XML grammar for constructing argumentative essays Kalli Benetos http://tecfa.unige.ch/perso/staf/benetos/ kbenetos@tele2.ch 08-07-2005 11-07-2005 XML grammar, argumentation, persuasive, essay, writing, structural aid Argumentative Essay Markup Language is an XML based markup language using a Relax NG schema developed during this project to explore the potential of XML as a cognitive tool in the construction of argumentative essays. Using existing schemas, rubrics and lesson plans as a reference for the underlying structure, a schema was developed and an XML editor with some pre-configured transformations using the ArgEssML schema for viewing results, were used as computer-based tools for constructing and editing an argumentative essay. XML proved to be capable of representing the structure of an essay and useful in helping a writer structure his/her essay, but the lack of a tool with a user-friendly interface that does not demand that the user edit within a window containing XML code, proved to be a hindrance to determining whether a computer-supported aid could potentially be more helpful than an equivalent paper representation of the underlying structure of an argumentative essay as presented by ArgEssML.

project definition

From conceptual maps that aid in idea generation (Isnard and Piolat, 1993) to writing aid by providing a structure to help in the linearization of generated arguments during the writing process (Erkens, Kanselaar, Prangsma, Jaspers, 2002), evidence for the potential positive effects structural aids for argumentative and persuasive writing is clear. The goal of this project has been to develop an XML grammar that can be used as the base in the development of a computer-supported tool to help children learn to construct persuasive essays and argumentative essays that also consider the opposing view and in the process to look at the possible strengths and weaknesses of a computer-supported aid to argumentative writing.

1. Can an XML grammar adequately represent the structure of a basic five-paragraph argumentative essay?

The construction the RelaxNG schema that defines the XML grammar for an argumentative essay for children learning to write argumentative essays, began by looking at existing tools available for structuring argumentative and persuasive writing in order to analyze the framework used to guide young writers. While there are several elaborate tools for Computer Supported Collaborative Writing (CSCW) that are aimed at higher education levels, there was only one computer supported tool found for structuring argumentation in a persuasive essay aimed at children: Persuasion Map, a rudimentary yet powerful an online mapping tool in the form of a Flash application. It's hierarchical tree structure and limited character input fields allow only for simple outline of ideas and provide only some cue words as guidance, but were easy to incorporate into an XML grammar and this created the first elements in the schema.

Although literature revealed that there are existing schemas that have been written to represent frameworks for argumentation, none found were appropriate for structuring written argumentative essays. Argument Markup Language (AML), an XML schema for representing structured arguments, was all encompassing but too large and complex to be easily translated into terms and an XML template structure that could be introduced to young writers. Mini-ArgML on the other hand focuses only on the components of a stand-alone argument. Both were consulted for a detailed analysis of the make-up of arguments and argumentation.

Since the users at this stage of development would have to work with XML elements in an XML editor, the naming of the elements was of key importance. The elements had to be semantically relevant not only to the structure of the essay but also use the terminology of the user. Writing guides (Harvey 1999, Darling 2004) and rubrics (Goodrich-Andrade 2005, Harland 2003) consulted proved useful in this respect.

2. Can such a grammar be expanded to include opposing view arguments?

One of the key omissions in Persuasion Map that has been addressed by the XML grammar and schema developed in this project is the possibility to consider an opposing argument to a presented argument and to counter it. Considering an opposing view and countering it is a difficult skill for students learning to write persuasive texts and is highly valued by rubrics used to grade persuasive writing (Goodrich-Andrade 2005, Harland 2003).

ArgEssML represents the underlying structure for flexible argumentative writing that can choose amongst models for simple arguments, counter-arguments and a combination of both for more complex and mature argumentation.

3. Could rubrics be used to create the XML grammar so that it provides structural support for the writing task?

Consulting writing guides for argumentative writing (Harvey 1999, Darling 2004) and rubrics used to evaluate argumentative essays (Goodrich-Andrade 2005, Harland 2003) helped to round out and simplify the structure to the most essential components of an argumentative essay without neglecting the importance of the use of counter-argumentation strategies in effective argumentative writing (Harvey 2003).

Argumentative Essay Markup Language (ArgEssML): The basic structure.

Relevant documents - (ArgEssML.zip)

Used in its simplest form, ArgEssML can be used to create the basic argumentative essay: thesis >> supporting arguments >> supporting evidence with the possibility of adding and editing multiple statements and multiple supports for each statement and a conclusion. There are five basic parts that make up the argumentative essay represented by the elements <header/>, <introduction/>,<arguments/>, and <conclusion/> (See Appendix A for complete documentation - or download pdf).

In the following XML template example the basic structure is presented with all the optional elements (marked with a '?') and the 3 types of arguments that can be constructed (<simple-argument/>, <counter-argument/> and a combination of the two).

In the following XML template example the basic structure is presented with all the optional elements (marked with a '?') and the 3 types of arguments that can be constructed (<simple-argument/>, <counter-argument/> and a combination of the two). The 3 types of arguments can be presented in any order and any combination thereof. There are few optional elements as the structure is very rudimentary as is. The schema includes some attributes that serve as meta-cognitive aids within the argumentation process. These include support sources, and selection of the type of evidence (support) being given, the type of attack being presented as a counter-argument and the strategy being used to defend against the attack. The purpose of the attributes is to give the writer the opportunity to reflect upon the validity and authoritativeness of an arguments presented.

<root>
    <heading>
        <author>
            <first_name/>
            <last_name/>
        </author>
        <essay_title>
            <title/>
          ? <subtitle/>
            <date/>
        </essay_title>
    </heading>
    <introduction>
        <thesis>
            <state_thesis/>
            <outline>
                <importance/>
              ? <conflict/>
              ? <solution/>
            </outline>
        </thesis>
    </introduction>
    <arguments>
        <argument id="1">
            <simple-argument>
                <state_argument/>
                <support source="xxx" type=""/>
              ? <relate/>
            </simple-argument>
        </argument>
        <argument id="2">
            <counterargument>
                <state_counterargument attack_type=""/>
                <come_back strategy=""/>
            </counterargument>
        </argument>
        <argument id="3">
            <simple-argument>
                <state_argument/>
            <support source="xxx" type=""/>
            </simple-argument>
            <counterargument>
                <state_counterargument attack_type=""/>
                <come_back strategy=""/>
            </counterargument>
        </argument>
    </arguments>
    <conclusion>
        <thesis_summary/>
        <arguments_summaries>
            <argument_summary />
            <argument_summary />
        <consequences/>
    </conclusion>
</root>

4. What tools could be used to present the structure in an interface that can be easily used by children (10 and over)

There were no tools or editors found that could present the XML-based structure in an interface that would be user-friendly for children 10 and over. For the purpose of testing the XML structure developed, OxygenXML editor was used with an adult tester. Three transformation scenarios, using different style sheets, were created to render the XML file the user worked with, so that the user could see his progress in a browser window. The first (essay.xsl) renders the user's entries but also shows instructions to help the user with his task. The second (essay_done.xsl) presents the text only. The third (essay_print.xsl) presents the text in a print friendly version.

The user had no knowledge of XML (but was familiar with the concept of HTML tags) and no formal knowledge of argumentative essay writing. A blank template XML file was presented with generic entries that help explain the type of content expected and comments to guide the user. After considerable instruction on the basic concepts of argumentative essay writing (as opposed to expository or narrative), a presentation of the underlying structure (argument types and components), a brief introduction to XML, OxygenXML editor and how to run the transformation scenarios and view the results, he began to write. The process was difficult at first and the user was frustrated and distracted from his task by the 'tools', but after 20 minutes had managed to master the essentials, but relied on copying and pasting the template elements for inserting new arguments and validating frequently. With children having no concept of tag elements and code views, the cognitive load would be too large to effectively learn the skills of argumentative essay writing.

After seeing entries rendered with the instructional text displayed, the user expressed that if the light coloured text areas were text fields into which he could insert his text directly, he would be more readily inclined to use a computer-based aid to write an argumentative text. To make a computer-supported tool that uses ArgEssML to help children structure and write argumentative essays the tool would have to have an interface that is much like the XML file rendered using the style sheet that displays instructional text to guide the writer but behave like a browser-based form with editable text fields and drop down menus for input and buttons to view the resulting text only and print friendly versions. The interface would also allow the user to add their choice of argument model and remove them with simple button clicks.

More problematic are concerns that such a tool may result in formulaic, disjointed text. The intention of the ArgEssML schema and any tool that uses it would be to teach the principles of writing an argumentative essay, and beginner writers would not easily be able to vary their wording and sentence structure to make for a more flowing and interesting text. The schema can be adapted to allow for the insertion of an optional 'free phrase' with each existing model, whose purpose is to connect arguments and make the text flow, which students would be allowed to use only after mastering the basics.

5. Is there a need for such a tool? What are the possible advantages over a pen and paper scenario?

If the amount of literature, lesson plans and writing guides available are in any way representative of the difficulty children have writing well-constructed argumentative essays, then the need for tools that aid in the task is significant. Whether these tools need or should be computer-supported is the more important question. The schema developed and the tool envisioned from the results of the preliminary testing are intended to be used as a tool and teaching aid, and not to be auto didactic, so with the same instruction and guidance prior to the task, the advantages over a paper and pen scenario are not easily discerned.

The immediate advantages such a tool may have over a pen and paper scenario is that the product is digital, therefore edited more easily and can be exported and then further edited if needed. The less obvious advantages may lie in its ability to guide the development and structuring of arguments, as well as the structural flow of the text, from outline to expounded text in one visual field (with some scrolling), where as with pen and paper, ideas would have to be mapped and outlined, then developed, structured and expounded linearly. This is difficult with out a lot of paper and a lot of rewriting. A very devoted student may use different sheets for each argument and then rearrange them until a definitive version is reached, but more likely is the 'outline, then expand the text linearly from the outline version' scenario.

6. What are the main obstacles to using XML as a cognitive tool and what can be done to overcome these obstacles?

The main obstacle encountered in this project has been the unavailability of a WYSIWYG XML editor. Without an authoring tool that requires significantly less cognitive and teaching resources than the task to be accomplished using the tool, it would become impossible to convince teachers to use a computer-supported tool to teach the writing of argumentative essays even if the initial investment in learning to use the tool were proven worth the effort given the resulting benefits (which remain to be discerned and proven).

Another obstacle that arose with the adult tester but may be less significant with children is the resistance of writers on principle, to accept a predefined structure to their writing, even if it is entirely natural and easy to use.

A preliminary test with one user showed ArgEssML to be, in itself, useful in revealing to the user the structure and components of an argumentative essay and in guiding the writing process. The user admitted to being helpfully guided to reflect upon the nature and quality of his thesis, arguments and supporting evidence, the variety of argument types used and the ordering of the arguments. The possibility to outline and later fill in precise wording was also appreciated. This shows that XML can be successful at representing the structure of an argumentative essay, but does not prove that a computer-supported authoring tool developed using ArgEssML would be more effective than a paper and pen scenario that presented the equivalent structural information. A fully developed user-friendly authoring tool and some comparative studies to look into which support best allows writers to construct better essays and more importantly, later construct argumentative essays without any support may prove revealing.

OxygenXML homepage and download (30 day free trial).

Darling, C. (2004) "Composition Patterns: Developing an Argument." Guide to Grammar and Writing. Capital Community College Foundation. <http://cctc2.commnet.edu/grammar/composition/argument.htm> (accessed June 22, 2005)

Erkens, G., Kanselaar, G., Prangsma, M., Jaspers, J. (2002) Using Tools and Resources in Computer Supported Collaborative Writing, Dept. of Educational Sciences, Utrecht University, The Netherlands <http://newmedia.colorado.edu/cscl/149.html >. (accessed June 20, 2005)

Goodrich Andrade, H. (2005). Appendix C: "Using Rubrics to Promote Thinking and Learning", Science Grade 8 Exit Project Guide, NYC Department of Education <exitproject.htm> (accessed June 20, 2005)

Harland, D (2003) Using asTTle persuasive writing: acase of teaching argument writing. asTTle Technical Report, #29, University of Auckland/Ministry of Education

Harvey, G. (1999) "Counter-Argument." The Academic Essay: A Brief Anatomy. The Harvard Writing Center. Harvard University. <http://www.fas.harvard.edu/~wricntr/documents/Counterarg.html> (accessed June 22, 2005)

Householder, M.The Silent Debate. lesson plan found on the personal website of Elizabeth Losh, of the Electronic Educational Environment of University of California, Irvine. http://eee.uci.edu. <http://eee.uci.edu/faculty/losh/Shakespeare/householder.html>

The Five Paragraph Essay

Isnard, N., Piolat, A. (1993) The effects of different types of planning on the writing of argumentative text. University of Provence, Aix en Provence. < IsnardPiolat1993.pdf> (accessed June 20, 2005)

Levelt, W.J.M. (1988). Speaking: from intention to articulation. Bradford Books, MIT press, Boston MA, USA. in Erkens, G., Kanselaar, G., Prangsma, M., Jaspers, J. (Ed.) Using Tools and Resources in Computer Supported Collaborative Writing, Dept. of Educational Sciences, Utrecht University, The Netherlands, 2002. <http://newmedia.colorado.edu/cscl/149.html> (accessed June 20, 2005)

SRI International. (2002) Argument Markup Language. Menlo Park, CA <http://www.ai.sri.com/~seas/aml/index.html> (accessed June 20, 2005)

Suthers, D. (1995) Towards an Interlingua for Information Objects. Position Paper presented at CHI ?95 Research Symposium, Denver, Colorado U.S.A <http://lilt.ics.hawaii.edu/lilt/papers/1995/suthers-chi95argml.pdf> (accessed July 8, 2005)

Appendix A - ArgEssML documentation (PDF)