• translation
• proofreading
• pricing and ordering
• resources
• references
• contact us
• home

CHAPTER IV

METHODS AND PROCEDURES

            This chapter describes the database, both how it was collected and what type of subjects participated in the project.  It also explains the procedures used in the analysis.

Subjects 

            The subjects of this study were all students of the English Language Institute at the University of Texas at Arlington.  This is an intensive program attended mostly by students planning to enter an American university and whose language skills are not sufficient to perform adequately in a university setting.  All of the students have completed a course of study at least equivalent to high school.  Many of them have finished their undergraduate work in their home countries and want to enter graduate school in the U.S. 

            Students in the program study English 5 days a week 4 hours a day.  Instruction takes a variety of forms: grammar, composition, speaking, listening, and reading.  Ten hours per week is spent on grammar/writing skills with the rest of the time divided evenly between oral skills and reading.  The program has six levels, 1 being the lowest and 6 the highest.  Only intermediate and advanced students (levels 3-6) were chosen for the study since students below that level have limited English proficiency and their production of relative clauses is low at best and usually non-existent. 

            All of the students were given a battery of placement exams during the first week of class.  The results of these exams determined what level a student would be in for a particular skill.  A student is seldom in the same level in all of the skills.  The skill and corresponding level which I used to group the students was their placement with regard to their grammar/writing class.  A description of the levels and the criteria established by the institute and used to determine what level a student belongs in is included in Appendix A. 

            The Spanish speakers in the study came from a variety of countries: Columbia, Ecuador, Guatemala, Honduras, Mexico, and Spain.  The Korean students were all from South Korea.  A breakdown of the number and nationality of the students by level follows.  Some students wrote more than one essay in a given level (on different topics), so the number of students does not match the number of compositions.

Table 4.  Number of students and compositions, by level.

Data

            As in Schacter's original study, this project is primarily concerned with how learners from these two particular language backgrounds use relative clauses in free compositions.  This type of unrestricted data allows the researcher to see how the learners have integrated relative clauses into their IL competency.  Other types of data, for example, grammaticality judgment tests and sentence combining, prejudice the learners towards making judgments about or using relative clauses.  Thus, while useful for some types of research, they do little to inform us about the frequency and function of relative clauses in the production of the learners.  The compositions used in this study were collected over an 18 month time span and were all written with the same constraints, which are described below.

            To control for the possible effect of genre, the students were limited to expository topics.  Some of these were compare and/or contrast essays while the others were more straightforward exposition.  The more common writing topics were "Describe and explain what you believe are the most important requirements for happiness."  and "Choose two places where you have lived and write an essay in which you compare and/or contrast these two places."  A complete list of topics is given in Appendix B.  As well as controlling for genre, this helped insure that students could pick topics which they were more comfortable with in terms of vocabulary and background knowledge and which were more interesting for them.  The students were allowed to use dictionaries and thesauruses while writing.

            Composition length might also be a factor in the frequency of RC production.  Students were limited to two hours, but very few students wrote for more than one and a half hours.  In addition, a count was made of the number of T-units in each composition (Hunt 1965).  A T-unit is an independent clause with all of its attending dependent clauses.  Although this does not control for length in an absolute way, it does give some indication of how long the compositions were.  Additionally, the T-unit count allows for a more specific statement regarding the number of relative clauses per independent clause. 

Analytical Procedures

Scope of analysis

            The analysis was restricted to relative clauses with finite verbs; in other words, no reduced relative clauses of the following type were included:

            (1)  The man standing on the corner was hit by a truck.

This decision was made for a variety reasons.  First, these constructions are frequently ambiguous as to whether they are functioning adjectively or adverbially.  

Secondly, I wanted this study to be comparable to other studies of this sort and they do not usually include reduced relatives.  Lastly, in pilot counts, this construction was found to be extremely infrequent in the student's production.  Complement clauses or what some call Headless relatives were also not included in this study.

            (2)  What she likes best is chocolate. 

            All other relative clauses, both restrictive and non-restrictive were included.  Constructions were included as long as it was clear that they were intended as an RC.   

            (3)  Seoul has four seasons which spring, summer, fall, winter.

These instances all included the relative pronoun, but left out the verb, always a copula.

Now, we will turn to how these were analyzed.

Grammatical Function 

            The relative clauses were classified with respect to the grammatical function of the Head Noun in the main clause.  There were five possibilities: Subject (S), Object (O), Other (X), Subject of Existential (EX) and Predicate Nominal (PN).  The Other category is almost exclusively Prepositional Phrase Objects, but includes Indirect Objects, Locative, Time, and some Head Nouns which were not the argument of a verb because they were in a sentence fragment.  

            The NP in the relative clause which is coreferential with the Head Noun was divided into 4 types:  Transitive subject (A), Intransitive subject (S), Object (O), and Other (X).  The grammatical functions OPREP, GEN, and OCOMP were collapsed into one category, Other.  One reason for this is that Genitives and OCOMPs are simply infrequent even in native speaker usage. In addition, Gass (1980) has demonstrated that L2 learners tend to avoid relativizing on these positions.

Definiteness of Head Noun

            The definiteness of the Head Noun is a factor that several researchers have found to play an important role in the formation and function of relative clauses, among them De Haan (1987) and Fox and Thompson (1990).  A Head Noun was considered definite if it was preceded by a definite article, a demonstrative pronoun, a specifying genitive, or a possessive pronoun.  All proper nouns and demonstrative pronouns were also considered definite.  All other NPs were considered indefinite. 

Type of Relative Clause

            The relative clauses in the data were all divided into several categories depending on their function in terms of the type of information encoded by the relative clause with respect to the Head Noun and the matrix clause. 

            The first distinction is the traditional restrictive versus non-restrictive distinction.  This is a semantic distinction which is marked in English by an intonation break in spoken discourse.  This usually corresponds to the commas used to offset the relative clause in written text.  Another characteristic of NRRCs in English is the use of a WH subordinator.  In other words, that  and Ø are not allowed.   L2 students frequently make mistakes in their production of NRRCs because many languages do not make a morphosyntactic distinction between restrictive and non-restrictive RCs.  Korean does not differentiate between RRCs and NRRCs, but  Spanish does.  Relative clauses are most often non-restrictive if the Head Noun is proper and hence does not need any identification, or if context demands a non-restrictive interpretation.  Because of the mistakes made by students when using NRRCs, some interpretation of the data was needed, but in this study, RCs were only considered non-restrictive if the orthography warranted or if there was no other possible interpretation because of the semantic value of the Head NP (e.g. all RCs on proper nouns are nonrestrictive). 

            Restrictive relative clauses were divided on the basis of whether they represented new information or old information (Bernardo 1979) .  The latter group are called identificatory relatives by Bernardo and will be so called here.  To be considered an identificatory relative, two requirements must be met.  (1) The Head Noun must be Given.  In almost all cases, this meant prior mention in the discourse, but there were instances where the Head Noun was more generic and could be inferred from the discourse.  (2) The information in the relative clause must be information already known to the reader.  All other relative clauses were considered informative, and, as recent research has shown, this is the primary function of relative clauses in English. 

            The informative relatives were then further divided into two groups: Asserted and Presupposed (Ziv 1975, De Haan 1987).  To distinguish between the two I used a slightly modified version of the separability test applied by Bernardo (1979).  Bernardo claimed that a sentence like (4) could adequately be replaced by (5).

            (4)  There’s a man in the field who’s picking pears.

            (5)  There’s a man in the field.  He is picking pears.

He also said that information encoded in the RC in (6) could not be separated from the main predication.

            (6)  She knocked the hat he was wearing off on the ground.

            (6a)  *She knocked the hat off on the ground.  He was wearing it.

            While Bernardo is right in a sense, it seems that the question of presupposition is really at the heart of the matter, for, if we reverse the propositions, separability turns out to be no problem.

            (6b)  He was wearing a hat.  She knocked it off on the ground.

De Haan (1987) makes a similar point about the presuppositionality of information, especially with relation to definiteness.  Presupposition seems to be primarily a matter of informational hierarchy.  Some information is logically prior to or required for the interpretation of other information, and while an RC is one way of encoding this type of information, it is not the only function of an RC.  Consider (7).

            (7)  John drives a car.  The car is a wreck.

            (7a)  The car John drives is a wreck.

            (7b)  John drives a car which is a wreck.

Given these two propositions, a speaker has a choice as to how to present the information.  In (7b), the RC is not presupposed, while in (7a) it is.  If the information in the RC is presupposed, the Head NP is definite since it is now considered identifiable.  In (8a), the information in the RC does not necessitate the use of a definite article since it is common knowledge that we all know many people and hence this information does not specify the referent to the point of definiteness.  The definite article in (8b), though grammatical, definitely requires specialized contextual support, and could not be used to introduce a referent. 

            (8a)  A man I knew said ...

            (8b)  The man I knew said ...

Nature of Head Noun

            Some languages have a tendency to relativize more frequently on nouns from a particular class (e.g. Japanese on personal pronouns).  All occurrences of Head Nouns belonging to any of the following categories were noted: personal pronouns, indefinite pronouns, demonstrative pronouns, and, what I have termed Filler Heads, e.g., thing, way,  and place. 

Discourse function of relative clause

            This is perhaps the most difficult part of the analysis.  To my knowledge, no one has developed even a simple taxonomy of the different discourse functions of relative clauses in expository discourse.  There have been some characterizations of relative clause function with respect to information flow and the presentation of new versus old information (Fox and Thompson 1990).  There has also been some work done on the functions of relative clauses in narrative discourse (Hwang 1990), but unfortunately many of these categories do not apply to expository discourse.  In addition, our database is composed of the IL of L2 students, and it is very likely that their use of relative clauses does not reflect native usage.  No one has, as far as I know, attempted a characterization of the discourse functions of relative clauses in L2 production.  It has, therefore, been one of the goals of this thesis to see if such a characterization could be made and if it is useful in solving the puzzle raised by Schacter’s first observations concerning relative clause use and avoidance.

Next: Findings