國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
CHAPTER 2
THE SEMANTICS OF THE OF-CONSTRUCTION
This chapter will begin with a literature review on two major topics: the word of and the semantic relations between the nominal groups mediated by of. A discussion on previous accounts of the meanings of of will be presented first, and then how semantic relations could work to establish the meanings of the of -construction will be explored. The last section considers the discourse functions of of -construction in academic writing.
2.1 The Meanings of Of
What does of mean? The traditional account of of highlights the syntactic roles that it plays in language, serving as a grammatical functor, i.e., a preposition, without delving into the meanings it carries (Napoli, 1989; Aarts, 1998). However, this view has been expanded as a plethora of prepositional studies have identified the meanings of these function words by means of image schemas, or the basic representation of our bodily experience (e.g., Lakoff, 1987; Langacker, 1999; Tyler & Evans, 2003). The meanings of of appear to be best described along with its neighbors, or its preceding and following lexical items (i.e., N1 and N2). In other words, to investigate the meanings of of, it is necessary to consider the entire construction (e.g., Kim & Sells, 2015; Schonthal, 2016). In addition to its co-occurring neighbors, semantic relations that capture the relation between of ’s two neighbors appear to be important. The following discussion will first survey previous work from three perspectives: grammatical, cognitive, and phraseological. Then, Section 2.2 is dedicated to discussing semantic relations that are pertinent to of -constructions. Finally, the last section will consider previous studies on how of -constructions are applied and realized in the written discourse by examining their discourse functions.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
2.1.1 A Grammatical Account
Previous grammatical approach to the of -construction concentrates on two of its functions: its possessive and postmodifying functions. For the first function, the possessive function, Quirk et al. (1985) indicate that a number of instances of of -construction can be paraphrased by its genitive alternative as shown in (2.1).
(2.1) a. the city’s population b. the population of the city
(taken from Quirk et al., 1985: 1276)
Apart from syntactic difference, the difference of (2.1a) and (2.1b) is the word order of the two NPs. The authors also consider factors affecting the NPs, including
‘lexical’ (e.g., animate, inanimate), ‘relational’ (e.g., partitive, quantitative),
‘perspective’ (e.g., objective, subjective), ‘syntactic’ (e.g., premodification, postmodification), and ‘communicative’ (e.g., end-weight, end-focus) that may influence one’s selection of the construction. A corpus-based study by Kreyer (2003) elaborates on Quirk et al.’s consideration and suggests that processability and the degree of human involvement may represent two polar-end factors influencing one’s choice of construction. Processability, according to Kreyer, concerns weight and syntactic complexity such as the presence of postmodification following N2. Example (2.2) is a case where the of -construction is heavily postmodified (shown in bold).
(2.2) The aggressive ardour of the professional golfer who might try to cut the slight dogleg and set himself up for an easier shot into the two-tier green (CS4) (Kreyer, 2003: 194; emphasis added)
Human involvement ranges in a continuum, called “the personality scale” (Kreyer, 2003: p.175) ranging from individual (e.g., proper names such as Peter) to non-personal (e.g., inanimate entities such as chair). He found that the of -construction is selected when the degree of human involvement is low such that in Example (2.3a) and when more complex syntactic structure is involved such as postmodification by preposition phrases or premodification on N2 (2.3b).
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
(2.3) a. the condition of the market in which it is to be sold (ARY) (p.192) b. the potential of his drab and neglected establishment (EA5) (p.198) (Kreyer, 2003; emphasis added)
There are a number of previous studies on genitive alternation like Kreyer’s that determine the underlying factors that possibly control a user’s choice between the genitive and of -construction (see Rosenbach (2014) for a comprehensive review).
Although there are factors listed that may have helped characterize of -constructions, the picture may not be comprehensive enough as non-alternating of -constructions have not been considered such as those specify a part-whole relation (e.g., the roof of this house), measurement (e.g., a tin of soup) or type (e.g., this kind of work) (Kreyer, 2003: 170-171; or Quirk et al., 1985: 703 for additional examples). There may also be other types yet to be discovered.
Now turning to the second function of of -construction, i.e., the postmodifying function, Quirk et al. (1985) state that“[t]he most common preposition, of, occurs chiefly as a postmodifier in NPs in a function similar to that of the genitive” (p. 703).
To illustrate of as a postmodifier, consider Example (2.4).
(2.4) a. the gravity of the earth b. the earth’s gravity
(Quirk et al., 1985: 703, emphasis added)
On the basis of genitive alternation, (2.4a) is considered to be equivalent to (2.4b), indicating that the of -phrase (of the earth) functions like a modifier (the earth’s) with the only difference in syntactic position: a premodifier in (2.4b) and a postmodifier in (2.4a). Throughout Quirk et al.’s volume, at least four senses and three relations are identified including ‘a sense of cause/means’ (He died of hunger, p.700), ‘having’ (a man of courage, p. 704), ‘concerning’ (a story of a princess, p.710), and ‘material’ (a bracelet of solid gold, p. 711). The three relations include ‘partition’ ((a) part of the city), ‘quality’ (a kind of wood), and ‘quantity’ (a lot of people) (p. 703). Furthermore, two additional functions, appositive and possessive, involving of -constructions are discussed. According to the authors, apposition describes a relation between two NPs
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
that “must normally be identical in reference” (p. 1301). As shown in (2.5), both Amy and my best friend are co-referential.
(2.5) Amy, my best friend, was here last night.
(Quirk et al., 1985: 1301)
While there are various grammatical means to express apposition, of -construction is among one of them as shown in (2.6).
(2.6) a. the city of Rome b. Rome is a city.
c. The city (that I mean) is Rome.
(Quirk et al., 1985: 1284)
The construction in (2.6a) is related to its copular alternatives in (2.6b) and (2.6c) where the city and Rome have the same real-world referent.
Classifying of as one of the postmodifiers like other prepositions (e.g., in, at, on, and to) is a coherent and logical thinking, because of along with in, at, on, and to belong to the same functional category of preposition, implying that their functions are also similar. However, the peculiarity of of has been noticed by a number of researchers (e.g., Langacker, 1982; Sinclair, 1991). Ample evidence shows that of functions more than a postmodifier (Sinclair, 1991). In fact, those of -constructions which depict relations of partition, quantity, and quality as discussed above all illustrate a non-postmodifying status, because the primary semantic content resides in the of -following NP (e.g., the city rather than part in part of the city). As a matter of fact, Sinclair (1991) argues for a special status for of as a preposition from a phraseological perspective, which will be explored in more detail in the following section.
2.1.2 A Phraseological Account of Of -construction
The phraseological approach has been incorporated into research in various academic fields including lexicography (e.g., Kjellmer, 1994), second language acquisition (e.g., Pawley & Syder, 1983), conversation analysis (e.g., Coulmas, 1981),
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
and collocations (e.g., Renouf & Sinclair, 1991), just to name a few. All of the research share an underlying assumption which is well-captured by Sinclair’s (1991:
110) idiom principle:
[A] language user has available to him or her a large number of semi-preconstructed phrases that constitute single choices, even though they might appear to be analyzable into segments.
The idiom principle underscores the role of conventionalized phrases or prefabricated chunks in language use (Erman & Warren, 2000). In other words, the major difference between a phraseological account of of and that of a grammatical account (Section 2.1.1) is that the former considers the entirety of an of -construction (e.g., the city of Rome) rather than dividing the construction into a phrasal head usually at N1 (e.g., the city) and a postmodifying of -phrase (e.g., of Rome). The importance of identifying conventionalized phrases has also been recognized in academic discourse (e.g., Adel &
Erman, 2012; Biber et al., 1999; Oakey, 2002). Biber et al. (1999), for example, take both deductive and inductive approaches to investigating of -constructions in complex noun phrases (CNP). Here, the deductive approach is aimed to identify the lexico-grammatical patterns and can be regarded as a problem-solving strategy which targets at one particular grammatical structure (in this case, the of -construction) in order to identify its collocating patterns. On the other hand, the inductive approach to pattern finding does not target at any linguistic forms or structures, and is also employed by Biber et al. (1999) under the research called lexical bundles (chapter 13).
Table 2.1 An illustration of major components of the complex noun phrase determiner premodifiers head noun postmodifiers
the increasing orientation of western society to information and information processing activities
Example taken from Biber et al. (1999:643)
In the deductive approach, Biber et al. (1999) regard the of -construction as a CNP with four major components including a determiner, a premodifier(s), a head noun,
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
and a postmodifier(s) as illustrated in Table 2.1. Note that in this example, the authors follow the conventional grammatical approach to classifying of by considering its role in postmodification and identify the collocating patterns of various types of NPs including ‘collective nouns’ (e.g., batch of and pack of ), ‘unit nouns’ (e.g., pair of, piece of, and trace of ), ‘quantifying nouns’ (e.g., box of, cup of, and packet of ), and
‘species nouns’ (e.g., sort of, kind of, type of, and class of ). In comparison with other prepositions, the authors found that the use of of is the most pervasive, occupying about 60-65% of all the prepositional modifiers. The collocation patterns of each type of NPs are also examined in all four registers including conversation, fiction, news, and academic writing. One special finding about academic prose is its prevalence of species nouns such as type(s) and kind(s). The authors reason that these nouns signal an important function of classification in academic discourse. Another variation is found in fiction where collective nouns are sometimes metaphorically expressed (e.g., a flock of messages, and a swarm of panicked men, p.249), unlike the other three registers (conversation, news, and academic writing) where collective nouns are found to be specific to one semantic category such as people with crowd and gang, animals with flock and herd, and inanimate objects with batch and set.
In the inductive approach, Biber et al. (1999) consider lexical bundles which is referred to as “extended collocations” or “bundles of words that show a statistical tendency to co-occur” (p.989). In their study, lexical bundles are determined by frequency of occurrence and are limited to co-occurring word strings with at least three words. The most common sequences identified by the authors are three-word bundles, followed by four-word bundles. The authors found that register variation between academic prose and conversation can be identified by examining the distribution patterns of lexical bundles. First, the number of lexical bundles is higher for conversation (over 80,000 instances per million words) than academic prose (over 60,000 instances per million words). Second, the bundle types diverge between the two registers. There are far more of -containing lexical bundles in academic prose than in conversation, as the former register is closely associated with NPs (e.g., the use of )
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
and prepositional phrases (e.g., as a result of ) and the latter with verb phrases (e.g., I don’t know). Moreover, a large number of these lexical bundles in academic prose contain an of -phrase fragment (e.g., the end of the, the position of the, as a result of, and about the nature of ). As most of these fragments constitute the first part of the of -construction (N1 and of ), it remains unknown to what extent the N2s in these bundles have in common.
Another data-driven approach to of -constructions can be found in Francis, Hunston, and Manning’s (1998) seminal work on English grammar. In their pattern-oriented corpus study, Francis et al. identify a total of 39 categories with the pattern [N of n]. This work is not only comprehensive but also provides a large inventory of words under each category as partially represented in the examples in Table 2.2. Yet, the categorization allows room for further work as there are some instances categorized under ’others’ that could not have been sorted into the previously mentioned categories.
Sinclair’s (1991) work on of -constructions is one of the earliest to show concern on a treating of merely as a postmodifying device. His concern is contingent on his analysis of headedness for of -constructions. On the basis of corpus data, he maintains that the semantic head, defined as the most crucial element(s) to convey meaning of a phrasal structure, of an of -construction varies with the types of nominal groups involved. In other words, the semantic head of an of -construction does not necessarily correspond to the nominal at the N1 position, designated as an N1 head, as would have been viewed from the of -phrase-as-a-postmodifier perspective (Section 2.1.1).
Sinclair supports his argument with corpus examples and his analysis shows that three types of semantic heads (N1 heads, N2 heads, and double-heads) can be identified (see Table 2.3). While N1-heads contain a postmodifying of -N2-phrase, N2-heads contain N1 groups with lighter semantic content including partitive and quantifier nouns (e.g., middle, evening). The double-heads are exemplified by titles, nominalizations, and premodified N1 (e.g., British premodifies view).
Embarking from Sinclair’s work, Chen and Chung (2015) continued the
‧
Table 2.2 Francis, et al.’s categorization of [N of n] (1998: 176-199)
Category Examples
cause and result agent, root, result
rise and fall absence, arrival, downfall flow arrival, entry, return era age, hour, life, term piece amount, block, bundle
rim area, boundary, continuation fabric stage core, heart, remnant others the 10th anniversary of her mother’s
death, a choice of three dishes
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table 2.3 Sinclair’s (1991) Types of Semantic Heads
Semantic heads Examples
N1-head operations of this sort, the age of eight
N2-head the middle of a sheet, the evening of 5th
August
Double-heads the British view of the late senator, the new President of Zaire
(Examples taken from Sinclair, 1991)
investigation of of -constructions by conducting a near-synonymy study of two reporting verbs, namely, demonstrate and show. We hypothesized that the types of of -constructions in the object position of these two verbs would differ due to the subtle semantic variation between the two verbs. The results affirmed our hypothesis in two ways. First, semantic headedness can be used to indicate the relative information load that each construction carries. We would expect a double-headed construction (e.g., the possibility of spontaneity) to carry more information than a single-headed one (e.g., the range of possibility). It was found that demonstrate tends to co-occur with a higher proportion of double-headed constructions (both of the nominal groups are considered as a head) than show, implying demonstrate’s heavier role in expressing more abstract notions. The second finding concerns with the fact that the verbs exhibit selective restrictions on the types of of -constructions. Demonstrate associates closely with abstract notions, particularly for evaluative and modal purposes (e.g., demonstrate the validity of this principle, demonstrate a harsher reality of that desire). Show, on the other hand, associates closely with evidential function and measurement (e.g., show conventional signs of mild disapproval, show the wide range of conditions). While Sinclair’s semantic head approach presents one way to classify the of -constructions, more work appears to be needed if we want to understand the of -construction as a whole. One aspect of the research is to investigate the relationships that the head and non-head nominals maintain in each of -construction. For example, the N2-headed construction parts of the body would indicate a part-whole relation which would be
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
quite different from any double-headed constructions that are more likely to denote an abstract notion such as titles (e.g., the Duchess of Bedford) and nominalization (e.g., the widespread avoidance of call-up, the design of nuclear weapons) (examples taken from Sinclair, 1991: 90-91). To some extent, the meaning of the of -construction is maintained by stimulating a relation between the two nominal groups as having been addressed by cognitive linguists presented in the next section.
2.1.3 A Cognitive Account
A major tenet in cognitive grammar is that we use general cognitive mechanisms, such as memory and categorization to process language (e.g., Croft & Cruse, 2004;
Evans & Zinken, 2007; Schmid, 2007). Another mechanism called attention allocation, or the concept of perceptual salience, has been successfully demonstrated in the studies of prepositions (e.g., Lakoff, 1987; Schmid, 2007; Zlatev, 2007). As explained by Schmid (2007), cognitive linguists have adopted the concept and terminology of Gestalt psychology, a branch of psychology that studies perception which stems from the idea that the whole is not the same as the sum of its parts (Rock
& Palmer, 1990), Figure/Ground alignment, to distinguish different degrees of salience. According to Talmy (2000), “[t]he Figure is a moving or conceptually movable entity” and “[t]he Ground is a stationary reference entity” (p.289). Such a distinction clearly demonstrates the pairing of the Figure with primary salience and the Ground with secondary. The Figure/Ground terminology is sometimes referred to as trajector/landmark to be used specifically for prepositions by cognitive linguists (Schmid, 2007: 128). While ample research has been given to spatial prepositions such as in and over, much less attention has been paid to non-spatial prepositions or prepositions like of with both spatial and non-spatial senses (Tyler & Evans, 2003).
Despite the lack of interest, of has been analyzed by Langacker (1982; 1999) to demonstrate another fundamental claim of cognitive grammar or the premise that grammar simply consists of symbolic units which form an arbitrary linking between a semantic structure and a phonological form (Langacker, 1987; 2008). Elaborating
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
from this position, Langacker (2008) posits further that “all constructs validly posited for grammatical description (e.g. notions like “noun”, “subject”, or “past participle”) must in some way be meaningful” (p. 5). The meanings of of have been demonstrated by Langacker (1982; 1999) to support this point of view.
In his analysis of the of -construction, Langacker (1982) claims that grammatical morphemes such as of “make active semantic contributions to the expressions in which they appear” (p.32). He describes of as “[profiling] a relationship between two entities such that one of them (the trajector) constitutes an inherent and restricted subpart of the other (the landmark)” (p. 74). Langacker (1982, 1999) explicates the meanings of of by making a stark contrast with other prepositions. For example, of differs from on and to in the sense that the latter two provide a link for two separable entities as shown in (2.7).
(2.7) a. the {bottom/?label/?lid} of the jar b. the {?bottom/label/lid} on the jar c. the {??bottom/?label/lid} to the jar (taken from Langacker, 1999: 74)
The example in (2.7) shows that of is specific for profiling, or designating, an intrinsic relation between two non-separable nominal groups, i.e. the bottom and the jar, as exemplified in (2.7a). Profiling describes a particular way of organization where the Figure is represented by the meaning of an entity which is designated as a profile and the Ground is represented by the conceptual framework of the Figure which can be designated as a base. In contrast, the other two trajectors (the label and the lid) which are physically separable from the landmark (the jar) stipulate a relationship different from that of of as exemplified in (2.7b) and (2.7c). In (2.7b), on profiles a sense of physical attachment, and in (2.7c), to profiles a sense of belonging or “an integrated assembly”
(p.74), which establishes an extrinsic relationship between two distinct entities. In a similar vein, the meaning of of is distinguishable from that of in as shown in (2.8).
(2.8) a. the tip {of/*in} my finger b. the splinter {*of/in} my finger
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
(taken from Langacker, 1999: 75)
Whereas of in (2.8a) designates an intrinsic relation that is between the tip and my finger, in, along with on and to, specifies a relation that is extrinsic or links two distinct objects in (2.8b). In the case of (2.8b), a splinter is considered as an alienation to a finger, since they do not naturally belong to each other unless something unanticipated forces their union. Thus, the relation between the two is established as extrinsic or at random. In contrast, any part-whole relation such as the tip or the top of a finger would be paired naturally as a finger always has a portion referred to as its tip or top. Overall, Langacker makes use of an intrinsic relation, epitomized by a part-whole or an inherent relation, to refer to any random conjoining of two entities by some external force.
Another influential perspective taken by cognitive linguists such as Lakoff and Johnson (1980) and Lakoff (1987) is conceptual metaphor. In a cross-linguistic study of Indo-European languages, Nikiforidou (1991) argues that the genitive (defined as a
Another influential perspective taken by cognitive linguists such as Lakoff and Johnson (1980) and Lakoff (1987) is conceptual metaphor. In a cross-linguistic study of Indo-European languages, Nikiforidou (1991) argues that the genitive (defined as a