Parallel conjunctive relations in EFL.

ABSTRACT

Parallelism is a well-recognized norm in English sentences; Dr. Johnson, Frances Bacon, and Walt Whitman are but few examples of creative writers who distinguished their styles with parallelism. Because if gives texts an effect of balance, disrupted parallelism spoils writer--reader interaction, disorienting the reader and hampering the writer's self-expression. Parallelism creates redundancy that is essential to textual predictability, an element responsible for reader's facilitation of information mental processing. This paper explores the theoretical concept of parallelism and presents all empirical investigation of EFL learners' use of this phenomenon. It surveys faulty parallel structures, classifies them into categories, explores their prevalence, and investigates the degree of difficulty that each poses. The experiment concludes that the degree of accuracy in learners' use of parallelism does indirectly reflect proficiency and that parallelism categories are ranked in ascending order of difficulty as follows: verb phrases, adverbs, noun phrases, adjectives, correlative conjunctions, clausal, and comparative structures.

0. Introduction

One of the main defects in sentence structure that may disorient the reader, hamper self-expression, and disrupt writer--reader interaction is lack of parallelism. Sinclair (1981) is of the view that written and spoken language alike tan be described at one and the same time at interactive and autonomous planes of discourse, where the interactive is concerned with negotiations between discourse participants, whilst the autonomous is concerned with the recording of experience. Widdowson (1979) demonstrates how a written monologue is a covert dialogue between a writer and reader; he establishes that the writer assumes the roles of both interlocutors. For EFL learners to accomplish the goals of self-expression and writer--reader interaction, they need to conform to the norms and structures of English sentences, paragraphs, and essays; parallelism is a well-recognized norm that English discourse tends to conform to. It gives text an effect of balance. English literature abounds in parallel structures (e.g., the Psalms and the writings of Dr. Johnson, Frances Bacon, and Walt Whitman).

EFL nonparallel structures often result from the teaching practice where foreign language learners are encouraged to vary sentence length and to increase sentence complexity in an effort to reduce the degree of foreignness in their writing. They, consequently, try to combine nouns, verbs, phrases, and clauses in one series, in one single sentence. Such combinations more often than not, render their sentences unacceptable due to violation of syntactic parallelism, a norm in English writing. Lack of parallelism results in the creation of unnecessary information processing obstacles which often cause reader disorientation. To illustrate this point, let us consider the following sentences that have been written by Arab EFL university students:

1) Nowadays we can see women doctors, teachers, nurses and in many other professions.

2) The place should be green, beautiful and it has attractive views.

3) There are problems facing old people such as disease, being accepted by young generations, and how we are supposed to solve them.

4) Old people made very good things for us, so we must reward them by taking care of them and didn't leave them alone.

These examples reveal an attempt at expanding sentence length and increasing structural complexity. The resulting sentences are erroneous, primarily because of lack of structural parallelism. In (1), for instance, the student conjoined a prepositional phrase to three noun phrases. In (2), two adjectives are joined to a clause. The conjoined items in (3) are of three different grammatical categories, a noun, a participial phrase, and a clause. The conjoined items in (4) are also non-congruent in category; one is a participial phrase, whilst the other is a verb phrase. All these examples lack the balance and predictability that parallelism creates.

It is the aim of this paper to present an experimental study that (a) surveys faulty parallel structures in the writing of some Arab EFL learners, (b) investigates the prevalence of such problematic structures at two distinct levels of competence, and (c) rank-orders, in terms of difficulty, the various types of parallel structures. The ultimate goal is to formulate recommendations for the teaching of EFL to Arab learners. Before commencing, however, let us define the central topic, parallelism.

1. Parallelism in linguistics and TEFL 

Quirk et al. (1973) hold that parallelism involves the use of identical forms in successive clauses or sentences to convey a certain rhetorical effect, or to create cohesion in texts. Memering and O'Hare (1984) maintain the same view when they assert that parallelism is a form of repetition. Similarly, de Beaugrande (1984: 170) defines parallelism as a process involving "filling the same surface patterns with different patterns, or re-using a surface format with different components". This implies that the form is repeated, but with new content each time. De Beaugrande also maintains that the use of parallelism creates a type of redundancy that is not in concepts nor in expressions, but in grammatical structure. In discussing the formal links that create text cohesion, Cook (1989: 15) views parallelism as a cohesive link and defines it as "a device which suggests a connection, simply because the form of one sentence or clause repeats the form of another".

Parallelism is replication of grammatical form and alteration of content with each repetition aimed at the creation of textual cohesion and redundancy, two factors that lead to an element of predictability in texts and improve comprehensibility.

Parallelism may be syntactic or textual. Syntactic parallelism may be at word, phrase, or clause level and it may often be marked by the use of coordinating conjunctions. Textual parallelism, on the other hand, transcends the sentence. Tadros (1985) mentions enumeration as a prediction category that involves textual parallelism, a category which often results from statements such as "there are four types of ...", "the following are ...", "they comprise ...", "the main factors were ...", etc.

Parallelism varies in prevalence across text genres. Some texts rely more heavily than others on parallelism although no text is likely to be free from it. Parallelism has been found by Mu (2001) to be, for example, a distinguishing feature of legal English.

At the heart of the modern concept of parallelism is de Saussure's notion that language is structured paradigmatically and syntagmatically (de Saussure 1921). Every linguistic unit operates at two levels: it is sequentially structured with other items in its immediate context; i.e. it is syntagmatically related to other linguistic units, and hence it has "conjunctive relations" with them. On the other level, this linguistic unit is paradigmatically related to other units that hold a "commutative", substitutive relationship with it, units that can occupy the same spot in the syntactic structure. This dual structuring of linguistic units, de Saussure (1921) asserts, corresponds to "deux formes de notre activite mentale", human mental activity. Most 20th century linguistic theory subscribes to the paradigmatic-syntagmatic notion of structuring, which is clearly relevant to the concept of parallelism.

Influenced by the interplay between the syntagmatic and paradigmatic levels of linguistic expression, Jackobson (1960) carried this notion further when he demonstrated how poetry is distinguished by the placing of paradigms on syntagms to achieve periodicity and how periodicity is present in all discourses as a structuring device. For him, this underlying parallelism is at the heart of any linguistic system. He views syntactic parallelism as that which enables the addressee to place one linguistic unit in a semantic equivalence or opposition relationship with another unit within the discourse.

Parallelism is also used for rhetorical purposes. In English stylistics, deviation from the plainest expression of meaning and structure is considered to be a figure of speech; hence, the artful departure from parallelism, from the norm in syntax, creates a marked structure which is often credited with special meaning. As Jakobson demonstrated the relationship between parallelism and poetic form, several linguists observed the relationship between syntactic parallelism and ritual language across a large range of cultures. The Biblical psalms and Quranic Makkan suras are the foremost examples that come to mind, but so are Indonesian Rotinese religious chants, American Indian Tzotzilian prayers, Mexican Chamulan songs, and Panamanian Cuna curing chants (see Bauman and Scherzer 1974).

Structural parallelism is found to facilitate the mental processing of linguistic utterances. Frazier et al. (1984) carried out some cognitive experiments and established that reading rime for the second clause of a conjoined sentence was faster when the clause was structurally similar to the first clause than when the clausal structures differed. They round "parallel structure" effect for several types of structures, including active vs. passive constructions, direct object vs. sentential complement, agent vs. theme, and animate vs. inanimate noun phrases.

Through syntactic repetition, parallelism reflects the logical similarity in meaning between linguistic units. It, thus, improves the clarity and efficiency of language discourse. When two conjoined linguistic units are of functionally different syntactic structures, they result in faulty syntactic parallelism, a phenomenon that often causes lack of transparency and precision in sentences and hampers understanding.

Faulty parallelism is highly recognizable in texts. Barton et al. (1998) analyzed student sentences marked by teachers as awkward and round parallelism to be a leading cause of awkwardness, together with embeddedness, syntax shift, and direct--indirect speech.

Faulty parallelism is characteristic of lower level linguistic competence. Nelson and Murphy (1993) looked into whether low-intermediate ESL students tan identify and discuss areas of needed revision in the writings of other students and they concluded that such students could identify macro-level problems of topic sentences and text organization and development, but were less adept at identifying problematic sentence-level parallel structures.

2. Objectives of the study

This study focuses on sentence-level structural parallelism. It attempts to answer the following questions:

* Do learners exhibit differences in the use of parallelism because of differences in linguistic competence?

* Is there a significant difference between learners' ability to recognize faulty parallelism and their ability to produce sound parallel structures?

* Do students of the same level of competence exhibit any skill difference between passive and productive knowledge of parallelism?

* How do the various categories of structural parallelism tank in terms of difficulty?

* Over the four years of B.A. study, how much progress do students make in each category of parallelism?

Answers to these questions are hoped to promote awareness of some important difficulties that EFL learners encounter in writing English. Once these problems are identified, they can be handled more efficiently by teachers and textbook writers alike. The identification of errors and resolution of difficulties lend support to the trend in language pedagogy that views writing as a process rather than a product. Current teaching methodology suggests the need for retracing a writer's steps in order to gain insight into their areas of difficulty.

3. Subjects

The subjects of this study are 84 students studying English at the University of Sharjah. They constitute the entire populations of the freshman and senior level language skill courses. Fourteen papers have been discarded because they would have invalidated the results; some were submitted blank or incomplete, others were done by non-freshmen or non-senior students. Therefore, the final number of students was 70, divided equally between freshmen and seniors.

4. Experimental procedure

The experimental data for this study were elicited through a test, especially designed to measure students' ability to recognize faulty parallel structures and to correct them. Grammar constituted the primary lotus of this test; it consisted of seven types of faulty parallel structures represented by the total number of 42 test items. For each category of parallel structures, there were six tokens: five erroneous test items and one grammatically correct distracting item. The six sentences on each category were hot presented as a block but rather randomly scattered in the test. The subjects were asked to underline the part of sentence that rendered the test item grammatically unacceptable. This might have been a word, a phrase, or a clause. The test duration was limited to one hour.

The test comprised the parallelism categories presented in Table 1 below. Each category is illustrated by one example, with the faulty parallel structure being underlined.

When the repeated syntactic pattern consists of verbs, the phenomenon is labeled here Verb Phrase Parallelism. In (1), for example, all the verbs are correct if each is considered in isolation. The unacceptability of this example stems primarily from the distortion of balance created by the fourth verb in the series; "swims", "plays", "jogs" are all simple present tense verbs, whilst "playing" is a present participle phrase.

If the series of repeated structural patterns involves nouns of the same form, then that phenomenon is noun phrase parallelism. The example in (2) depicts faulty noun phrase parallelism which is caused by the incongruity that the present participial phrase "abusing children" brought about when contrasted with the abstract nouns, "reform" and "waste".

Series of adjectives or adverbs are repetitions of a single grammatical categories; they, therefore, constitute what is termed Adjective Parallelism in one case and Adverb Parallelism in the other. (3), above, illustrates how the verb phrase "can be counted on" breaks the parallel structure that was created by the single word adjectives, "friendly" and "dedicated". On the other hand, (4) demonstrates how a prepositional phrase that could have, otherwise, functioned as an adverbial modifier distorted the pattern that was formed by the single word adverbs, "rhythmically" and "beautifully".

Correlative conjunctions create Correlative Parallelism by the sheer fact that these conjunctions always consist of two parts, each introduces one member in a pair of linguistic units; the units conjoined this way forma correlative parallel structure. In (5), the conjunction "either ... or" is a correlative conjunction; therefore, faulty parallelism resulted when the conjoined units were the incongruous verb phrase, "can enroll in English 101", and the noun phrase, "English 102".

Clause Parallelism is when a sentence conjoins two or more clauses of comparable structures. The example in (6) illustrates faulty parallelism that resulted from the lack of similarity between the conjoined subordinate structures, "because she worked hard" and "because of her intelligence". The first consists of Subordinator + NP + VP, whilst the second consists of Subordinator + PP.

The structural patterning entailed by comparative adjectives of equivalence constitute a type of Comparison Parallelism. Expressions of equality of the structure "N + as + Adj + as + N" form fixed parallel patterns that allow no structural variation. Hence, the example in (7) violates parallelism because "than" substituted for "as". The intrusion of the parenthetical phrase, "if not cleverer", impeded the parallelism entailed by the expression of equality as well.

5. Data analysis

Test papers were marked and student errors identified and tabulated. The principles followed in the classification of errors constituted the primary variables in the study. These were as follows:

* Students' level of language competence as reflected in their academic status at university: freshmen vs. seniors.

* Nature of error: recognition vs. production errors.

* Category of structural parallelism: Verb Phrasal, Noun Phrasal, Adjectival, Adverbial, Correlative Conjunction, Clausal, and Comparative.

To gauge any discrepancy between the competence and performance of the two groups of subjects, it is necessary to establish the significance of differences between their ability to recognize faulty parallelism and their ability to produce correct parallel structures. Recognition is symptomatic of linguistic competence, whilst production is indicative of performance. Therefore, a recognition score was calculated per parallelism category for each of the experimental groups, and so was a production score. These scores were simply sums of the individual scores attained by group members per parallel structure category as round in Table 2 below. Then the Chi-square was applied to the lists to test whether the differences between the various sets of scores were substantial enough to constitute statistical significance.

When freshmen and seniors are compared with regard to their ability to recognize faulty parallelism, they stand apart. Chi-square shows a highly significant difference between the two groups' recognition marks [[chi square] (6) = 241.24, p<0.0001], thus indicating that the mark differences between them are not attributable to accident but rather to language training.

The two groups are also found to be clearly distinct in their ability to rectify faulty parallel structures [[chi square] (6) = 209.821, p<0.0001]. They portray different abilities in producing parallel structures.

The sums of category marks for the two experimental groups, therefore, testify that they are distinctly different in both performance (production of accurate parallelism) and competence (recognition of faulty parallel structures).

Looking into the differences in passive and productive knowledge of parallelism that each group exhibited, we round that there is no significant difference between the freshmen's passive ability to recognize faulty parallelism and their ability to correct it. The freshmen showed compatible recognition/production abilities, with the Chi-square value being [[chi square] (6) = 4.075, p<0.6667], thereby indicating non-significant difference between their recognition and production scores. Senior students, on the other hand, exhibited significant difference between their passive knowledge of recognizing parallelism and their ability to produce correct parallel structures [[chi square] (6) = 18.437, p<0.0052]. Seniors appear to have known more about structural parallelism than they were prepared to demonstrate. Freshmen, on the other hand, had less theoretical knowledge but were better able for their level to practice the little that they knew.

To find out whether the various categories of structural parallelism are similar in difficulty, we assumed that the marks attained by the subjects were inversely related to difficulty; the higher the mark, the less difficult the category is. The sums of points attained by each group of students in the various categories were then converted into percentage points and rank-ordered, thus depicting the groups' levels of mastery in each category. The clustered columns in Figure 1 below show the categories' values ranked-ordered by difficulty.

The Spearman Rank Correlation Coefficient was calculated for the percentage marks that the two groups achieved in recognizing and in producing each type of structure. The results show that the two groups had exactly the same level of difficulty in recognizing these structures, [rs(7) = 1.0], with verb phrase parallelism being the easiest to recognize followed by adverb, noun phrase, adjective, correlative conjunction, clause, and then comparison being the hardest to recognize. In terms of ability to produce the various types of parallel structures, the two groups had similar, though un-identical, levels of difficulty [rs(7)=0.964], with correlative rather than clause parallelism being more difficult for freshmen than for senior students.

On the basis of this finding, one wonders whether this order of difficulty is generalizable to the larger population of EFL learners.

To assess the magnitude of improvement in the use of parallel structures that students make over their entire B.A. program, assume that the freshmen group would perform in four years rime the same way as the present experimental senior group. Their improvement may be indexed, then, by the difference between freshmen and senior percentages per category of parallelism. The degree of improvement made by the present experimental groups was gauged in the same method. The improvement attained in the recognition of parallel verb phrases, for example, was calculated by deducting the percentage that the freshman group attained in recognizing VP (i.e. Y1(R) = 60%) from the percentage that the senior group attained in the same category (i.e. Y4(R) = 91%), thus giving the difference of 31 percentage points. Table 3 shows, for each of the categories, the magnitude of improvement in percentage points that freshmen would make by the time they have become seniors and quantifies the two groups' recognition--production differences in performance in the various parallelism categories.

Students appear to make significant improvement as they progress in their B.A. They make the most improvement in noun phrase parallelism, not only in terms of recognition but also production, followed by parallelism in adjectives, adverbs, and verb phrases. They make the least improvement in recognizing and correcting comparative parallelism. Parallelism expressed in correlative conjunctions and clauses appear to be the second hardest to learn.

Both freshman and senior groups perform better on recognizing parallelism than on producing it. (See 3rd and 4th columns in Table 3). First year students, however, have a slightly better ability in recognizing than in producing the various categories of parallelism, with the discrepancy between recognition and production being the biggest for verb phrase parallelism. Fourth year students, on the other hand, have a greater discrepancy between the two abilities; there is an 11 percentage point difference between their abilities to recognize and to produce correlative parallelism, for example. It is evidently clear that fourth year students find it easier to recognize than to produce parallelism, whilst first year students' recognition and production abilities are more compatible. This by no means implies that freshmen are better than seniors at parallelism. It simply indicates that when advanced EFL students demonstrate a larger repertoire of parallel structures, their abilities to recognize and to produce such structures are not identical.

6. Conclusions and implications for TEFL

Because syntactic parallelism creates redundancy that leads to textual predictability which in turn results in the facilitation of readers' information mental processing, and because parallelism is a useful discourse structuring device that creates textual cohesion, it is one important aspect of text development that EFL learners need to know and manipulate well. EFL teachers and textbook developers need to also be aware of the type of problems learners encounter with parallelism and to know which categories pose the most trouble.

If the studied sample were representative of EFL learners, which is a gross assumption to make, or of the Arabs amongst them, then this study may be taken to make the following conclusions:

* The degree of accuracy in the use of parallel structures by EFL learners may indirectly reflect their language competence; the higher the degree of accuracy in structural parallelism, the higher the learner's overall level of language proficiency.

* Disparity between EFL learners' passive and active linguistic knowledge is also manifested in the use of structural parallelism. Learners' ability to recognize parallel structures and their ability to produce accurate parallelism maybe symptomatic of this passive-active knowledge disparity.

* Less proficient EFL learners tend not to exhibit much difference in performance between recognizing and producing parallel structures. Advanced learners, on the other hand, demonstrate better command of parallelism but more contrast between the abilities to recognize and to produce parallelism.

* The various categories of structural parallelism appear to pose the same level of difficulty for less proficient and advanced learners alike. The difficulty being least with verb phrase parallelism and increasing with adverbial, noun phrasal, adjectival, correlative, clausal, and comparative.

The most important implications of these findings for TEFL are three. First, cumulative learning and lengthy exposure to the foreign language are useful in mastering aspects of the language but they do not necessarily lead to full compatibility between passive and productive language skills. It is inevitable that learners would know more than they are capable of demonstrating. In fact, the more competence they develop, the more divergence there is between their passive and productive skills. Second, the easier to learn categories of parallelism should be taught first, then the other categories introduced gradually according to their levels of difficulty, with correlative, clausal, and comparative parallel structures being left until last. Such a strategy would guarantee that students have mastered the concept of parallelism before the complex types are introduced. Third, assessment and evaluation must respect the gradual and ongoing nature of language development.

Since writing is a complex process, EFL learners need to progress in it through a number of possibly non-linear levels to experience it. EFL teachers would achieve better results if they were to focus upon exploring and understanding what their students actually do throughout the writing task, appreciating the difficulties their students have with parallel structures. They need to guide them through the stages of writing and help them to understand writing as a communication process, alerting them to the value of the balance and predictability that parallelism creates in texts. By observing the norms of writer-reader interaction, EFL learners will tome to realize that parallelism is actually used to facilitate the comprehensibility of their written discourse.

Table 1. Categories of faulty parallel structures

Ser.   Faulty parallel                Example
          structure

1)     Verb Phrase       In summer, John swims, plays tennis,
                         jogs, and playing football.
2)     Noun Phrase       Prison reform, abusing children, and
                         toxic waste are three issues which
                         concern citizens nowadays.
3)     Adjective         The new employee is friendly,
                         dedicated, and can be counted on.
4)     Adverb            The dancer moved with grace,
                         rhythmically, and beautifully.
5)     Correlative       A student either can enroll in English
       Conjunction       101 or English 102.
6)     Clause            She succeeded in the final exam
                         because she worked hard and because
                         of her intelligence.
7)     Comparison        Mary is as clever, (if not cleverer),
                         than John.

Table 2. Freshmen and senior group marks per parallelism category

Type of       FreshmenR *    FreshmenP **    SeniorR *    SeniorP **
parallelism

VP                126            116            191          176
Adverb             85             81            159          152
NP                 67             61            153          143
Adj                63             59            142          138
Correlative        46             40            109           86
Clausal            42             38            104           90
Comparison         35             30             72           52

* (R) stands for "recognition", ** (P) stands for "production".

Table 3. Improvement in performance

Category         Sen(R)--      Sen(P)--       Fresh(R)--     Sen(R)--
                 Fresh(R) *    Fresh(P) **    Fresh(P)       Sen(P)

VP                  31             29             5             7
Adverb              36             33             1             4
NP                  41             39             3             5
Adjective           38             38             2             2
Correlative         30             22             3            11
Clause              30             25             2             7
Comparison          17             11             3             9

* (R) stands for "recognition", ** (P) stands for "production".

REFERENCES

Barton, Ellen--Ellen Halter--Nancy McGee--Lisa McNeilley

1998 "The awkward problem of awkward sentences", Written Communication 15/1: 69-98.

Bauman, Richard--Joel Scherzer

1974 Explorations in the ethnography of speaking. Cambridge: Cambridge University Press.

Cook, Guy

1989 Discourse. New York: Oxford University Press.

De Beaugrande, Robert

1984 Text production: Toward a science of composition. (Advances in Discourse Processes Ser. 11). Westport, USA: Greenwood Publishing Group.

de Saussure, Ferdinand

1921 Cours de linguistique generale. Paris: Payot.

Frazier, Lyn--Lori Taft--Tom Roeper--Charles Clifton--Kate Ehrlich

1984 "Parallel structure: A source of facilitation in sentence comprehension", Memory and Cognition 12/5: 421-430.

Jackobson, Roman

1960 "Concluding statement: Linguistics and poetics", in: Thoman A. Sebeok (ed.), 353-377.

Memering, W. Dean--Frank O'Hare

1984 The writer's work. A guide to effective composition. (2nd edition). Paramus, USA: Prentice Hall PTR.

Mu, Kejuan

2001 "On the lexical and syntactic characteristics of legal English", Fujian Waiyu: Foreign Languages in Fujian 1: 19-22.

Nelson, Gayle L.--John M. Murphy

1993 "Writing groups and the less proficient ESL student", TESOL Journal 2/2: 23-26.

Quirk, Randolph--Sidney Greenbaum--Geoffrey Leech--Jan Svartvik

1973 A university grammar of English. London: Longman.

Rizvi, S. N. A. (ed.)

1981 The two-fold voice: Essays in honour of Ramesh Mohan. Salzburg: University of Salzburg.

Sebeok, Thoman A. (ed.)

1960 Style in language. Cambridge, Mass.: MIT Press.

Sinclair, John McH

1981 "Planes of discourse", in: S. N. A. Rizvi (ed.), 70-89.

Tadros, Angela

1985 Prediction in text. (Discourse Analysis Monograph 10). Birmingham: University of Birmingham Printing Service.

Widdowson, H. G.

1979 Explorations in applied linguistics one. New York: Oxford University Press.

SANE M. YAGI--SHEHDEH FAREH

University of Sharjah, United Arab Emirates