Investigating Summary Typology : Considerations For Classification

The professional writing teacher who undertakes to teach practical summary-writing soon realizes that condensed texts take many forms and that summarizing skills must be adapted to specific contexts and fields. The need for a clear description and classification of summary types quickly becomes apparent. This paper explores methods of describing and classifying summaries on the basis of structural, metatextual, cognitive, and contextual criteria in an attempt to provide a multidimensional framework for classifying summaries. 37

SUMMARY-WRITING IS OF INTEREST TO EDUCATORS, communication specialists, and researchers in a number of fields.Teachers use summarywriting tasks to foster and evaluate cognitive and linguistic skills in their students; abstractors and subject specialists write abstracts of articles and documents; and professional communicators are turning more and more frequently to summaries in their struggle to transmit ever-expanding amounts of information.Researchers have investigated summary-writing to gain insight into discourse comprehension: linguists have studied summarizing to learn more about memory and recall and have analysed summaries in their attempts to represent the semantic macrostructure of texts (van Dijk andKintsch 1983), Technostyle Vol. 11, No.314 1994 and information specialists have studied abstracting strategies, formats, and practices (Collison 1971;Cremmins 1982;Ratteray 1985).
Since professional writers are frequently called upon to summarize texts for various purposes, the summary is of particular interest to teachers of professional writing.Moreover, the process by which a summary is produced involves strategies used in many forms of written communication: selection of important ideas, analysis and synthesis of information, and reformulation and condensation of material.The professional writing teacher who undertakes to teach practical summary-writing soon realizes that condensed texts take many forms and that summarizing skills must be adapted to specific contexts and fields; the need for a clear description of summary types quickly becomes apparent.
On the most practical of levels, a preliminary distinction between types of summaries can be made on the basis of their use and the nomenclatures used to distinguish these uses (Russell 1992).For example, the precis, or school-sponsored summary, is used as an academic exercise to develop a student's writing skills.The abstract, perhaps the most closely studied and most standardized form of the summary, briefly presents the contents of a document in order to facilitate research and dissemination of knowledge in a specialized field.A summary record condenses the content of a speech or oral statement and is used as an official document in organizations such as the UN.Minutes record the proceedings of meetings.Executive summaries enable busy business people to decide whether to read adocument and are designed to facilitate consultation of specific sections of the document.Popularized summaries condense and adapt specialized material for a general readership.In the field of journalism, headlines, leads, and press releases are all forms of summaries, and journalistic articles themselves summarize events.These terms help to distinguish the various practical applications of the summary, but more theoretical approaches can also be used to differentiate summary types.Thus, I propose to explore methods of describing and classifying summaries on the basis of structural, metatextual, cognitive, and contextual approaches.

The Structural Approach
The field of structural linguistics provides traditional text typologies that can be used to differentiate between summaries on the basis of the text type of the source text (that is, the text that is to be summarized.)Beaudet  ,, . .
.*'"""""'"""'-''~~~~M~t"•"'' • • - ( 1994) describes structural models of text types and their application to the classification of source texts in summarizing.This classification distinguishes between such types as descriptive, narrative, expository, argumentative and instructive texts, in part on the basis of the logical organization of information in the text.This structuralist approach is reflected in a number of Frenchlanguage books on summary-writing (Valentine 1990, Moreau 1977, Boret and Peyrot 1969).
When teaching summarizing, the professional writing teacher may find it useful to familiarize students with such text types and to classify the source texts accordingly.The student's familiarity with text types enables him or her to anticipate and process information more easily when reading the source text.Recognition of structural types facilitates and shapes the comprehension and analysis of texts, which are key steps in summarizing.
Once a reader detects the typical structure of a text type, he establishes a mental frame of what is likely to be included in the text and is then able to concentrate specifically on the semantic content of the text.A schema aids and speeds the comprehension of texts by establishing expectations of what will follow in the text and how the information is to be interpreted (Liddy 1988, 25).
Such text types are thus useful for classifying the source text.But what about the summary text itself?Is the typology of the source text useful in classifying the summary text?Does the summary text simply reflect the structure of the source text?Or can summary texts themselves be classified on the basis of their own internal structure?Abstracts are one summary type which do seem to exhibit a distinct internal structure: they commonly consist, in very general terms, of scope and purpose, methods, results, and conclusions (Cremmins 1982, Collison 1971).In an in-depth analysis of 276 empirical abstracts, Liddy ( 1988.) describes the structure of this summary type and identifies recurring lexical clues that are evidence of such a structure.Such observations on the typical discourse structure of abstracts apply specifically to empirical abstracts, which summarize empirical studies that are already structured by scientific methodology.Thus the structure of abstracts of empirical research articles reflects the structure of the source discourse and of scientific method itself.However, not all abstracts exhibit this structure, for not all abstracts summarize empirical articles.For example, abstracts published in the Computer Select database exhibit an entirely different pattern: they summarize articles that describe new computer products and software, and thus they give descriptions of different.products and sys-terns rather than following the pattern of the empirical abstract.
Like other summaries, abstracts reflect, to a certain extent, the structure of their source texts.It is apparent that any attempt to describe and classify summaries must take into account the existence of both the source text and the summary text, and must reflect the relationship between the two.

The Metatextual Nature of The Summary
The summary is a metatext -that is, it is a text about a text, a "secondorder discourse" (Kintsch and van Dijk 1978, p.376).The metatextual nature of the summary can be a source of a number of choices on the part of the summarizer and can lead to the production of different types of summaries.Thus the nature of the relationship between the source text and the summary text can provide a possible theoretical framework for classification.

Metatextual Reference
A number of researchers have distinguished between summary types on the basis of this relationship between source text and summary text.For example, Ratteray ( 1985) baseshis classification in part on the psychological distance .between the two texts: he uses the term "surrogate" to describe texts prepared "as if by the hand of the original author or speaker," and contrasts these with texts in which the summary-writer acts as an intermediary between the original writer and the ultimate user.Gaillard and Lau nay ( 1979) use a similar distinction to differentiate between the resume, in which the summarizer substitutes for the original author, and the analyse, in which the summarizer explains the original author (12). 1  Metatextual reference in summaries can take various linguistic forms.In his article on the linguistic genre of abstracts ( 1991 ), Michael Jordan observes that descriptive•abstracts tend to contain clauses that report what types of The following example illustrates this difference of approach in summarizing.Source: I am delighted with the work done by this committee; the members have worked long and hard, and have been tireless in their efforts.Summary 1 -Summarizing in the voice of the original author: I congratulate the committee on its work.Summary 2 -Distancing oneself from the original author: The speaker praised the efforts of the committee.
Technostyle Vol.11, No.3 I 4 1994 information are given in the original document, whereas informative abstracts are more likely to be summaries which give the information itself.Descriptive abstracts frequently contain clauses that describe the paper and make metatextual reference to the original-either by referring directly to the document itself, for example by using it "as the grammatical subject of a select number of verbs in the active voice," ( 1991, p.510) , or indirectly through untriggered passives.
Glaser (1991, pp.9-10) also reports linguistic evidence of metatextual reference in abstracts: she found a high percentage of passive verb forms in her corpus (as high as 58.3% of verb forms in one journal), as well as a high number of metacommunicative phrases in the initial sentence position.Basham ( 1986) defines metatextual elements as "those elements in a summary that serve as 'text about text,' that is, those elements that frame the summary in relation to its source text" (110).She identifies gist statements that include "formulaic references to the title and/or primary author of the source text" ( 110).She goes on to distinguish between summaries on the basis of the degree of personal involvement of the summary-writer in the summary text.Laurent ( 1985) distinguishes clearly between two ways that the summary can represent the source text.On the one hand, it can restrict itself to the information in the source text (in which case the summary text is co-referential with the source text); in this case, the summary text is presented as if the circumstances of its production were the same as those of the original's (Laurent 1985, p.84).On the other hand, the summary can include not only information from the source text proper, but also information about the communicative situation of the source text, in which case the summary-writer distances himself or herself from the original.
The distinction between these two types of summary is of concern to the professional writing teacher, since they require different approaches and strategies.Laurent states that the texts that are the easiest to summarize are those which are not dependent on contextual information for comprehension, but rather involve "conceptual" information.More difficult are texts which contain indicateurs d 'individualite dependent on the spatio-temporal circumstances of the source-text production (such as first-and second-person pronouns, demonstratives with exophoric reference, and adverbs indicating spatial and temporal relations.)

Parallels Between Source and Summary Text
Further distinctions can be made between summaries on the basis of the relationship between the source text and the summary text.The distinction between a summary that follows the linear development of the source text and one that resequences material has led researchers to distinguish between precis and synthesi::ed summaries (Hidi 1983), or between sequential summaries and synthesizing summarie.1. Ratteray 1985, p.458), or between the resume and analyse (Gaillard and Launay 1979).Maldidier and Normand ( 1982) differentiate between a summary based on a deconstruction and reconstruction of the original, and one based on linear reduction; they contend that these two types of summaries reflect two contrasting conceptions of the comprehension process: the rationalist model, in which meaning is deconstructed through analysis and then reconstructed through synthesis; and the empiricistmodel, in which meaning is grasped as a whole and then reduced as if by an optical instrument (114).
The use of new wording in a summary also provides a basis for distinction between types of summaries.Virtually all books on the precis, or the school-sponsored summary, state that the student must write the summary in his or her own words.Indeed, this is perhaps the main distinguishing feature of the school-sponsored summary.In summaries written for professional rather than pedagogical purposes, however, using the wording of the original where appropriate is not only tolerated but often recommended.Abridgements, for example, are designed to capture some of the wording and style of the original.Minutes must report the exact wording of motions and resolutions.In abstracts, the abstractor is not only allowed but often encouraged to use the wording of the original where appropriate, in order to avoid unnecessary paraphrasing that could lead to distortion (Collison 1971, p.3).
Perhaps the most formal attempt at classifying summary •types on the basis of such relations between source text and summary text was undertaken by Pouzet ( 1981 ), who described nine summary types (including the resume scolaire, the r~sume informatif, the resume critique, and the synthese) and then distinguished between them on the basis of eight distinctive features, including invariance informative, distance enonciative (distancing), linearite (the absence of resequencing), and reprise des signifiants (the use of the original signifiers).Pouzet's typology is useful in that it highlights some of the differences and similarities between types of summaries.However, her attempt to establish a firm typologytends to oversimplify the possible variations within Technostyle Vol.11, No. 3 I 4 1994 ..,., _________________________________ ... summary types: she reduces the complex and variable configurations of summaries to the simple presence or absence of specific features.

Cognitive Considerations
Different summarizing strategies produce different types of summaries.Thus, as suggested in the preceding section, an examination of the cognitive dimension of the summary-writing process will throw light onto differences between the resulting summary texts.
In recent years, much research has been done into summarizing strategies; most of this research has taken as its basis the work of Kintsch andvan Dijk (1978, 1983).In Kintsch and van Dijk's processing model of discourse comprehension, a text's semantic structure is represented by an abstract text base, in terms of sequences of propositions.During discourse processing, the reader employs a number of macrorules (deletion, generalization, construction) to construct the macrostructure of the text.This model has been used by researchers (Hidi 1983;Brown and Day 1983;Sherrard 1986Sherrard , 1989;;Johns 1985;Winograd 1984) to explain the differences between "mature" and "immature" summaries on the basis of the strategies used to produce them.Mature summaries are characterized by the combination and integration of ideas across paragraphs, the rearrangement of material by topic cluster, and the statement of the gist in one's own words (Brown and Day 1983, p.2). Mature performance involves "deep-level transformation of a text, departing from surface wording and original order of propositions" (Sherrard 1989, p. l) and the recombination of ideas "into novel configurations" (Hidi 1983, p.4), the result of higher-level macrorules.The mature summary contrasts with summaries produced by weaker or inexperienced subjects, in which the summarizer focuses on the surface structure of the original and reproduces verbatim elements of the original -thus using the lower-level copy-and-delete strategy.Ratteray ( 1985) warns of the dangers of researchers analysing summarywriting performance on the basis of a set of rules without taking into consideration the conception that the subject might have of the task required.He advises researchers to make explicit distinctions between summary types in order to ensure that they do not "interpret with one set of criteria the data generated from subjects who summarized unwittingly by a different set of criteria" ( 469).Professional writing teachers should also take this advice into consideration when evaluting their students' summaries.
The purpose of the summary can also provide grounds for classification.Once again the field of abstracting provides an example: in the traditional distinction between the indicative (or descriptive) and the informative abstract, the purpose shapes the content.The indicative abstract is "a brief description written to help the user understand the scope of the original document, without giving him a detailed step-by-step account of it" (Collison 1971, p.27); it thus describes the purpose and method of the original but does not contain extensive data.The informative abstract is designed to give more detailed information on the content of the original and in some cases to substitute for the source text; it thus gives more extensive data and includes results, recommendations and conclusions.
Further distinctions can be made on the basis of the intended readership of the summary.A technical text could be summarized by an abstract, but it could also be popularized.An abstract is aimed at an expert audience with specific needs, whereas a popularized summary simplifies specialized material for a non-specialist readership.Thus two distinct summary types could be produced from one source text, each aimed at a different audience and having a different purpose.
Mode or channel of communication is also a consideration in describing summary types, and in describing the source-text-summary-text relationship.For example, summary records and minutes of meetings communicate in written form the content of spoken discourse events, and thus involve a change in mode in their production.
Summaries can also be differentiated on the basis of their field or domain: they can be classified according to the type of professional or occupational activity that is being engaged in.According to this criterion, summaries could be classified as scientific/technical, administrative, political, religious, literary, journalistic, legal, or promotional.The discourse community for which the summary is produced is a similar consideration in description and classification.A full description of summary types as genres remains to be done; this issue invites further research.

Conclusion
In my attempt to differentiate between types of summaries and to suggest a basis on which they can be classified, I have adopted a multidimensional approach.Teaching the summary with such an approach can help sensitize the student to the complex and varying requirements of the task.Rather than Technostyle Vol.11, No.3 I 4 1994 restricting classroom work to the traditional academic precis, the professional writing teacher can help students learn to produce professional types of summaries as well.Summaries vary in the extent to which they can be characterized along different dimensions; a combination of the approaches I have suggested can produce a "kind of reference grid to be used flexibly and selectively" (Emery 1991,p. 575).By understanding the complexity of the summarizing task and the variation of summary types, the teacher is better able to help students learn to master the undeniably important skill of writing a summary.