The Choice between Prenominal and Postnominal Possessors
|1. It enables the complete identification of the possessee.|
The referent of the possessor is a specific, definite entity whose identification can be made by both the speaker and the hearer. This type of reference point enables the hearer to restrict the set of possible referents of the possessee to one particular referent: Barbara's bike ('the bike owned by Barbara').
|2. It restricts the number of possible referents of the possessee by excluding referents with other possessors.|
For example, in Slovene constructions like nek Barbarin prijatelj ('a'+'Barbara's'+'friend' >> 'a friend of Barbara's') the referent of the possessee is a specific, indefinite entity. Although the (specific, definite) possessor cannot facilitate the complete identification of the possessee, it can be regarded as a reference point since it excludes certain referents ('those who are not Barbara's friends') and thus restricts the set of possible referents of the possessee to a subset ('only those who are Barbara's friends').
Similarly, in English constructions like [some student]'s bike the referent of the possessor is a specific, yet indefinite entity. Despite its indefinite status the possessor has the function of a reference point: it excludes certain referents ('bikes that are not owned by students') from the set of possible referents of the possessee.
|3. It restricts the set of possible referents of the possessee to one particular class.|
This type of reference point is present in possessive compounds, which are characterized by the classifying function of the possessor. For example, a [lady's bike] denotes a type of bike, i.e. 'a bike designed for women' and not 'a bike owned by some particular lady' (in fact, the possessive compound [lady's bike] does not exclude the possibility that the bike in question is owned by a man: Bob's [lady's bike]). The most striking difference between possessive compounds and other possessive constructions is the fact that possessors in possessive compounds are non-referential, i.e. they have no referents in the real world. In the case of a [lady's bike] only the possessee (bike) has a referent, either a specific (e.g. She bought a new lady's bike. The lady's bike she bought was quite expensive.) or a non-specific one (e.g. A lady's bike is easy to ride.). The possessor (lady), on the other hand, does not refer to any particular lady or ladies in general; it merely states the type of bike. The situation would be different if the phrase a lady's bike was not a possessive compound but a prototypical possessive construction: [a lady]'s bike ('the bike owned by some particular lady'). The two constructions, however, are not unrelated. If we ignore the fact that ladies' bikes are not popular with all women, we can conclude that if some lady owns a bike ([a lady]'s bike) it will be a bike designed for women (a [lady's bike]). If we start the other way round, we can now provide a definition of the possessee in our possessive compound on the basis of comparison with the possessee in the homonymous (prototypical) possessive construction: a [lady's bike] = a bike of the same type as [a lady]'s bike, i.e. the bike owned by some particular lady. The possessor in the possessive compound is, notwithstanding its non-referentiality, at least indirectly in the function of a reference point: it restricts the set of possible referents of the possessee to a particular class, the class being the same as the class the possessee in the homonymous possessive construction belongs to.
As argued in the study, all the above functions of the possessor can be described as reference-point functions. Extending the notion of reference point makes it possible to associate the reference-point function with any prenominal possessor, irrespective of how far it may be removed from the prototype.
2. The empirical part of the study
2.1. The hypotheses
The study addressed the question of choice between prenominal and postnominal possessive constructions in English and Slovene. In accordance with the reference-point analysis of the possessive construction, two hypotheses were put forward:
|1. The prenominal possessor has the reference-point function by default. (H1)|
|2. The identification value of the possessor is the main factor determining the choice between prenominal and postnominal constructions. (H2)|
For either language, the hypotheses were checked against empirical data. The data were compiled from the B(ritish) N(ational) C(orpus) (a corpus of modern English) and the corpus FIDA (a corpus of modern Slovene). A brief description of the sources is given in the section below.
2.2. The sources
Contrastive studies often involve analyses based on parallel corpora, i.e. collections of source-language texts together with their translations into the target language(s). Data obtained in this way, however, do not always reflect actual use because a particular translation variant may be chosen for a variety of reasons. Therefore the study made use of referential corpora instead: the BNC served as the source for the "English" part of the analysis and FIDA for the "Slovene" part.
Both the BNC and FIDA can be characterized as:
(i) 100-million-word corpora,
(ii) monolingual corpora (each dealing with one language only although foreign language words do occur in it),
(iii) synchronic corpora (containing texts from roughly the same period, i.e. the late twentieth century),
(iv) general corpora (including many different styles and varieties; not being limited to any particular subject field, genre or register).
They differ in two aspects regarding text selection:
(i) FIDA is a written corpus only whereas the BNC includes both written and spoken language.
(ii) FIDA includes all its texts in full whereas the BNC is a sample corpus: with the exception of shorter texts and multi-author texts, samples of 45,000 words are taken from various parts of single-author texts. Sampling allows for a wider coverage of texts and avoids over-representing idiosyncratic texts.
The actual source for the English part of the analysis was the BNC Sampler, which includes 184 texts selected from the BNC, with the variety of text-types and genres preserved. The Sampler contains one million words of written text and one million words of spoken text. In order to maintain the parallel with FIDA, the spoken component of the Sampler was excluded from the analysis. From the written component, 500 examples of the prenominal possessive construction (i.e. with a Saxon genitive phrase as the possessor) and 500 examples of the postnominal one (i.e. with an of-phrase as the possessor) were chosen at random to serve as empirical data for the analysis.
Similarly, only a small portion of FIDA served as the actual source for the Slovene part of the analysis. The problem of FIDA is that the majority of the texts included are translations from foreign languages, predominantly from English. For the purposes of the study, originally Slovene texts had to be selected. The selection followed two criteria: time (the texts had to be published in the last decade) and domain (they had to cover different kinds of writing). Thus 24 texts were chosen, none of them dating back further than 1996: 12 from fiction, 6 from the field of natural sciences, 6 from the field of humanities. For each text, the first 100 examples of the prenominal possessive construction and the first 50 examples of the postnominal one (a smaller amount of empirical data was needed here because the analysis focused on the use of the possessive adjective) were extracted. In the case of short texts, of course, these numbers were smaller. Besides, 200 examples of the prenominal construction and 100 examples of the postnominal one were taken from national newspapers and periodicals. In total, the Slovene part of the analysis comprised 1954 examples of the prenominal construction (i.e. with a possessive adjective as the possessor) and 1200 examples of the postnominal construction (i.e. with a postnominal genitive phrase as the possessor).
2.3. The analysis
Focusing on each language separately, the analysis comprised 500 examples of the English prenominal (Saxon genitive phrase) construction, 500 examples of the English postnominal (of-phrase) construction, 1954 examples of the Slovene prenominal (possessive adjective) construction and 1200 examples of the Slovene postnominal (genitive phrase) construction.
The English part of the analysis was carried out on a smaller scale because there had already been similar empirical studies, all of them confirming the tendency for prenominal possessors in English to be high in topicality. For Slovene, on the other hand, this was the first corpus-based analysis of the kind.
In the case of each construction, both factors contributing to the identification value of the possessor were considered: topicality and informativity. Therefore each example was classified according to (i) the semantic properties of the possessor (building up its topicality) and (ii) the semantic properties of the possessee and the semantic relation between the possessor and the possessee (building up the informativity of the possessor). The classification followed the following frame:
1.1 PRENOMINAL POSSESSOR: the Saxon genitive phrase
1.1.1 Possessor (<> topicality)
1.1.2 Possessor <==> Possessee (<> informativity)
1.2 POSTNOMINAL POSSESSOR: the of-phrase
1.2.2 Possessor <==> Possessee
2.1 PRENOMINAL POSSESSOR: the possessive adjective
2.1.2 Possessor <==> Possessee
2.2 POSTNOMINAL POSSESSOR: the genitive phrase
2.2.2 Possessor <==> Possessee
Besides its semantics, the following aspects of the possessive construction were relevant for the analysis:
(i) the form of the possessor (i.e. the grammatical number and the modifiers)
(ii) the form of the possessee (a noun, a deverbal noun, a gerund)
(iii) the syntactic function of the possessor (relevant in cases where the possessee was realized by a deverbal noun or a gerund)
2.4. The findings
The analysis of the empirical data has shown that (1-3):
|1. In both languages the prenominal construction is characterized by a high identification value of the possessor, which is reflected in the fact that (1.1.) the possessor is highly topical, i.e. mentally accessible, and (1.2.) the relation between the possessor and the possessee is such that the possessor emerges as a highly informative participant in the relation.|
1.1. The prenominal possessor is realized predominantly by:
(i) Personal names:
(1) Joan's hesitation, the Simpsons' flat
(2) Žigov naslov ('Žiga's address'), Pogačnikova ponudba ('Pogačnik's offer')
(ii) Kinship terms and nouns denoting mutual interpersonal relations:
(3) his wife's persistent chattering, her friend's letter
(4) očetov obraz ('her father's face'), sosedino vedro ('the neighbour's bucket')
(iii) Other nouns denoting people:
(5) the old woman's death, the architect’s own design
(6) starkina odločnost ('the old woman's determination'), doktorjevi gostje ('Doctor's guests')
Less frequent, however not unlikely to occur as prenominal possessors, are names of institutions, places and countries, common nouns referring to these entities (e.g. city, country, firm, institute, island, nation, school, state, town) and expressions of time. Their common denominator is that they all relate to human activities and interests, which makes them relatively high in topicality. Abstract concepts, on the other hand, have the lowest possible degree of inherent topicality and are as such not used as prenominal possessors.
1.2. The prevailing relations between the possessor and the possessee are the following:
(i) The possessee denotes some part of the possessor's body or psyche:
(7) Elizabeth Danziger's face, Richard's mind
(8) Irenine oči ('Irene's eyes'), Erazmova duševnost ('Erasmus's mind')
(ii) The possessee denotes some part of the possessor's life or identity:
(9) his father's death, the girls' switched identities
(10) hčerina poroka ('her daughter's wedding'), sestrino ime ('her sister's name')
(iii) The possessee denotes some quality of the possessor:
(11) Hamlet's lunacy
(12) Janezova hvalisavost ('John's boastfulness')
(iv) Between the possessor and the possessee exists a kinship relation or a mutual interpersonal relation:
(13) the old farmer's daughter, Rory's pals
(14) Aleševa sestra ('Alex's sister'), hčerkin fant ('her daughter's boyfriend')
(v) Between the possessor and the possessee exists an "employer-employee" relation:
(15) Gorbrandt's men, the young king's faithful chamberlain
(16) očkov šef ('Dad's boss'), Oblakovi fantje ('Oblak's boys')
(vi) The possessor is the owner of the possessee:
(17) Karen's things, the boy's sandshoe
(18) Kajini čevlji ('Kaja's shoes'), Martina bajta ('Martha's shack')
(vii) The possessor is the performer or carrier of the activity or state denoted by the possessee:
(19) Mr Gorbachev's policies, the group's main strategy
(20) Kristusovi čudeži ('Christ's miracles'), Andrejeva kariera ('Andrew's career')
(viii) The possessor is the author of the thing denoted by the possessee:
(21) Rupert Brooke's poems, the architect's own design
(22) očetov prejšnji film ('his father's previous film'), Einsteinova formula ('Einstein's formula')
(ix) If the possessee is realized by a deverbal noun or a gerund, the possessor performs the function of an intransitive subject or a subject that does not denote a prototypical agent (i.e. a volitional initiator or causer of an action that affects the state of another participant in the process):
(23) white farmers' fears (<< White farmers fear something.), his wife's persistent chattering (<< His wife chatters.)
(24) otrokov psihofizični razvoj << Otrok se razvija. ('the child's mental and physical development' << 'The child develops.'), Sovino hukanje << Sova huka. ('Owl's hooting' << 'Owl hoots.')
What all these relations have in common is the fact that in each of them even a non-relational possessee nominal starts to behave as a relational one. Thus the possessor becomes part of the semantic structure of the possessee, the result being that the possessor gains in its informativity.
|2. Unless their use is precluded by some morpho-syntactic factors, possessors with high identification values do not occur in postnominal constructions.|
In English, the genitival -'s can attach to both complex and plural possessors. In the following cases, however, a postnominal construction is normally used (even if the possessor has a high identification value):
(i) There's an apposition describing the possessor:
(25) daughter of John Neville, a Dublin merchant
(ii) There's a relative clause describing the possessor:
(26) the help of Dr Blackwell, who needed a moral support now
(iii) The possessor is rather long:
(27) the future of new director general John Birt
(iv) The possessor is coordinated with another possessor:
(28) the names of the newly elected Mayor and the officials appointed by him
In Slovene, on the other hand, the complexity and the grammatical number of the possessor seem to be crucial: as soon as the possessor is complex or plural (dual) in form, the derivation of a possessive adjective is not possible. An attempt was made in the study to associate even these two factors with inherent topicality. The fact that personal names and kinship terms are used almost exclusively as prenominal possessors indicates that they tend to be simple in form. This tendency can be understood as a reflection of their topicality: as highly topical entities they do not need further description in order to be conceptualized by the hearer. Other types of nouns denoting people, which are slightly lower in topicality, are more often modified, the syntactic consequence being their use as postnominal possessors. The need for modification and topicality are obviously related.
It is worth noting at this point that topicality is also connected with semantic classes of nouns that can(not) serve as derivational bases for possessive adjectives: most of the nouns that have a low degree of inherent topicality cannot perform this function. This is a reflection of the fact that the reference-point function of all prenominal possessors originates in the function of a prototypical possessor, i.e. the single, human, definite and mentally accessible owner of a particular thing. The analogy with the prototype seems to be so strong that the derivation of possessive adjectives is restricted mostly to nouns denoting people, i.e. nouns that are compatible with the function of a prototypical possessor.
|3. Possessors with low identification values are normally used in postnominal constructions.|
The only exception seems to be the type of construction generally known as possessive compounds, where the function of the possessor is classifying: the possessor does not function as a prototypical reference point but merely states the type (class) of the possessee (e.g. a [boy's school]). Representing a considerable extension from the prototype, possessive compounds
may be excluded from the above generalization.
For both languages, the findings confirm the hypothesis that the prenominal possessor has the reference-point function by default (H1). This makes the prenominal possessive construction a perfect device for introducing new concepts into the discourse: a new entity is introduced through an entity that has already been conceptualized by the hearer. Using the prenominal construction, the speaker makes it explicit to the hearer that the possessor is able to perform the reference-point function and thus facilitate the identification of the possessee.
The findings also confirm the hypothesis that the choice between prenominal and postnominal constructions depends primarily on the identification value of the possessor (H2). At first sight it seems that in the case of Slovene constructions this factor is overridden by the complexity and the grammatical number of the possessor; in addition, a possessive adjective cannot be derived from any semantic class of noun. As proposed in the study, even these factors can to some extent be explained in terms of identification value, namely (inherent) topicality.
Finally, the findings confirm the universal nature of some basic concepts and constructions. The fact that the reference-point analysis of the possessive construction may be applied to languages as different as English and Slovene suggests the potential for further application of the reference-point model to cross-linguistic studies.
 Part of my doctoral dissertation (Lipovšek 2002).
 Pronounced /’fi:da/.
 For more information see http://info.ox.ac.uk/bnc.
 For more information see http://www.fida.net.
 The examples are listed in Lipovšek 2002.