This collection comprises a selection of eleven papers presented at the workshop Morphosyntactic Categories and the Expression of Possession, held in Manchester, UK, on 3–4 April 2009. In addition, a useful introduction by the volume's editors, Kersti Börjars, David Denison & Alan Scott, summarizes the main points covered in the papers. Discussions of English dominate, with seven of the papers devoted to various aspects of English possession, although Stephen R. Anderson's paper (‘The marker of the English ‘Group Genitive’ is a special clitic, not an inflection’) introduces data from some much less known languages (Kuuk Thaayorre, spoken in Queensland, Australia and Heiltsuk, spoken in British Columbia, Canada). Three other languages are the main topics of papers by Liliane Haegeman (West Flemish), Maria Koptjevskaja-Tamm (Swedish) and Tina Bögel & Miriam Butt (Urdu), while Catherine O'Connor, Joan Maling & Barbora Skarabela take a cross-linguistic approach, comparing English with languages of the Germanic, Slavic and Romance families.
It is not immediately apparent what holds this collection together, other than the topic of possession. The theoretical approaches are diverse, ranging from generative and Optimality Theoretic (Anderson) and Lexical-Functional Grammar (Bögel & Butt) treatments to usage-based models (O'Connor et al.) and cognitive grammar (Hudson). The data sources are diverse. For example, several authors use various contemporary and historical corpora but there are also papers based on native speaker intuitions (Haegeman), and mixture of contemporary corpus, internet and solicited examples (Koptjevskaja-Tamm). The analytical methods are diverse, and various types of quantitative and qualitative approaches are brought into play: variationist (Szmrecsanyi), logistic regression analysis (Börjars, Denison, Krajewski & Scott, as well as O'Connor et al.), percentage distribution (Allen, as well as Juvonen) to mention a few. The analytical goals are also diverse, and even the types of constructions studied differ from paper to paper; some authors consider a specific narrow context, e.g. John Payne's analysis of the oblique genitive, also known as the ‘double genitive’ e.g. a friend of John's, others look at all instances of the possessive morpheme, e.g. Juvonen also considers compounds such as townspeople.
Given this eclectic set of papers coupled with the uneven focus on English, the editors do well to point out that the emphasis should be on the questions. Though the English genitive has been well studied at least since Jespersen (Reference Jespersen1905), the questions its study raises are some of those central to contemporary linguistics: What is the nature of affixes and clitics and should they have a more fine-grained treatment? Is grammaticalization a one-way process? What conditions the choice between two (or more) alternatives in the grammar? Another challenge for the editors was how to organize the volume. It appears that the historically-based papers have been placed at the beginning followed by those with a heavy emphasis on statistical methods, with the non-English studies placed at the end.
The first three papers address data from Old, Middle and Early Modern English. First, building on extensive earlier work on English genitives, Cynthia Allen's ‘Dealing with postmodified possessors in early English’ focuses on prenominal possessor nouns that are postmodified by a prepositional phrase, as in the three constructions exemplified in (1) below.Footnote 1 In present-day English these may be expressed by the ‘group genitive’, as in (1), but in Middle English they often appear as a ‘split genitive’, (1b), in which a prepositional phrase bearing a thematic relationship with the possessor noun is seemingly extraposed. The example in (1c), is often termed the ‘his-genitive’.
- (1)
a. the King of France's daughter
b. the King's daughter of France
c. the king of France his daughter
Allen's general question is one which is also taken up by several other authors in the volume. When speakers (and writers) seem to have a choice between two or more constructions, what conditions that choice? As Allen explains, it is natural to assume that that the Middle English split genitive results from the loss of case in Old English and that it developed from constructions of the type in (2) by replacing genitive case on the possessum (frances) with the preposition of.
(2)
However, examination of the morphosyntax of Old English finds that split genitives of the type in (2) are unusual in Old English, whereas other types of splits (in which the extraposed material does not bear a thematic relationship to the head noun) are not unusual. In most of the examples it is not clear whether the possessum has simply been extraposed within the possessor phrase (the nominal) or, as was common in Old English, right dislocated to a clause boundary. Ambiguity of this type may be seen with the genitive phrase in the example below.
(3)
Moreover, all the examples of this type occur in late Old English texts. This ambiguous environment would appear to make reanalysis possible. Moving to Middle English, Allen argues against the assumption that the group genitive developed out of the split genitive and also against the assumption that the his-genitive (which she prefers to call the ‘separated genitive’) is analogous to possessor doubling constructions found in many Germanic languages (see example (5) below), arguing that it is an orthographical variant of an attached marker. The split genitive becomes frequent in early Middle English and its competition with the group genitive does not start until the end of the 14th century. Through careful analysis of split and group genitives, their frequency and the type of marking, Allen concludes that the main factor influencing choice is ‘structural complexity’ (p. 29). The most frequent group genitives tend to be names, titles and fixed expressions with a simple structure (det/poss N of N). If the possessor is premodified or if the possessum becomes more complex, the tendency is to use a split genitive. Of course, it is generally thought that the split genitive died out in the Early Modern period leaving speakers to find other ways of dealing with complexity.
In her article Allen comments on that by Kersti Börjars, David Denison, Grzegorz Krajewski & Alan Scott (in the same volume), as they make the intriguing claim that the split genitive has not, in fact, died out in present-day spoken English. Although not a historical study (the data come from the spoken part of the British National Corpus) and although they use a very different analytical method, Börjars et al. investigate a related question: What governs the choice modern speakers make between the possessive ’s and the of-construction? Their particular interest is in the debate over the affix or clitic status of the possessive ’s, the main evidence for its clitic status being the ability to appear at the right edge of a possessive phrase. It is well known that animacy, topicality and core possession (e.g. inalienable possession) are factors that favour possessive ’s and these they look at first, before going on to look at the influence of structural factors. They discuss, and then by-pass, the thorny question of how to measure structural complexity by taking length (in words) as its equivalent. This is a discussion that Allen appears to have evaded and it is not clear whether, for her, number of words or number of phrases is relevant; possibly, by implication, both are. The results show that the most influential factor when it comes to speaker choice is weight, and it turns out that heavy postmodification of the possessor has more influence than premodification. Börjars et al. also claim (p. 141, contra Quirk et al. Reference Quirk, Greenbaum, Leech and Svartvik1985:1282) that split constructions, such as that below, are grammatical:
(4) We don't know the gentleman's name with the tape recorder.
Allen (p. 34), however, questions whether these constructions are similar to those found in Middle English where the extraposed material is still clearly within the nominal. Many modern examples may, instead, involve extraposition to the right edge of the clause. Nevertheless, this is a strategy to avoid possessive’s when there are heavy postmodifiers. The conclusion reached by Börjars et al. is that the possessive ’s has an ambiguous status and there is no clear-cut distinction between affixes and clitics. Somewhat more controversially, they touch on questions relating to the nature of the mental grammar, suggesting that the native speaker's knowledge of grammar might extend to knowledge of statistical preferences.
In the second historical article, ‘Variation in the form and function of the possessive morpheme in Late Middle and Early Modern English’, Teo Juvonen looks at possessive marking in a selection of texts. Partly aiming to complement Allen's study, he skirts the affix vs. clitic debate to look more closely at register. In a more fine-grained analysis than others who consider simple animacy and agency as factors that favour possessive ’s, he breaks his possessor types into: human, your (as in your graces pleasure, which only occurs in letters), king, God, collective, animal, and inanimate, and his text types into: sermons, history writing, and letters, looking then at the distribution of split-genitives, group genitives, complex possessors and simple possessors. He suggests that pragmatic, stylistic and sociolinguistic factors account for the his-genitive, whereas the change from split genitive to group genitive happened across the board.
In the third historical article, ‘The great regression: Genitive variability in Late Modern English news texts’, Benedikt Szmrecsanyi is also concerned with specific text types in an analysis of British newspapers taken from ARCHER (A Representative Corpus of Historical English Registers). The investigation is into the choice between possessive ’s and the of-genitive and the challenge is to keep apart effects from fluctuation in topic or within the genre (what she calls ‘environmental’, i.e. language-external factors) from genuine diachronic changes in the grammar. The statistical analysis turns up a number of detailed interesting results. Perhaps the results of interest to the widest audience are those related to the question of grammaticalisation and possible degrammaticalisation. The frequency of use of the inflected genitive was found to be fairly stable from the 14th to the mid-18th century, after which, it decreased sharply until the first half of the 19th century, then started to become more frequent and increased rapidly in the 20th century. Using logarithmic regression analysis to abstract away from grammar-external factors, these results are then interpreted as showing grammaticalisation in the 19th century and degrammaticalisation in the 20th. A somewhat surprising methodological choice, adopted without discussion in this paper, is that heaviness or complexity, i.e. weight, is calculated not as number of words (as did Börjars et al.) but as number of graphemes.
Moving to the end of the volume, the last three papers deal with languages other than English, though, for the most part, they address similar questions. The issue of two competing constructions in a language is again taken up, this time by Liliane Haegeman, in ‘Two prenominal possessors in West Flemish’. This data-rich discussion has the goal of describing in detail the similarities and differences between two patterns for expressing possession, as in (5a) and (5b):
(5)
In (5a), we have a ‘possessor doubling’ construction and the possessive pronoun zenen agrees with the noun oto, whereas in (5b), sen is an invariant marker which has a phonologically conditioned variant se. This paper is one that takes a very wide view of what a possessive construction might be, basically all constructions with these syntactic forms regardless whether the thematic relationship between the prenominal possessor head noun and the possessum is that of possessor, agent, experiencer or theme. After a detailed survey of the two constructions along with some comparisons with Afrikaans it is suggested that one of the most significant differences is that the sen construction has a rigid adjacency requirement, unlike the doubling construction, leading to the conclusion that the two constructions cannot have an identical syntax.
Following this, comes Maria Koptjevskaja-Tamm's ‘A Mozart sonata and the Palme murder: The structure and use of proper-name compounds in Swedish’, where a very specific and rarely studied type of construction is described, i.e. a noun–noun compound in which the first noun is a proper noun. Two research questions are addressed (i) Are these types of compounds referential?, and (ii) To what extent is there a distinction between instance specification (a property of s-genitive constructions) and type specification, (normally attributed to compounds). The answers to both questions appear inconclusive and further research is called for.
The final paper in the volume is Tina Bögel & Miriam Butt's ‘Possessive clitics and ezafe in Urdu’. Again, we have a language where there is a choice between two constructions, one employs a genitive marker, analysed as a clitic, whereas in the other it is uncertain whether ezafe, borrowed from Persian, is part of the nominal morphology or a clitic. A useful brief section introduces readers not familiar with Urdu to the basics of the language, and there follows a description of, and comparison between, the two constructions. Although a number of the papers in this collection make some reference to Zwicky & Pullum's (Reference Zwicky and Pullum1983) criteria for distinguishing clitics and affixes, Bögel & Butt are very systematic, and as part of their analysis they examine each criterion in turn. The overall conclusion with respect to ezafe is that it is a clitic. Also interesting in this paper is the discussion of various other analyses of ezafe, particularly how it is handled under different theories (e.g. Head-driven Phrase Structure Grammar, HPSG).
Of the four papers that have not yet been discussed, two others are not exclusively about English. First, Catherine O'Connor, Joan Maling & Barbora Skarabela's paper ‘Nominal categories and the expression of possession: A cross-linguistic study of probabilistic tendencies and categorical constraints’. Again, the focus is on factors that influence a speaker's choice between two constructions, here the choice in English between ’s as in the earth's interior and the English of as in the interior of the earth. Three factors known to influence the choice are taken into account: animacy, weight (calculated in number of words) and discourse status; prenominal possessor NPs tend to be animate, low-weight and discourse-old. Using a tagged version of the Brown Corpus, the statistical tendencies were calculated for each factor and it was found that animacy was the most influential, followed by weight. Then O'Connor et al. look at the construction which they call the Monolexemic Possessor (MLP) construction, i.e. a one-word, prenominal, animate, discourse-accessible possessor, in 17 different languages, and argue that the same probabilistic tendencies show up as an accessibility hierarchy, which is, in order of most accessible to least: pronoun > proper noun > kinship term > common noun (p. 111). In the conclusion they argue for a usage-based model of grammar and see their results as supporting Hawkins’ (Reference Hawkins2004) Performance Grammar Correspondence Hypothesis.
As mentioned earlier, although not apparent from the title, Stephen R. Anderson's ‘The marker of the English “Group Genitive” is a special clitic, not an inflection’ appeals to some lesser known languages to support the arguments. The discussion centres around two different approaches to the group genitive, whether it is special clitic generated by a rule: ‘Adjoin /z/ to the final syllable of a DP bearing the feature [Poss]’ (p. 197) or an ‘edge inflection’ (p. 198), a word-level inflection that is transmitted along the phrase (or rather down the tree) until it reaches the edge. English is compared with Heiltsuk, which clearly has a special clitic, and Kuuk Thaayorre, which clearly has edge inflection and the special clitic analysis is argued for in English. A consequence of this analysis is that ’s in English would be situated at the right edge of the DP in the specifier of DP, a somewhat controversial suggestion given that many syntacticians argue that ’s is on a par with determiners such as the indefinite article, and is the head of DP.
Finally, two papers dealing exclusively with English are John Payne's ‘The oblique genitive in English’ and Richard Hudson's ‘A cognitive analysis of John's hat’. Hudson looks at ’s (and briefly at of-constructions), from a language acquirer's perspective, and which he analyses as sometimes a clitic and sometimes an affix. His conclusion is along the same lines as that of Börjars et al., except, for them, ’s in English is something in-between the two prototypes, whereas for Hudson it is an either/or situation, sometimes one and sometimes the other. Many linguists would find his conclusion, ‘If a multiple, messy and incomplete analysis is good enough for a language learner, it should certainly be good enough for us’ (p. 175), rather depressing and defeatist.
John Payne's article, ‘The oblique genitive in English’, discusses the construction that is more familiar to many as the ‘double genitive’, (6a). The construction has been linked with three other constructions, presented in (6b–d):
(2)
Drawing on the available literature, Payne first compares the oblique genitive with each of the three other constructions in turn. This is followed by a usage-based analysis drawing on a search of the British National Corpus for the string N + of mine. The 842 examples found were analysed for semantic relations and 13 different categories were found. Also, the frequency of the various determiners possible was analysed. It is concluded that the oblique genitive overlaps with each of the three other constructions in a fairly narrow space. It is not a variant of any of the three constructions but is an independent construction with its own independent semantics.
With such an eclectic collection, it is difficult to specify an audience for this volume, though each paper is interesting in its own right. Some authors, in particular Haegeman, Allen, and Bögel & Butt, have presented a wealth of examples which readers will find useful whatever their theoretical inclinations. Others, e.g. Börjars et al., Szmrecsanyi, and O'Connor et al., have taken care to describe their method and analysis in detail which could serve as a guide to others seeking to replicate the method. Still others, e.g. Hudson, and Bögel & Butt, have taken care to explain in detail their particular theoretical assumptions, useful for readers not familiar with those approaches.
Finally, all the papers are well written and presented. The following minor editorial issues were noted: Footnote 6 on page 22 is repeated verbatim as footnote 17 on page 24; three references on page 147 (Newmeyer 2003, Clark 2005 and Gahl & Garnsey 2006) are not in the reference list; and there appears to be an error in the gloss on page 219.
