國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
suitable methodology for the analysis and monitoring of such transfers which hitherto has been lacking. (HJ1-19131)
d. As we did not perform endoscopy in this control group, we can not rule out the inclusion of a few patients with gastritis. (HU4-6279)
e. To identify a cause of difference as genetic or environmental in origin, although it is a clear dichotomy, does not save the original Lorenzian dichotomy. (GU8-467)
The examples above demonstrate three possible sites for a postmodifier to act on.
First, there are two ambiguous cases, (5.24a) and (5.24c), where the postmodifiers could be considered to act either on N2 or on the entire of -construction. In contrast, the entire of -construction in (5.24e) appears to be postmodified. While (5.24b) is a nominal derived from the prepositional verb apply to, the prepositional to phrase does not really perform the normal function of a postmodifier. In (5.24d), the prepositional with phrase only modifies N2 rather than the entire of -construction. Although postmodifiers are not as confined as premodifiers which is more limited in space and type of phrases, ambiguity could also arise from the structure. One question arising from the discussion here was whether the two types of modification would affect each other. To put it more specifically, would premodification reduce the chance of postmodification or vice versa? Would the chance of premodification at N2 be reduced in the presence of postmodification? To answer these questions, we turned to HCFA to determine the statistically significant configurations for discussion in the next section.
5.3 HCFA: Premodification and Postmodification
Four factors were subjected to an HCFA analysis, namely, premodification at N1 (premod N1), premodification at N2 (premod N2), postmodification (postmod), and syntactic position of the of -construction (syntactic position). For the first three factors, there are only two levels, Y stands for an occurrence of modification and N stands for no occurrence of modification. The final factor, syntactic position, includes eight levels notional and non-existential subject, direct and prepositional object, complement of preposition, subject and object complement, and others. Significant HCFA results with a p-value at least smaller than 0.05 and the effect size greater than
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
0.01 were obtained and presented in Table 5.7. There were 25 significant configurations, where 12 of them represent the cases whose observed frequency is higher than the expected and 13 were those whose observed frequency is lower than the expected. As shown by the configurations for single levels without interaction with other factors, modification is not the norm. In fact, the majority of instances are unmodified. Three syntactic positions, complement of preposition, object, and non-existential subject, dominate other positions, as they are significantly less in frequency. From the two-way interactions, as shown by postmod(Y) interacting with syntactic(object) and postmod(N) interacting with syntactic(sub_nonex), we can interpret that when the of -construction serves as an object, it is very likely to be postmodified and when the of -construction takes up the subject position, it is less likely to be postmodified. However, the situation with the of -construction as object complicates further with a four-way interaction: premodN1(N) X premodN2(N) X postmod(Y) X syntactic(object). This situation is when an of -construction as object tends not to be premodified at all but is postmodified.
‧ 國 立
政 治 大
‧
學tio Na
l C na e h
g n
c h i U n
iv ers i ty
Table 5.7 Significant HCFA configurations for premodification and postmodification
premodN1 premodN2 Postmod syntactic Freq Exp Cont.chisq P.adj.Holm Q Obs-exp Dec
N . . . 3381 2440.5 362.4422 9.30E-164 0.385 > ***
. N . . 3022 2440.5 138.5545 8.01E-63 0.238 > ***
. . N . 3149 2440.5 205.6842 9.38E-93 0.29 > ***
. . . comp_prep 1497 610.125 1289.1576 4.45E-243 0.208 > ***
. . . object 1085 610.125 369.6067 1.57E-78 0.111 > ***
. . . sub_nonex 1040 610.125 302.8765 1.43E-65 0.101 > ***
. . Y object 481 385.0072 23.9336 7.51E-06 0.021 > ***
. . N sub_nonex 738 670.9609 6.6982 0.03441227 0.016 > *
. N Y object 306 238.3716 19.1869 0.00028636 0.015 > ***
N . Y object 332 266.689 15.9944 0.0012159 0.014 > **
N N Y object 217 165.1166 16.3029 0.0030419 0.011 > **
Y . N sub_nonex 263 206.1957 15.6489 0.00169991 0.012 > **
Y . . . 1500 2440.5 362.4422 9.30E-164 0.385 < ***
. Y . . 1859 2440.5 138.5545 8.01E-63 0.238 < ***
. . Y . 1732 2440.5 205.6842 9.38E-93 0.29 < ***
. . . object_comp 4 610.125 602.1512 6.91E-273 0.142 < ***
. . . sub_exist 65 610.125 487.0498 3.03E-189 0.128 < ***
. . . other 273 610.125 186.2787 2.55E-58 0.079 < ***
. . . obj_prep 404 610.125 69.6374 5.41E-21 0.048 < ***
. . . comp_sub 513 610.125 15.4612 9.42E-06 0.023 < ***
. . N object 604 699.9928 13.1639 0.00053132 0.023 < ***
. . Y sub_nonex 302 369.0391 12.1782 0.00149417 0.015 < **
. N N object 365 433.3904 10.7922 0.00733034 0.015 < **
N . N object 383 484.8752 21.4046 8.70E-06 0.023 < ***
N N N object 235 300.2034 14.162 0.00212703 0.014 < **
139
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
In sum, the HCFA results helped us to elucidate the interaction between premodification and postmodification at different syntactic positions. Two situations were identified: first, at the object position, the of -construction tends to be postmodified but not premodified; second, at the subject position, the of -construction tends to be premodified at N1 but not postmodified at all.
5.4 Discussion
To answer the second research question on what functions of -constructions have in academic writing, we examined the of -constructions going beyond the construction level, i.e., to investigate the types of clauses and syntactic positions where the of -constructions are commonly found. The results indicate that the of -constructions mostly appear in the complement of preposition (29.09%), followed by direct object (28.25%) and subject (22.00%). In the complement of preposition, the preposition in was found to head the prepositional phrases containing an of -construction. The reason for such a high proportion of the in prepositional phrases could be due to the fact that in is the second highest preposition in English. The complement of preposition was found to be used as an apposition (5.25a) or an adverbial of circumstance (5.25b).
(5.25) a. Lawton (1973), in his analysis of Britain’s nineteenth-century population, observed that... (FB2-318)
b. Inhibition of viral replication in chronic hepatitis B is usually accompanied by a transient rise in the activities of aminotransferases.
(HJ2-4058)
It was also found that a copular be construction containing an of -construction could be effectively used to express a writer’s stance. First, the of -constructions may appear in the subject position in a copular be construction (5.26a) where it is evaluated in the object position. Second, attributive adjectives functioning as premodifiers for N1 or N2 could be used as shown in (5.26b). Lastly, deadjectival nouns could be chosen to express evaluation in a more subtle way in (5.26c).
(5.26) a. Evaluation of the cost effectiveness of drug treatment is in its infancy…
(EC7-886)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
b. …one important function of research is to indicate what is not relevant,
… (CN6-536)
c. The beauty of such methods is that they can never be proved wrong since, ... (CS0-1371)
In addition to the use of evaluative adjectives as premodifiers in the copular be construction, this type of adjectives are also found in other clausal structures. The evaluative adjectives can be found either before N1 (5.27a) or N2 (5.27b).
Interestingly, there are more evaluative adjectives preceding N1 than N2. The same disproportion can also be applied to evaluative nouns serving either as an N1 or N2 head.
(5.27) a. No matter whether the stochastic or a deterministic form is used, the successful use of the model will depend on two factors,... (HR3-1569) b. In the meantime, we will conclude by drawing attention to a number of
serious weaknesses in the strategy on which it is based. (FBC-1441)
There are far more evaluative N1 heads as compared to N2 heads. Examples (5.28a) and (5.28b) illustrate each. Other evaluative nouns identified from the corpus data include both positive and negative evaluations. Negative ones (e.g., misuse, neglect, suspicion, trap, villain, irrelevance, failure, monopoly, barbarization, condemnation, racism, disadvantage, incompetence, eccentricity, dismissal, dishonesty) tend to predominate. In contrast, only a handful of positively evaluated or neutrally adopted N1 heads were found including nouns like significance, importance, glamour, adequacy, and frivolity. Moreover, a few of -constructions display a heavy-duty evaluative function with both evaluative premodifiers and nominals as exemplified by (5.28c-e).
(5.28) a. These characteristics clearly add extra dimensions to MDC ’s urban regeneration strategy, raise central questions about local accountability and highlight the thorny problem of uneven development. (BN8-699) b. The significance of these beliefs in creating a commonsense culture of
taken-for-granted racism in Britain is difficult to underestimate, ...
(CLW-656)
c. It is a well-established form of parasitism, resting on bits of biblical dogma such as ”the poor always ye have with you” (John, xii.8). (FPJ-1638)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
d. In Birmingham there was a virtually total neglect of anti-semitism.
(CS6-1049)
e. If there are any cases in which the victim of unfair pressure may avoid a contract even though the threat is not of unlawful action, there seems no possibility of any concurrent tort liability. (FSS-1247)
Going beyond the construction, the most important functions of the of -constructions involve modification patterns. As presented in Chapter 5, when the of -construction acts as the subject, it tends to be premodified but not postmodified.
When the of -construction acts as the object, it is more likely to be postmodified but not premodified. The premodifiers have been shown to be an important site for evaluative purposes and it is shown here that the premodifiers at N1 are crucial at the subject position. Another important discourse function of the of -construction at this position is to encapsulate previous propositions and, thereby, allowing the reader-writer dialog continue to flow. This function is useful for academic writers and may serve as a cohesive device that links the preceding proposition with the succeeding one.
5.5 Chapter Summary
This chapter investigates the of -constructions in a broader context by observing how it functions at different syntactic positions in various clause types. First, some of -constructions were found to belong to what Levin (1993) refers to as ”oblique”
subject alternations. This type of structure minimizes the appearance of an agentive subject and allows for a tighter structure, both of which conform to the characteristics of academic writing. In terms of the of -construction as the object, it was found that there is a high proportion of objective of -constructions, followed by attributive and measurement of -constructions. When the of -construction acts as a complement, it either functions as a complement of preposition to add additional information such as specifying location or time or functions as a topic-framing device fronted in a sentence.
Regarding the of -constructions found in the subject position of a passive construction, they tend to be evaluative either found in the premodifier or nominal group. We also employed a multivariate statistical analysis (HCFA) to determine how premodification
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
may interact with postmodification. It was found that when the of -construction acts as the subject, it tends to be premodified but not postmodified. When the of -construction is at the object position, it is more likely to be postmodified but not premodified. It is suspected that such a modification pattern is oriented towards the reader to facilitate reading comprehension. Overall, the results of this chapter show that of -constructions carry various functions in conformity with their syntactic positions.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
CHAPTER 6
CONCLUSIONS AND PEDAGOGICAL IMPLICATIONS 6.1 Summary of Findings
This study sets out with the aim of assessing the actual use and functions of the of constructions in academic writing. The most important means of characterizing the various types of the of -constructions was found to be semantic relations. The of -constructions were examined in two levels, first at the local construction level and then at the broader sentential context. At the construction level, we used the correspondence analysis to examining the 14 subcategories with semantic categories classified from WordNet. There were only three significant configurations identified between N1 and N2 (N1-‘animate’ + N2-‘spatio-temporal’, N1-‘spatio-temporal’ + N2-‘group’, and N1-‘technical/concrete’ + N2-‘technical/concrete’), indicating a limited use of semantic categories to consider of -constructions. A second approach to the of -construction was a corpus-based categorization of semantic relations between the two nominal groups in the of -construction. A total of four major categories, namely, ‘part-whole’, ‘quality’, ‘action’, and ‘derivative’, were identified along with 17 subcategories. It was found that the ‘objective’ relation (e.g., acquisition of an ability) under ‘action’ predominates in academic writing, followed by the ‘attributive’
relation (e.g., the accuracy of models) and ‘topic‘ relation (e.g., three images of the gall bladder) under ‘quality’. Altogether, these results illustrate the importance of using semantic relations to capture the meanings of of. We also looked at the of -construction beyond the construction level in different types of clauses. First, for the simple subject-verb-object constructions we found that syntactic positions may have an effect on the writer‘s choice of premodification or postmodification. Two modification patterns involving both premodifiers before N1 and N2 and postmodification following N2 emerge. It was found that when the of -construction occupies the object position, postmodification is more likely to take place when both N1 and N2 are not premodified. Another modification pattern was when the of -construction occupies the subject position, its N1 is more likely to be premodified
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
without any postmodification. Second, copular be constructions were found to be used with the of -constructions to carry out characterizing, identifying, and evaluative functions. Three important functions of the of -construction were identified. The first appears to be a hedging function by avoiding the use of an agent, which is referred to as agent demotion. The of -construction itself functions as the subject to avoid the use of an animate entity. Another function of the of -construction is encapsulation (Sinclair, 1993) by packaging information into a clausal element which plays a role in maintaining textual cohesion. Halliday and Martin (1993) also point out that this is an important function for academic writers to build new knowledge on top of the old, taken-for-granted ones. Lastly, the of -construction can be used to express evaluation in at least three ways. These include using the nominal group at N1 (e.g., problem, significance), premodifiers preceding N1 or N2 (e.g., important, impossible) and both (e.g., a thorny problem). In the next section, we would like to demonstrate how different types of semantic relations associated with the of -construction can be applied to the analysis of sections in a research article.
6.2 Application to Pedagogy
Given that we employ an academic corpus for the present study, our findings will be applicable to teachers and learners in the field of English for Academic Purposes (EAP). Unfortunately, unlike previous findings on lexical bundles or phraseological approach to academic writing where a list of the most frequently used of -constructions can be immediately used for teachers as material for curriculum design and for learners as a potential learning aid, the huge amount of possible combinations of the of -constructions seems to be underrepresented by any list of less than a hundred or so.
One solution to compensate for this difficulty of transforming research results into teachable materials was inspired by our covarying collexeme analysis results.
Observation of the CCA results reveals many discipline-specific pairs such as breach of contract and Court of Appeal. These pairs could not be adequately explored as they reflect disciplinary variations. For the benefit of teaching academic writing, a more
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
specific approach could also be useful. Previous research also has shown the benefits of compiling a specialized corpus for teaching academic writing (e.g., Flowerdew, 2004, 2008; Tribble, 2002). The next section, Section 6.2.1, will present the use of a mini corpus constructed from the BNCweb to be applied to raising learner’s awareness of the of -construction. Section 6.2.2 will provide additional examples to serve as instructions to teaching agent demotion in academic writing.
6.2.1 A Mini Corpus of Nucleic Acid Research
While there are varying functions that are potentially possible with the of -construction, a pedagogical application by using specialized corpus data is presented here. The objective of this analysis is to see how of -constructions are distributed in various sections of empirical research articles. A mini corpus of ten research articles all of which come from the journal Nucleic Acid Research was selected from BNCweb involving a few steps from the main menu to select one of the user-specific functions to make a subcorpus. In this case, three text files (FTB, K5N, K5R) were selected for a query for the concordance lines containing an of -construction. After downloading the query results from the website, some manual work was necessary to identify the positions of the concordance lines within a research article. The selection criterion for a research article from the corpus was mainly based on its organization where five sections including abstract, introduction, methods, results, and discussion were all present. Each section was recognized by its section heading. An additional section referred to as ‘others’, which contain all the captions for tables and figures was also included. The results generated 1,088 instances of of -constructions. However, only 139 of them has a frequency of more than 2. A covarying collexeme analysis was applied to identify their collocational strengths. A total of 82 collexeme pairs with a collocational strength greater than 2 were analyzed and classified into semantic relations. Table 6.1 presents the results of corpus examples in each section. For example, in the abstract section, there is only one collexeme pair, expression of the variants, that has a frequency of more than 2.
‧
Table 6.1 List of N1-N2 collexeme pairs generated from ten Nucleic Acid Research articles
‧
Table 6.1 – continued from previous page
Section N1 N2
In addition, we can find a highly disciplinary-specific of -constructions from the data shown in Table 6.1. With only one exception, lines of evidence in the discussion section, which is an expression of research discourse possibly used by academic writers across disciplines, the rest of the N1-N2 collexeme pairs are all discipline-specific involving technical nouns such as promoter, genes, mRNA,
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
bacteriophage, exon, and polypeptide. However, a quick query for the of -construction lines of evidence from BNCweb generates 23 hits. The distribution function indicates that the use of this phrase is highly academic with a rate of 14 out 23. The remaining nine instances are found in books. The text domains where lines of evidence can be found are mainly scientific (21 out 23) and two instances found in arts and world affairs. In other words, lines of evidence is more of a scientific academic discourse than the humanities. As discussed earlier in Section 4.1, technical nouns prevail in N2.
N1, on the other hand, involves more action nominals such as analysis, preparation, detection, and attributive nouns such as presence, cooperativity, and mobility.
Table 6.1 also demonstrates variations among different sections. In terms of a quantitative comparison, the abstract section is least likely to find the of -construction as compared to other sections. This section only has one N1-N2 collexeme pair, that is, the expression of variants, which qualifies our criteria for collocational strength and observed frequency.
If we inspect each collexeme pair by means of various types of semantic relations, differences among sections would be apparent. Figure 6.1 provides the breakdown of semantic relations identified for each N1-N2 collexeme pair in all sections except for abstract. From the graph above we can see that the introduction section is predominated by action of -constructions such as sequence analysis of the PPT promoter and transcription of protein-coding genes. Both the methods and results sections, in contrast, have the highest proportion of quality of -constructions. The quality of -constructions include both attributive ones like a complete loss of activity, the mobility of each polypeptide of the expressed proteins and quantitative ones like 22 bases of exon P4 and the low steady-state levels of RNA. The results section has far more attributive of -constructions than the methods section. The discussion section is where we can find the most evenly distributed types of semantic relations with the part-whole of -constructions as the highest. Part-whole of -constructions like the largest subunit of mammalian RNA polymerase II or members of the C/EBP family are commonly used in the discussion section by the academic writers to validate their
‧ 國
立 政 治 大
立 政 治 大