5. Results of Information Structure in Sentence Patterns
5.1 Distribution of Information Status in the Sentence Patterns
The chapter is structured as follows. In section 5.1, the distribution of the information status in the sentence patterns is presented. In section 5.2, the results of the ‘discourse-new’ NP in the sentence patterns are further illustrated. In section 5.3, the results of the ‘discourse-old’ NP in the sentence patterns are introduced. Lastly, in section 5.4, the impact of information structure and construal on the sentence patterns is discussed.
In the analysis of information structure, the sentence patterns in the active transitive and the passive construction were all included (except [NPIN + Verb + NPOUT]), as in Table 5.1. The NP whose information status is investigated is bolded.
5.1 Distribution of Information Status in the Sentence Patterns
Table 5.1 presents the distribution of the information status of the NP in the active transitive construction and the passive construction. The results show that, as for the NPs in the direct object position of the active transitive construction, 80.4% of them are ‘discourse-new’ while only 19.6% of them are ‘discourse-old’.
Table 5.1 The distribution of information status of NP in sentence patterns Sentence Patterns Information
‧
On the other hand, as for the NPs in the subject position of the passive construction, 56.4% of them are ‘discourse-new’ and 43.6% of them are ‘discourse-old’.
These two results suggest that the direct object NP in the active transitive construction strongly prefers ‘discourse-new’ NP and that the ‘discourse-new’ NP preference is less so in the subject position of the passive construction.
From the perspective of the information status in information structure, Table 5.2 demonstrates the preferred position for the ‘discourse-new’ and ‘discourse-old’ NPs in the constructions. On the one hand, when the NPs are ‘discourse-new’, 71.5% of them occur in the direct object of the active transitive construction, which means that the
‘discourse-new’ NPs prefer the post-verbal position in the active transitive
construction rather than the pre-verbal position in the passive construction. On the other hand, when the NPs are ‘discourse-old’, 55.8% of them occur in the pre-verbal position of the passive construction in contrast to 44.2% in the post-verbal position of the active transitive construction. ‘Discourse-old’ NPs show the preference for the pre-verbal position of the passive construction.
Table 5.2 Information status of NP in sentence patterns of SUBSTITUTE Information
Status Sentence Patterns Number Total
Discourse-new
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
The results show that the choice of sentence patterns generally conforms to ‘old-before-new principle’ (Chafe, 1994; Ward & Birner, 2004). When the NP is
‘discourse-new’, the NPs prefer to occur in the post-verbal position, that is, the sentence pattern [NPAGENT + Verb + NPIN/OUT] (You can substitute margarine in the recipe). In contrast, when the NP is ‘discourse-old’, the pre-verbal position is preferred, as in [NPIN/OUT + be + Verb-pp] (Margarine can be substituted in the recipe).
These findings relating to the preferred position of the NP in information structure echo with the proposal of Preferred Argument Structure (Du Bois, 2003).
Specifically, the preferred sentence pattern [NPAGENT + Verb + NPIN/OUT] of the
‘discourse-new’ NP is consistent with Du Bois’ (2003) argument that “the general pattern for two-place predicates is that only one core argument typically carries new information, and this argument is not the A” (p. 38), which means that the direct object tends to carry new information. Our finding that the ‘discourse-new’ NPs prefer the post-verbal position (i.e. direct object) of the active transitive construction is consistent with the preferred argument structure of transitive verbs in Du Bois (2003:38). In general, ‘discourse-new’ NPs prefer the post-verbal position of [NPAGENT + Verb + NPIN/OUT] (You can substitute margarine in the recipe) while
‘discourse-old’ NPs tend to occur in the pre-verbal position of [NPIN/OUT + be + Verb-pp] (Margarine can be substituted in the recipe).
In the next section, we further look into the relation between the preferred role of the NP and the information status of the NP in information structure. Specifically, the two hypotheses of this thesis will be attested. The ‘discourse-new’ NPs were predicted to show the preference for the NPIN while the ‘discourse-old’ NPs showed little
preference to either NPIN or NPOUT.
‧
5.2 ‘Discourse-new’ NP in the Sentence Patterns
Table 5.3 displays the ‘discourse-new’ NPs in different sentence patterns.
Regardless of the constructions, 90.6% of the ‘discourse-new’ are interpreted as NPIN
in the sentence patterns while merely 9.6% of them are NPOUT. The stark contrast between the NPIN and the NPOUT in ‘discourse-new’ NPs suggests that ‘discourse-new’
NPs prefer NPIN rather than NPOUT.
Table 5.3 ‘Discourse-new’ NP and sentence patterns in SUBSTITUTE Information
Status Sentence Patterns Number Total Grand
Total
This result supports our first hypothesis that the ‘discourse-new’ NP tends to be the NPIN. The tendency is anticipated in that using a ‘discourse-new’ entity as the NPIN is more common than the NPOUT to be replaced. To illustrate, in (5.1), the
‘AGENT’ I uses an alternatives thick oil (NPIN) to prevent leakage.
(5.1) I own a 1976 SWB Series III which is leaking oil from the seal round the offside swivel pin housing. As the housing is not pitted I cleaned it and replaced the seal, however it is still leaking some oil. Can I avoid renewing the housing by substituting thick oil to prevent further leakage? (AN2-652)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
However, it is less favorable to interpret thick oil as the NPOUT since replacing new information which just comes to readers/hearers’ mind may be against our live experience or intuition. Therefore, if we replace the oil in (5.1) with the thick oil, as in (5.2), the information status of the direct object NP (thick oil) of SUBSTITUTE becomes ‘discourse-old’ which in turn makes the interpretation of the NP more compatible with NPOUT rather than NPIN.15
(5.2) I own a 1976 SWB Series III which is leaking some thick oil from the seal round the offside swivel pin housing. As the housing is not pitted I cleaned it and replaced the seal, however it is still leaking. Can I avoid renewing the housing by substituting thick oil to prevent further leakage?
The other example of the ‘discourse-new’ in [NPIN/OUT + be + Verb-pp]
(Margarine can be substituted in the recipe) is exemplified in (5.3). The NP (A quality, dry pale sherry) is ‘discourse-new’ as a newly introduced alternative in the recipe, and thus the NP should be the NPIN.
(5.3) [Title: Rice wine] Used extensively for cooking and drinking in China, it is made from glutinous rice, yeast and spring water. A quality, dry pale sherry can be substituted, but cannot equal its unique, rich, mellow taste.
(G2D-750)
The corpus results support our first hypothesis in that the ‘discourse-new’ NP demonstrates the strong preference for the NPIN in natural language use. In the next
15 For here, we recognized that the context information also comes into play and induces our interpretation of the NP (thick oil) to be NP .
‧
section, the second hypothesis relating to relation between the ‘discourse-old’ NP and the role carried by the NP will be attested.
5.3 ‘Discourse-old’ NP in the Sentence Patterns
As for the relation between the ‘discourse-old’ NPs and the role carried by the NP, the second hypothesis that the distribution of NPIN should be on a par with NPOUT
in each sentence pattern is made. The hypothesis rests on the reason that the
previously mentioned or discussed NP should be able to be either NPIN or NPOUT. The less preferred situation of replacing new information demonstrated above is
precluded.
Table 5.3 shows that when the NPs are ‘discourse-old’, 55.8% of them are NPIN
and 44.2% of them belong to NPOUT. The close proportion of NPIN and NPOUT in
‘discourse-old’ NP suggests that ‘discourse-old’ NPs show the neutral preference to either NPIN or NPOUT, and thereby the second hypothesis is attested.
Table 5.3 ‘Discourse-old’ NP and sentence patterns in SUBSTITUTE Information
Status Sentence Patterns Number Total Grand
Total
These findings suggest that while the information status of NP could help identify the role of NP in question, only being ‘discourse-new’ is beneficial in
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
prediction. When the NP is ‘discourse-old’, readers/hearers still need to count on other clues in the context to figure out the role of NP.
The ‘clues in the context’ refers to the clues provided in the context for
readers/hearers to structure the whole situation by means of ‘inference’. To illustrate, in (5.4), the NP (the values) is ‘discourse-old’ due to the anaphoric reference to the marked working voltage which contrasts with the advertised value. In other words, there are two kinds of value: the advertised value in contrast to the value of the
marked working voltage. Then, since the ‘discourse-old’ NP (the values) demonstrates the neutral preference to either NPIN or NPOUT, information status helps little here. We are forced to turn to other resources in the context other than information status.
(5.4) This is the timely advice offered by Colin Pickwick. He recently purchased some capacitors (from a well established supplier) only to find (on close examination) that the marked working voltage was well below the advertised value. It would appear that either the values had been
substituted by the supplier (without warning) or the capacitors had been
mistaken for components having an identical capacitance value but with a much reduced voltage rating. (C91-488 to C91-491)
It is said that the marked value of some capacitors is below than the advertised value, and this may result from the replacement of the initial voltage value. In the normal situation, the initial voltage volume is assumed to be consistent with the advertised value. However, the supplier replaces the values (i.e. the initial voltage value) without warning, and the replacement cause the disparity between the
advertised and the initial value. Following the flow of inference displayed above, we can infer that the NP (the values) is an NPOUT.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
The two hypotheses have been supported by the corpus results in section 5.2 and 5.3. In section 5.2, the first hypothesis relating to the strong preference to the NPIN
when it is ‘discourse-new’ is attested. Then, in section 5.3, the second hypothesis that the ‘discourse-old’ NP shows the neutral preference to either NPIN or NPOUT is attested as well. Since ‘discourse-old’ NP may be less helpful to the prediction of the role carried by the NP, readers/hearers may rely on their ability of inference to figure out the role of the NP by the clues provided in the context.
Then, the next section presents that the impact of information structure is more than predicting the role of NP. Information structure may affect the configuration of the sentence patterns.