This study investigates the effect of sentence presentation format for optimal processing and takes into account previous research on the way in which people/readers/parsers parse and comprehend sentences. The study included 29 native speakers of English, who were presented with a cohort of sentences each comprising a verb that could be transitive or intransitive but designed as Late Closure sentences to elicit transitive readings or Early Closure sentences to elicit intransitive readings.
Results initially adhered to Late Closure in that Late Closure sentences were read faster than Early Closure sentences but further analysis proved that page presentation and line breaks had an impact on the Early and Late Closure sentences which impacted on the readers comprehension and thus posits that initial parsing is not based solely on syntactic rules but includes other factors such as punctuation, which in this case is represented by line break presentation. Introduction
Language is complex and sentences in natural languages are usually highly ambiguous and can be interpreted in many ways, but despite this complexity sentence comprehension seems to be easily and quickly achieved. When reading a sentence the sounds (or letters) have to be grouped into words or morphemes and meaning needs to be retrieved for those words or morphemes, syntactic information must be sought, analysed and integrated into syntactic structure, and semantic interpretation also needs to be designated; this process is called parsing, and it is the way in which parsing is undertaken that is of particular interest to researchers.
In other words – how do people parse? Mitchell (1994:375) tells us that although a great deal of research questions on the issue of sentence interpretation have been derived from an assortment of conceptual classifications, the chief inspiration has been Chomsky’s generative linguistics and his use of tree diagrams and phrase structure rules, which provide a basis from where to determine the different roles of people and objects within the sentence. The main problem however, “is to determine how people convert a string of words into something like a tree diagram to represent the structure of the sentence” (Mitchell, 1994:376).
Do they make an immediate guess about where the current word goes, even if it turns out to be wrong or do they wait until they know for sure before they decide on the syntactic structure of the whole sentence? Parker and Riley (2005:252) tell us that much “of the research in language processing has been concerned with how people resolve syntactic ambiguity” and that ambiguity can be observed in a ‘garden path sentence,” wherein the sentence appears to have one structure but later it turns out to have another.
As in the Garden Path Model adopted by Frazier (1987), many researchers suggest or take for granted that when confronted with structural ambiguity readers cope by following one analysis or interpretation. Frazier (1987) also adopted the core principles – Minimal Attachment and Right Attachment – from Frazier and Fodor’s (1978, 1980) Sausage Machine approach to explaining parsing preferences.
Frazier (1987:9-10), however, reformulated these principles, wherein Minimal Attachment accounts for the strategy used when putting new words into syntactic trees by using as simple a syntactic structure as possible, and Late Closure strategy replaces Right Attachment and accounts for continually adding new words to a syntactic constituent instead of locating another place for them if they are grammatically acceptable, while prolonging the closure of that syntactic constituent (clause or phrase).
Thus, as in Frazier and Fodor’s Sausage Machine, Minimal Attachment makes sure the parser establishes the most straightforward structure in the first analysis of ambiguous sentences and Late Closure certifies that new elements are directly fixed to preceding data, thereby curtailing the likelihood of excessive workloads on memory. It is the fact that the Garden Path Theory (Frazier, 1987) regards syntactic processing as modular – automatic and based purely on the basis of structural information – that has been the most controversial issue for researchers.
Other researchers provide evidence that other factors such as referential pragmatics (Crain and Steedman, 1985), lexical semantics and plausibility (Mitchell 1987), lexical preference (Holmes, 1987), lexical frequency and combinatory idiosyncrasy (MacDonald, Pearlmutter and Seidenberg, 1994), and prosody (Warren, 1996) have a quantifying outcome on sentence comprehension and determining sentence ambiguity.
Clifton, Frazier, and Connine (1984) conducted a study wherein sentences containing only verbs that are optionally transitive (some where a transitive reading and others where an intransitive reading is preferred) were presented to subjects, stopping after the first word following the verb for them to make lexical choices. Results showed that subjects acted more quickly when the word following the verb complemented its preferred argument than when it did not; thus giving proof for Clifton, Frazier and Connine that lexical information is accessed and used very quickly.
Holmes (1987), Kennedy et al. (1989) and Ferreira and Henderson (1990), all investigated the issue of another kind of ambiguity. They tried to account for ambiguity in sentences where the verb can take a direct object or a complement and found that different verbs possess lexical preferences and thus prefer different kinds of complements and an NP following a verb could be taken as a direct object which could lead to a Garden Path when the second verb is read, or as a subject of the complement.
In 1987 Mitchell conducted a study, results of which he interpreted as support for two stages in processing – a syntactic structure is built on the basis of major category information only – in the first stage and then more lexical information is used in the second stage to eliminate any incorrect attachments. Mitchell’s (1987) experiment gained support from an imitative study undertaken by Stowe (1989), using the same subject-paced word-by-word reading and an eye-tracking technique but results appear to contradict Mitchell’s findings and suggest that lexical information is used in the initial stage of parsing.
Thus we find that research undertaken to determine how people parse is far from conclusive and there remains differing viewpoints as to whether other factors other than syntactic factors influence the comprehension of a text when parsing. This study sets out to investigate whether page format – presentation of sentences – has any impact on time taken to parse and understand a sentence, in order to establish the best presentation for optimal processing, which would be relevant in advertising and education.