DOCUMENT RESUME ED 341 291 FL 020 061 AUTHOR Tirkkonen-Condit, Sonja TITLE Relational Propositions in Text Comprehension Processes. PUB DATE 91 9p.; In: Communication and Discourse across Cultures NOTE and Languages. AFinLa Yearbook 1991; see FL 020 041. PUB TYPE Speeches/Conference Papers (150) -- Reports - Research/Technical (143) EDRS PRICE MF01/PC01 Plus Postage. DESCRIPTORS Foreign Countries; Grammar; *Reading Comprehension; Rhetoric; Semantics; *Structural Analysis (Linguistics) *Propositional Analysis IDENTIFIERS ABSTRACT This study is based on the assumption that texts are composed of two kinds of propositions: lexical and relational. Lexical propositions account for semantic relations within a clause, and they can be described as semantic role relations between a lexical predicate and its arguments. Relational propositions account for functional relations among clauses, sentences, and passages of any size. The terms "rhetorical" and "relational prol.osition" are used synonymously. Lexical propositions are in the text and are lexically and grammatically signalled, whereas relational or rhetorical propositions are not unambiguously signalled. The question of how rhetorical relations are understood by readers of texts is addressed, and two empirical studies are reported briefly that seem to support the following hypotheses: (1) at least in some text comprehension tasks, relationship propositions seem to be identified with some degree of consensus; and (2) comprehension of relational propositions may figure more consciously and explicitly in some demanding text comprehens.lon tasks such as translation. It is concluded tnat more research is needed to determine whether it is realistic to assume that a battery of rhetorical relations can be delineated so accurately that readers can be taught to identify them with a great degree of consensus. (LB) ********************* ***** ********************************************* Reproductions supplied by EDRS are the best that can be made from the original document. ********************************************************************** RELATIONAL PROPOSITIONS IN TEXT COMPREHENSION PROCESSES Sonja Tirkkonen-Condit Savonlinna School of Translation Studio Introduction This study is based on the assumption that texts are composed of two kinds of propositions: and propositions lexical relational propositions. Lexical propositions account for semantic relations within a clause, and they can be described, for instance, as semantic role relations between a lexical predicate and its arguments, i5 in case grammar. Relational propositions account for functional relations among clauses, sentences, and passages of any size. The batteries of types of rhetorical relations vary, and relations such as Specify, Rackground, Justify, Evidence, and Summary are examples of relation types that ha% e been identified. These relations are also called rhetorical relations, and the terms rhetorical and relational proposition are used synonymously. Rhetorical structure theory was outlined in 1975 by Joseph Grimes, and it has been further developed in the 1980's by, eg., William Mann and Sandra Thompson. Lexical propositions are in the text, ie. they are lexically and grammatically signalled, whereas relational or rhetorical propositions are not unambiguously signalled. Their comprehension hinges largely on the inferences made by readers. will be concerned with the question of how rhetorical relations (relational ails& propositions) are understood by readers of texts. It might also be necessary to ask if it is justified to assume that the comprehension of relational propositions plays any role at all in text comprehension. Even if it may be agreed that the!" comprehension plays some role in text comprehension, it seems necessary to find out to what extent readers identify the same relations in a given text. will briefly report on the results of two empirical studies which seem to I J support two hypotheses: (1) at least in somc text comprehension tasks relational propositions seem to be identified with some degree of consensus, and (2) the 11- "PERMISSION TO REPRODUCE THIS 2 U.S. D(PARTMENT OE EDUCATFON Oucatonal Raftearct, in4 Improyanwl Offros 01 E MATERIAL HAS BEEN GRANTED SY EDUCATIONAL RESOURCES INFORMATION CENTER (ERIC) !fp Od UCOCi Si Oman drhIs docurnant rFil I cgganitst,0n raorsivaod 1rom Ma parson or onvnating C Minor ohanoas have Wan mad* lo imProvis COPY AVAILABLE twociuccon ausidy of view oi opinions 'talc! ,n th.s doct, Pants TO THE EDUCATIONAL RESOURCES I opiostrit OnrotaI do not Nroessartiy mant INFORMATION CENTER (ERIC)." poUcy OEM poedx.. 240 comprehension of relational propositions may figure more consciously and explicitly in some demanding text comprehen,ion tasks such as translation. Results of a text analysis experiment The experiment consisted of a text analysis task administered to a group of five subjects, who were to describe the rhetoriad structure of an editorial using the concepts defined in Mann and Thompson (1988). The experiment took place at Savonlinna in the context of a seminar in which this group participated. My contribution to the seminar was a lecture on Text and translation, accompanied by a group-work session. My lecture was focussed on rhetoricai cture theory, and I Introduced the idea of text analysis in terms of relational propositions. The relation types introduced and demonstrated in some detail were Background, Elaboration, Contrast, Antithesis, Concession, Evidence, justify, Emluation. Solutionhood, and Summary, and those introduced only cursorily were Condition, Purpose, Motivation, Enablemerq, Cause, Otherwise, and Restatement. These relation names derive from Mann and Thompson's 1988 article. The main purpose of the group-work session was to check whether the participants had learned the system of analysis to such an extent that they could apply it to new texts. There were two text analysis groups. They gave me permission to tape their discussions. When I listened to the tapes, it turned out that only one of the tapes was worth transcribing. There was very little talk on the other tape, and what was there was virtually inaudible. The first group's discussion, on the other hand, was lively and articulate, and therefore easy to transcribe. Thus my data originate from this "better recording and its transcription. The text which I had chosen for the subjects to analyze was an editorial of The New York Times of 30th December, 1988, The text is as follows, with sentence numbers added. 241 AGELESS, AND DRESSED LIKE AN ATHLETE (1) One day this week an elderly New Yorker was seen running for bus. (2) Running like s dere (3) How come this woman was so a fleet of foot? (4) Because said feet were encased in Nikes, or Adidases, or Reeboks. (5) Or something just like them. Along (6) with millions of other Americans old enough to remember Jesse Owens, she has discovered that wisdom lies in dressing like an athlete. (7) There is an 84-year-old New Englander, for instance, whose collection oi sweats rivals that of the heavyweight champion Mike Tyson. (8) She has then in pink and blue and red and gray, and she has them for all OCCA110115. On three-dog nights, when once she might have huddled in bed (9) dressed in a flannel nightgown, banked by the requisite trio of spaniels, she is serene in sweatshirt and pants. (10) And the ice-cube feet that used to make it through the night attired in hand-knitted booties are now toasted by sweat socks - the kind with two stripes at the top. (11) There is nothing new, of course, in drePing practically. (12) But as if you were in training is to do so without to dress sacrificing chic and secrecy. Nursie shoes, wedgies and the little numbers with (13) solaces for the footsore. (14) the tractor-tread sole are dusk But they proclaim the bunion, the callus and the cruel corn. (15) Put on running shoes, however, and who's to know if you're going to walk to work or simply have bum feet? (16) As for sweatsuits, they do what shawls, snuggles and long johns never could: provide warmth at the same time that they project action, of a five-mile Jog. energy and the possibility (17) From toddlers to totterers, millions of Americans now know happier feet and cozier days and nights because they're dressing like competitors for the Golden Gloves. (18) May this fashion never go out of style. The instructions given to the subjects were that they should describe the dominant rhetorical relations in the text. This group completed the task in 22 minutes. Before looking at the result.l of this experiment, I will give you my own description of the dominant rhetorical ilations of the text. The schematic description appears in Figure 1, in which the numbers refer to the sentence 4 '0111 242 numbers in the text. According to this analysis, the nuclear "thesis" of the text is in sentence 12, which expresses, in a nutshell, what good it does to you to "..lress as if you were in training." The Background and Concession, which precede this point of the text, and the Evidence, which follows it, are satellites which substantiate this nuclear thesis. Most of the topics that emerged in the tape-recorded discussion were related to the identification of rhetor;f al relations. The relation names Summary, Evidence, justify, Concession, Background, and Exemplification were expliciily referrers to. The relation which was identified most rapidly, unambiguously, unanimously, and without hesitation was Concession. Concession appeared as a discvission topic twice. This point in the group's analysis coincides with my own analysis. 1-18 BACKGROUND 1-10 11-18 11-12 -18 11 12 13-16 17-18 17 18 Figure 1. Schematic description of the experimental text 243 Another relation which appeared to be easy to identify and which did not elicit much disiussion was Summary. The other relations explicitly taken up in the aiscussion were considerably more difficult to identify. It took the group a long time to reach agreement on the dominant rhetorical structure of the text, and they ultimately approached the same kind of analysis that I presented above, according to which sentence 12 is nuclear in the entire text The group's global analysis approached my analysis, though it took them a relatively long time to reach. The hierarchically dominant Evidence relation, for instance, was not particularly easy to identify. The Background relation came out as Exemplification. Results of a hanslation experiment In the translation experiment my aim was to find out on what knowledge base translators make dedsions while translating. In particular, I was interested in the knowledge base of those decisions in which a translator makes a choice between two or more competing translation variants - irrespective of whether the choice is between lexical items, between alternative phrases or between alternative syntactic or textual solutions. As data for my analysis I was able to use the results from Tuija Pontinen's and Tiina Romanov's (1989) M.A. thesis. Pöntinen and Romanov had made think- aloud protocols of two translators' performance in translating a short LLBA abstract from English into Finnish. One subject was a professional trans:ator and a linguist, while the other subject was an expert in psycholinguistics but a lay person in translating. The text as shown in below, deals with lateralization of language functions in the brain. Danesi, Marcel, Lateralization, affect, metaphor, and language use, Interfaces, 1984, 11, 2, May, 41-46. While localization of many speech functions in the left hemisphere of the brain is well-documented, discourse-related and metaphorical language functions may result from interaction of left and right Research shows that emotive language hemisphere functions. programming, in both expression and elicitation, is content-controlled f1 244 by the right hemisphere and structured by the left. Experiments also show that metaphorization, which violates constraints imposed by semantics, cannot be attributed solely to the analytical functions of the left hemisphere. Research evidence overall seems to refute the view that the right hemisphere is totally inactive in language processing. I made an analysis of the knowledge that the subjects verbalized in their protocols. The verbalized knowledge was divided into three groups: linguistic knowledge, textual knowledge and extra-textual or world knowledge. The point of interest here is textual knowledge. It refers to knowledge which the subjects had extracted from the text, such as the professional translator's observations about .hetorical structure of the text. The professional translator pointed out, for istster...e, tut the text exposes an emphatic contrast between earlier research and recent research on lateralization. The following extract from her protocol (here translated from Finnish into En&'ish) shows that she makes an effort to get this contrast conveyed into the translation as well: "now I should look at these emphases here / the first sentence begins with WHILE LOCALIZATION OF MANY MM BLAAPBLAAP / and then there is some new stuff from DISCOURSE-RELATED EIV onwards / MAY RESULT / so that it seems that this has been only rerently subjected to research / in other words the subclause of the first sentence in fact tells us what is or should be conventional and shared knowledge to everybody in the field / so that this contrast should probably be expressed in the text / yes the text is so sh-at and therefore it would seem strange for instance if I started this sentence with the word "tosin" (admittedly) / there is an abundance of literature which deals with the origin of sev nal speech functions in the left hemisphere but discourse- related and metaphorical language functions ma in fact (. clitic particle -kin) ariSe / if only the style here were to be just a tiny bit more colloquial then the elide particle -kin would be very good here / as a result of the interaction of the left and right hemishperes / well y / or could I perhaps take the liberty of adding something here / but new research results show that / this may perhaps lead the reader better into this / but recent research / lets put It this way / lets give it a try / shows that di.scourse.zd and metaphorical language functions may in fact result from interaction of the left and right hemispheres / that's it then / the editor of the journal will then probably do what he likes to this text so that this is not necessarily the final thing" Such verbalivitions were not found in the lay subject's protocol. She seemed to deal with ea.:11 sentence of the text separately and did not pay attention to connectionS between sentences. 7 245 Another difference in the professional's working is worth noticing in this context. She ran through the entire text nine times, the first readings being mainly characterized by efforts io understand the text thoroughly, and the last times to make sure that the final translation made sense to the reader. It is no wonder, therefore, that this translator got a very good picture also of the rhetorical structure of the text. Conclusion The above observations lead Tri2 to suggest that explicit concern with the comprehension of relational propositions may be characteristic of demanding text comprehension tasks such as translation, abstracting, summarizing, etc., which require an analysis of what is said in the text. More research is needed, however, to find out whether it is realistic to assume that a battery of rhetorical relations can be delineated so accurately that readers can be taught to identify them with a great degree of consensus. I do not believe that it is possible to delineate an exhaustive battery of relations which are mutually exclusively defined. Aiming at such a battery makes an untenable assumption about the nature of language use: it assumes that language use is basically unambiguous, which it is not. It is the ambiguity and "fuzziness" inherent In language use - and the freedom of interlocutors to interpret language in ways which seem most relevant in the circumstances - that makes it impossible to create a battery of rhetorical relations which leaves no :::sidue when applied to concrete texts. BIBLIOGRAPHY Grimes, J. E. 1975. The thread of discourse. Janua Linguarum, Series Minor, 207. The Hague: Mouton. Mann, W. C. and S. A. Thompson 1988. Rhetorical structure theory: toward a functional theory of text organization, Text 8, 243-281. Pontinen, T. A. and T. M. Romanov 1989. Profes.3ional non-professional vs. translator: a think-aloud protocol study. M.A. thesis. Savonlinna School of Translation Studies, University of Joensuu. 246 Tirkkonen-Condit, S. 1990. How do readers infer relation& proposMons in an argumentative text an experimental study. Paper read in the Second ISSA International Conference on Argumentation, Amsterdam, June 19-22, 1990. Forthcoming in the series Pnignolks and disown analysis, Dordrecht Pods. Tirkkonen-Condit, S. 1990. The interaction of world knowledge and linguistic knowledge in the processes of translationl a think-aloud protocol study. Paper read at the Colloquium on Thearetical Aspects of Translation and Meaning Lodz, September 20-22, 1990. To appear in M. 'Mien and B. Lewandowska (eds.), Precendings of the 1990 M4:trick-14ft rolkquison on -Translation and meaning", Part IL Mautricht Euroterm.