someone engaged in fieldresearch. Inthe medical case described herein, the background knowledge attributed to the physician by the patient is takenfor granted or not made problematic.The TF’straining and clinical experience is not known to the patient.The TF’stalk andthe inferences she made were evaluated by the attending physician, and some kind of assessment is a normal part of the insti-tutional setting within which the exchange occurred.Manylexical items were used in an ambiguous or misleading manner,butsuch language use is normally not scrutinized by the attending physician. Theassessment made by the attending physi-cian usually includes a contrast between the account giventoher or him by the resident or training fellowand her or his ownindependent examination of the patient.

12

CRL NewsletterJanuary 1988Vol. 2, No. 3

Within a theory of compositional semantics, no attempt is made to examine differential background knowledge thatspeakers and listeners bring to their comprehension of utterances.Cognitive grammar approaches to language competenceand performance address background knowledge and can also specify ethnographic conditions that appear to be linked to lan-guage use and comprehension, and the participants folk or mental models that seem to be relevant. Inthe case of the work onthe ideal reader by Fillmore and Kay,theyexamine an informant’slinguistic performance during a comprehension task.Inthis paper,Ihav ealso added a number of additional contingencies that normally surround language use during decision mak-ing and the inherent comprehension that must prevail. For example, institutional conditions that affect the way people areprocessed organizationally,the asymmetry that exists between the physician and patient vis-a-vis their knowledge ofmedicine, and the asymmetry that can exist between physician as novice in a speciality or subspecialty and across such spe-cialties. Finally,there is the potential during speech events for participants to negotiate lexical and more complexlev els ofmeaning. Ourdiscussion can be pursued further by examining an exchange between the twophysicians when theydiscussedthe case.The exchange beganasfollows:

(1) TF: Ok,next is Elena Louis, (background voices(2) anyway,she’s44years of age and sent here(3) from(the?) oncology group.(4) Sothe past twoyears she has had episodes(5) initiallyof erythema followed by swelling(6) involving the second and third metacarpal(7) andPIP joints of both hands, alternating,(8) onetime this hand, one time this hand.(9) She’salso had arthritis of her ankles,

(10) whichincludes redness on a lateral border(11) ofthe lateral malleolus followed by(12) swelling.

Having already informed the reader about the TF and her limitations as a novice in the area of rheumatological dis-eases, the above passage nevertheless appears as a crisp, knowledgeable statement by an expert. Theredo not appear to beanyobvious constraints on the way the TF conveysher findings to the attending here.My observation of the setting sug-gested a relaxed atmosphere despite the formal mode of presentation that is part of a familiar bureaucratic routine for a teach-ing hospital.But my ethnographic experiences enables me to say that the TF is being evaluated by the attending and theattending’sattempt to sustain an informal exchange leads me to viewthe setting as somewhat less than relaxed. Otherreadersof the above narrative bythe TF have noted that the contents seem to makesense and were curious about the attending’ssub-sequent negative remarks about the TF’sperformance. Atthe end of the TF’slong narrative (of which only a fragment isshown here), the attending expressed some concern about the TF’stentative diagnosis of \"degenerative joint disease.\"Thesubsequent examination by the attending provedinconclusive.The patient was to have returned for x-rays and further tests.The attending’scritical remarks were expressed to me after the clinic session was overbut their depth did not materializeuntil after the attending had listened to the original TF-patient interviewand the TF-attending exchange partially quotedabove.

An examination of the TF’snarrative and the attending’sresponse (not shown here) from a compositional semanticsperspective should not in principle pose anyparticular problems.The technical terms, for example, should be amenable toeasy comprehension by reference to manytexts and medical dictionaries available. Theuse of the term \"oncology group\" inline 3, however, continues the earlier metonymical expressions of the patient and TF employed at the outset of the originalmedical interview. The metonymical phrase \"oncology group\" is not readily interpreted by reference to a compositionalsemantics domain.An additional problem for a compositional perspective can be found in line (8) where the TF states thatthe redness and swelling (line 5) of the lower joints of the fingers (lines 6 and 7) alternate as \"one time this hand, one timethis hand.\"Observational knowledge of the local setting is needed to confirm what could be inferred as the TF extended eachof her hands to the attending to illustrate her remark.Acloser look at line 9 reveals a reference to \"arthritis of her ankles.\"Decomposing this utterance poses a number of difficulties for the research analyst because the textbook and dictionarydescriptions and definitions of \"arthritis of her ankles\" create a number of problems of interpretation.Indeed, the attendingchallenged the TF’sobservation here as being diagnostically ambiguous.

13

CRL NewsletterJanuary 1988Vol. 2, No. 3

The research analyst is obliged to consult more than reference texts and dictionaries in order to comprehend andexplain the above narrative toareader.Knowledge about medical postgraduate education, institutionalized hospital practices,declarative and procedural basic science and clinical knowledge about medicine and rheumatology,local pragmatic languageusage, and organizational and interactional practices are presupposed in order to comprehend the TF’snarrative.

When the attending and I reviewed the transcript line by line and the tape, the attending was critical of the questionsused by the TF and directly challenged the validity of her claims about arthritis of the ankles (line 9) and the entire diagnosis(Cicourel, 1986).Specific lexical items were challenged by the attending.Although the formal meaning of such terms as\"erthema,\" \"third metacarpel and PIP joints of both hands,\" \"arthritis of her ankles,\" \"redness on a lateral border of the lateralmalleolus,\" seemed to be self-evident to the attending at the time of the exchange, the attending stated that such terms werenot linked to physical conditions such as tenderness and indications of joint malformation.Nor did the TF clarify what forthe attending were ambiguities in claiming a tentative diagnosis of degenerative joint disease (osteoarthritis)when there werespecific joints on the hand and problems with the ankles that could associated with rheumatoid arthritis.Manyofthesedetails are specified elsewhere (Cicourel, 1986).Ihav eunderscored the kinds of background knowledge required to claimthat there was a detailed, crisp, and informative narrative delivered by the TF.

The reader and patient would have noway of knowing that the initial interviewelicited information that was challengedby the attending, and that the subsequent oral history constructed by the TF for the attending had altered the often ambiguousmaterial obtained during the initial interviewwith the patient.The attending, of course, was unaware of the original inter-viewmaterial at the time the TF reported her findings about the patient.There were a number of complicated communica-tional issues that emerged in this case that I cannot pursue here.

Moving awaysomewhat from the hyponymy notion, I want to underscore the fact that the truth or falsity of thereported speech presented above isdifficult to discern, and anydirect attempt to pursue a formal semantic analysis of theexchange would be contingent on politeness conditions of House Staff-attending interaction, the attributions of knowledgethat are tacitly made during the exchange, and the institutional expectations and constraints that are always present.

Asecond case involves a medical resident (MR) and a patient suspected of having an infectious disease and possibleadditional complications associated with the infection.The initial interviewbythe MR was preceded by an extensive exami-nation of the patient’schart in my presence.The interviewbeg anasfollows:

(19) MR: Hi! I’mDr.X.

(20) P:(?)

(21) MR: That’sokay.This is [researcher] (C:Hello.).(22) I’mthe infectious disease doctor.(23) P:Oh, are you!(24) MR: Iwork with Dr.Y,she’smyboss.(25) P:I’ve heard about you.Okay.

(26) MR: Itlooks likeyour doctors are doing a real good(27) jobtreating that infection with penicillin.

The above interviewtook place in a medical ward of a university hospital.The MR and researcher were both wearing

white lab coats when the ward was entered and the interviewoccurred. Thenurses and aids on the ward did not question ourentrance nor our immediate attention giventothe patient.

The various uses of the term \"doctor\" (lines 19, 22, 24, and 26) by the MR does not tell the patient about his residentstatus and the fact that his \"boss\" of line 24 is the infectious disease attending who is responsible for supervising the MR’sdiagnostic and treatment activities. Thereference to \"doctors’ in line 26 is not explained to the patient.There had been twointernal medicine specialists who already interviewed and treated the patient after she was admitted to an intensive care unitand the ward noted above.Thus, at least three different attendings (each with a different subspecialty) had seen the patientand were involved in the treatment of her infection.The \"real\" expert here, however, was the MR’s\"boss,\" the infectious dis-ease attending.From my discussions with the patient, she was unaware of anyofthe distinctions I have noted above.Herfolk theory of medicine as instantiated during our discussions did not seem to include knowledge of the background of thedifferent physicians that had examined and treated her.

14

CRL NewsletterJanuary 1988Vol. 2, No. 3

Asubsequent speech event (Cicourel, 1987a; 1987b) between the pathologist in charge of the microbiology laboratory,the infectious disease attending mentioned earlier,the MR, and a fewothers who routinely attend the 11:00 A.M. lab roundsshould be noted here.The lab rounds are occasions at which each patient’scultures for the past 24 hours are examined formicroorganisms as part of the normal business and teaching activities of the microbiology laboratory.The three physicianswho discussed the case (described above)atthe lab rounds did not identify themselves to the others present (fourth year med-ical students, pediatric residents, and a pathology resident) as physicians with different kinds of expertise. Norcould medi-cally informed experts knowwhich physician could be said to be the expert in the following conversation unless theyweretold the titles of each speaker.

(1) PA: (?)(lowvoice level) Isthis the same one (we?)(2) (ya?)did yesterday?

(3) IDA: No. Thisis the eye lady.(4) PA: (?)(5) IDA: Cellulitis(6) PA: Oh.

(7) IDA: With group A strep..in shock

(8) PA: Inshock. (Slightrise in voice level) Howabout(9) that.

(10) IDA: I[t?]wasgonna be more interestingif she didn’t(11) MR:I’m (?)

(12) IDA: have bacteremia but (laughing and voice level

increasing) nowshe’shad bacteremiaso(13) MR:There’salittle, there’s

little (voice levelincreases) problem with thatthat I’ll, will go into more as far

When the reader is told that my observations of the microbiology laboratory revealed that the pathologist was the per-son with the authority to makefinal pronouncements about the nature of the organisms discerned from examining the culturesfor each patient, the speech event noted in lines 1-13 can be analyzed more carefully vis-a-vis their semantic content.Tellingthe reader that PAisthe pathologist, that IDAisthe infectious disease attending, and that MR is the infectious disease medi-cal resident adds information that goes considerably beyond simply reporting a conversation between three \"physicians.\"Unless we knowsomething about the identity of the speakers and the kinds of expertise theyare presumed to possess vis-a-vis the patient described in lines 19-27 above,itwould be difficult to discuss the terms employed in a coherent manner.Norwould we be able to attribute differential significance to what each \"physician\" said unless we knewsomething about theethnographic and organizational bureaucracyofmedical teaching hospitals.Knowing something about the bureaucraticorganization, the training that medical students and house staffreceive,the kinds of medical specialties that can be found inmedical organizations, makes it possible to engage in a semantic analysis of discourse that is impossible on the basis of thefolk knowledge of content alone.Amodel of formal semantics does not tell us what the words in lines 19-27 mean withoutthe context described above.

The twocases of medical discourse described briefly above underscore the role of the speaker or listener’sfolk modelof language, world knowledge, and the role of local knowledge conditions associated with or that emerge in the immediateenvironment overthe course of a speech event. TheMR and TF are \"physicians\" when theyare speaking with a patient, butare also novices when speaking with an attending.There are serious legalconsequences attached to the novice status ofHouse Staffand the extent to which theycan express their expertise. Similarconditions can be described among differentspecialists or sub-specialties within the practice of medicine.Conclusion

The tacit reliance on one’snative intuition to create formal definitions while seeking to satisfy conditions of perceivedpsychological reality attributed to speaker-listeners implies that a test of the claims being made (and independent of conven-tions adopted) can be found in the practices of all researchers.Research analysts are presumed to use their own native intu-ition in a similar manner to agree to or propose related or different interpretations perhaps based on different theoretical

15

CRL NewsletterJanuary 1988Vol. 2, No. 3

commitments and sociocultural experiences. Theresearcher’soften tacit ethnographic experiences, therefore, become a cen-tral resource in constructing utterances.In the materials presented above from medical settings, the researcher and partici-pants also must know, atleast tacitly,about hyponymy,metaphor,and metonymy in order to perform as competent membersof the speech events cited.Forasocial scientist without training or knowledge of linguistics, these concepts (hyponymy,etc.)must be raised to a self-conscious leveland learned explicitly as part of her or his research tools.

The notion of psychological reality implied here is consistent with a parallel distributed processing viewofschema the-ory or folk theory of mind.Manyresearchers working in cognitive science note that persons’ capacities for processing infor-mation occur under conditions of the spreading activation of the possible significance of lexical items or phrases for compre-hending utterances (McClelland and Rumelhart, 1981).Atheory of compositional semantics, therefore, becomes a concep-tual resource that is an integral part of the way humans are said to structure their perception of aspects of an object or eventand provide humans with the basis for constructing the meaning or interpretive sense of their experiences.

Environmental experiences activate our memories while the latter simultaneously guide and are influenced by what isperceivedand processed as instantiated schemata.The interactive nature of schemata, according to McClelland and Rumel-hart (1981), tends to be automatic as the environment produces newexperiences or data.From a sociocultural point of view,however, the researcher and speaker or listener’suse of compositional or combinatorial semantics is contingent on the wayknowledge is socially distributed in different cultures, within groups, and socially organized activities. Acknowledging therole of instantiated schematized knowledge requires the linguist to recognize changing inputs and the emergent reconstructionof folk models or a changing sense of what is happening in a local setting during language use.But linguistic theory and itsuse of the notion of psychological reality invariably posit primarily normative,idealized descriptive environments and are fur-ther limited by an obscure modeling of individual cognitive processing while ignoring the collective discourse or conversa-tions during which language acquisition and use occurs.

Discussions of hyponymy presuppose a system of classes or categories and rules and/or conventions for various enti-ties. Areference to classes or categories can include the researcher’scareful definition of entities that makeupaclass or cat-egory,but seldom includes the native’s folk understanding of the psycho-sociocultural reality of a class or category of entitiesor objects in locally emergent settings.Psycho-sociocultural reality in cognitive science, therefore, is based on theresearcher’stacit use of intuitive knowledge in the normative construction or reconstruction and decomposition of semanticdomains.

The formal semantic description of hyponymy is a reminder of the way intuitively informed meanings become an inte-gral part of a normative linguistic environment of objects whose compositional elements can be specified.The self-evidentuse of concepts like[boy], [girl], [male], [parent], and [mother], however, can obscure the role of folk knowledge as recon-structed by informants and researcher within the latter’scompositional viewofsemantics.

We hav elittle idea of lay-persons’ distribution of knowledge by sociocultural groups and within and between differentbureaucratic organizations. Thereis little information on the ability of individuals to recognize and compare concepts andlexical items as a function of their perception of and participation in locally emergent discourse and the extent to which par-ticipants engage in the kinds of componential decomposition and comprehension attributed to them.

Atheory of compositional semantics presupposes considerable familiarity with a language.Prior training or expertiseis required to enable students of language to transform their own educated and intuitive knowledge of semantic meaning intoaformal or objective format. Theutterances used as data, however, inv ariably lack the ambiguity and negotiated circum-stances of daily life use, comprehension, and decision making.Classes and categories are often not only ambiguous in dailylife exchanges, but attributions of meaning are also responsive toand embedded in interpersonal and organizational relation-ships and the constraints of local production.

Science, lawand medicine are examples of domains where a compositional semantics is part of a self-conscious effortby the practitioner to cleanse concepts and data of the cognitive limitations of individuals and groups in institutionalized,organizational, and locally managed settings.The intersection of these essential semantic domains with pragmatic languageuse in actual settings can help clarify the larger conceptual framework in which both formal and practiced and enforcedmeanings are negotiated.

Tw omajor issues of semantic analysis are the role of presuppositions and background knowledge. Theseissues canoverlap but in general presuppositions are logical entailments that can be inferred or deduced from the semantic facts beforeus. Tosay that \"John drove through the red light\" entails assuming some kind of vehicle (auto, bus, motorcycle, bicycle, car-riage) was involved and background knowledge about the meaning of \"the red light\" as an electronic device for regulating the

16

CRL NewsletterJanuary 1988Vol. 2, No. 3

flowoftraffic. Backgroundknowledge about the existence of streets and intersections and different types of vehicles and theskills that are presupposed in their use can also be presented.

The analysis of single utterances always assumes background knowledge and implies and makes use of some kind ofimplicit sociocultural theory that motivates or guides the interpretations constructed.Each reference to presupposition orbackground knowledge assumes a sociocultural world as tacit or normatively idealized states of affairs.

Students of language become \"archeological linguists\" when theyignore the natural history of the emergence and useof speech events. Formal criteria, and a narrow, compositional viewofsemantics, can obscure claims about meaning and canbecome obstacles to an understanding of language perception, comprehension and use.

Aresearcher’sability to identify presuppositions and her or his capacity to use background knowledge becomes a self-evident resource that motivates the formal analysis of hypothetical utterances.The researcher’sknowledge base plays a rolehere. Howdoweassess the researcher’sclaims about the informants’ use of presuppositions and background knowledge?Unless we examine these issues, the researcher’sidentification and use of presuppositions (the investigator’sknowledge base)is tacitly assumed to be what the speaker or listener can comprehend and employinthe utterances used for the analysis ofmeaning. Inthe present paper,Ihav eused materials from a complexsemantic domain to underscore the researcher’slimita-tions and the necessity of a compositional viewofmeaning that is linked to institutional, organizational and locally managedsettings.

17

CRL NewsletterJanuary 1988Vol. 2, No. 3

REFERENCES

Cicourel, Aaron V.1986. \"Thereproduction of objective knowledge: Common sense reasoning in medical deci-sion making,\" in G. Bohme and N. Stehr (eds.),The KnowledgeSociety.D.Reidel Publishing Company, pp.87-122.

Cicourel, Aaron V.1987a. \"Cognitive and organizational aspects of medical diagnostic reasoning.\"DiscourseProcesses,10, 347-367.

Cicourel, Aaron V.1987b.\"The interpenetration of communicative contexts: Examplesfrom medical encoun-ters,\"Social Psychology Quarterly,50(2), 217-226.

Fauconnier,Gilles. 1985.Mental Spaces.Cambridge, Mass.: MIT Press.

Fillmore, Charles J.1982. \"Idealreaders and real readers,\" inProceedings of the 32nd Georgetown Roundtableon Languagesand Linguistics,D.Tannen (ed.).Washington, D.C.:Georgetown University Press, 248-270.Goodenough, Ward. 1965. \"Yankee kinship terminology:Aproblem in componential analysis,\"AmericanAnthropologist,Vol. 67, No.5, October.(Special publication on Formal Semantic Analysis, edited by E.A.Hammel.)

Kay,Paul. 1979.The role of cognitive schemata in wordmeaning: Hedges revisited.Ms. 59pp.

Kay,Paul. December1982.Linguistic Competence and Folk Theories of Language:Two English Hedges.BerkeleyCognitive Science Report No. 3.

Lakoff, George. 1987.Women, Fire, and Dangerous Things:What Categories Reveal About the Mind.Univer-sity of Chicago Press.

McClelland, J.L., and Rumelhart, D.E.1981. \"Aninteractive activation model of context effects in letter per-ception: Part 1.An account of basic finding.\"Psychological Review,1981,88,pp. 375-407.

Rumelhart, D.E., et al.1986. \"Schemataand sequential thought processes in PDP models,\" in J.L. McClelland,D.E. Rumelhart, and the PDP research group,Parallel Distributed Processing: Explorations in the Microstruc-tures of Cognition Volume 2: Psychological and Biological Models.Cambridge, Mass.:MIT Press.

Schneider,D.M. 1965. \"Americankin terms and terms for kinsmen: A critique of Goodenough’scomponentialanalysis of Yankee kinship terminology,\"American Anthropologist,Vol. 67, No. 5.(Special publication on For-mal Semantic Analysis, edited by E.A.Hammel.)

18