can, themselves, be supported (e.g., the statistics from the NHTSA director). The claim holds the top position of an argument schema because all other elements of the argument are presented to either support or oppose this main proposition. As with other schema, awareness of an argument schema can allow readers to accurately represent the author’s intended meaning. It can also aid in organizing subsequent text and guide coherence inferences.
Two studies have looked directly at high-school (Chambliss, 1995) and grade school students’ (Chambliss & Murphy, 2002) use of argument schemas to guide argument comprehension. They presented participants instructional-type argument texts similar to the prototypic structure of argument texts taught in secondary composition courses. In particular, the texts were highly structured with strong topic sentences and introductory and concluding paragraphs summarizing the structure. These researchers found that advanced high-school readers and a majority of the younger readers were able to use their knowledge of an argument structure to correctly organize the basic element in the texts, especially as the external organizers were increased. These argument texts were specially written to conform to the structure taught in class and may have artificially increased performance. It is unclear how well young readers or adults would be able to comprehend authentic arguments of the type that they would encounter outside the classroom.
Much more is known, however, about children’s production of arguments. It has been shown that even young readers can use their knowledge of argument structure to produce arguments. Around the age of seven, children appear to become capable of rudimentary argument production (Golder & Coirier, 1994; Miller, 1986, 1987; Stein & Miller, 1993). For instance, they can state an explicit position and generally support their claim with a reason (Coquin- Viennot & Coirier, 1992; Golder & Coirier, 1994). By the end of their twelfth year, children begin to support their claims with more than a single reason, show an increased ability to detect invalid inferences, and begin to mention counter-arguments (Golder & Coirier, 1994). For many children, argument production ability continues to improve gradually with age, though several aspects of argumentation (e.g., counter-arguments and qualifiers) are still not mastered by the end of high school (Golder & Coirier, 1994; Knudson, 1992; 1994). Given that students can create simple arguments, they must have at least a rudimentary argument schema that could be used to comprehend written argumentative discourse.
Special lexical and syntactic elements in a narrative text can serve as processing cues for constructing a mental representation of the text (Graesser, Millis, & Zwaan, 1996). Blakemore (1987, 2000) points out that some words and constructions, such as “that is” “but” and “because,” do not contribute to the truth-conditions of a proposition but that they do provide processing or conceptual information. In addition to such connectives, discourse markers include headings, numbering, summaries, and typeface changes. These devices are used by writers to emphasize the structure of texts and to signal importance. Prior research has shown that readers’ representation of the structure of narrative and expository texts is aided by signals that clarify the structure (Beck, McKeown, Sinatra, & Loxterman, 1991; Kozmnsky, 1977; Lorch & Lorch, 1995, 1996; Lorch, Lorch, & Inman, 1993; Meyer & Rice, 1989; Millis & Just, 1994). This is especially true for challenging texts (Spyridakis & Standal, 1987) and for low-knowledge participants (Goldman, Saul, & Cote, 1995; Lorch & Lorch, 1996).
In a similar manner, there are discourse cues that can signal argument elements (e.g., claims and reasons) and their relationships. The types of discourse markers that can signal argument elements include connectives (e.g., because, therefore, although), headings (e.g., “evidence for”), and organizational markers (e.g., first, second, finally). These discourse markers can direct the reader to appropriately connect the various argument elements into a coherent structure. In fact, Britt and Larson (2003) found that statements marked by modals (e.g., should) and uncertainty markers (e.g., probably) signaled controversial assertions requiring support. Although attending to such features may be largely automatic for some students, they require inferences that may present less-able students with a significant challenge. Such students may require instruction and practice to become accustomed to attending to these cues.
2.2.3 Connectives, text types, and reading comprehension
There have been a number of studies investigating the role of coherence signals (connectives) in text processing in narrative, and expository text types. For example, Sanders and Noordman (2000) focused on the cognitive status of these relations. Using reading verification and free recall tasks, they investigated the type of coherence relation between segments (e.g., problem-solution vs. list), and the implicit and explicit marking of the relations by means of signaling phrases in expository texts. Both factors affected text processing. Explicit marking of the relations resulted in faster processing but did not affect recall. Carrel (1985) argued that explicit teaching of various aspects of text structure and rhetorical organization of expository texts significantly increased the amount of information ESL students could recall. Furthermore, Joyce, et al. (1998) examined the effects of text genre and repeated reading on written language comprehension in younger and older healthy adults. Participants verified four text-based statements (i.e. explicit, implicit, contradicted, and elaborated) after reading expository, narrative, and procedural texts. Text genre, statement type, and repeated reading produced significant effects. Text genre influenced reading time, with expository passages being read faster than narrative and procedural passages, irrespective of age. In the same vein, Laura and Fuchs (2002) examined the reading difficulty of secondary students with learning disabilities in expository and narrative texts. The participants were administered two expository and two narrative texts. The results indicated that students had more difficulty with expository text than with narrative text in terms of reading fluency and comprehension. Also, Degand and sanders (2002) investigated the effect of causal connectives and signaling phrases in expository texts that were manipulated with respect to the presence or absence of linguistic markers. In some texts they manipulated the presence or absence of causal connectives, in others the presence or absence of causal signaling phrases. The comprehension questions focused either on a manipulated relation or on other parts of the text. The results showed that the implicit condition differed significantly from the explicit condition while the explicit versions did not significantly differ from each other.
2.2.4 Two Approaches to Text Type Analysis
Nearly all classifications of texts by purpose or function derive directly or indirectly from Aristotle’s Rhetoric. However, there is no agreed palette of text types. The range of text types suggested by different linguists depends on the goal of the text typology and the approach with which this typology has been worked out (for an overview, see Lee 2001). The system presented here was inspired by two authors with opposite stances: Egon Werlich and Douglas Biber. Werlich belongs to the tradition of German text linguistics which has not filtered into the English linguistic tradition. Germany experienced a boom of research in text typology in the 70s. Unfortunately most of these studies remain untranslated into English. Werlich is unusual in this respect because he wrote a Text Grammar of English in English (Werlich, 1976). His approach is neither corpus-based nor quantitative; he follows the descriptive tradition based on selected examples extracted from exemplar texts. Werlich’s five text types are very intuitive and reflect cognitive processes: perception in space (description), description in time (narration), comprehension of general concepts (exposition), creation of relations among concepts through the extraction of similarities, contrasts etc. (argumentation); planning of future behaviour (instruction). He has the merit of methodically listing many linguistic and textual features that interact and co-occur in each single text type. His text analysis is based on presuppositions about a well-formed text in the adult’s mind. His approach is called deductive, because the interpretation of the actual text depends on premises about cognitive processes. Biber has instead an inductive approach. His text typology (Biber, 1988 and 1989) is based on his working corpus (the LOB corpus plus the London-Lund Corpus of Spoken English). He thoroughly reviews previous (socio-) linguistic studies and selects 67 linguistic features that can be interpreted functionally. For example, nominalizations co-occurring with passives convey abstract information (Biber 1988: 227). His basic assumption is simple and powerful: if certain features co-occur consistently, then it is reasonable to think that they share an underlying function that encourages their use. In this way the functions of a text, instantiated in text types, are not posited on an a priori basis; rather they are required to account for the observed co-occurrence patterns among linguistic features. He uses factor analysis to identify these co-occurring patterns. His approach was novel and suggestive, but, leaving aside the statistical doubts that factor analysis might raise, it has two main weaknesses. First, his text typology is too corpus-dependent. For instance, the instructional text type is not included in his typology because