LINGUIST List 24.3673

Thu Sep 19 2013

Review: Linguistic Theories; Morphology; Syntax: Börjars, Denison, & Scott (2013)

Editor for this issue: Monica Macaulay <>

Date: 25-Jun-2013
From: Sebastian Sulger <>
Subject: Morphosyntactic Categories and the Expression of Possession
E-mail this message to a friend

Discuss this message

Book announced at

EDITOR: Kersti BörjarsEDITOR: David DenisonEDITOR: Alan ScottTITLE: Morphosyntactic Categories and the Expression of PossessionSERIES TITLE: Linguistik Aktuell/Linguistics Today 199PUBLISHER: John BenjaminsYEAR: 2013

REVIEWER: Sebastian Sulger, Universität Konstanz


The book ''Morphosyntactic Categories and the Expression of Possession''collects 11 papers that were originally presented at a workshop which tookplace in Manchester in April 2009. The theme of the workshop, the realizationof the concept of ''possession'' (across languages, with a focus on English)through various morphosyntactic constructions, has long been a challenge tolinguistic theories. The papers all investigate aspects of the morphosyntacticmarking of possession from the perspective of a variety of linguistictheories. Since in all of the surveyed languages there are differentconstructions available for realizing possession, particular attention is paidto the distribution of the relevant constructions, using corpus data andstatistical analysis.


In the introduction, Kersti Börjars, David Denison and Alan Scott set thestage for the volume by touching on the historical and theoreticalimplications of the English ''s''-genitive. They mention its usage acrossdifferent constructions, and further note that it has received much attentionin the literature, as it provides a window on a range of issues that influencethe way we think about the architecture of grammar. They sketch thedevelopment of the marker as well as its theoretical treatments over time - asregards the latter, they draw attention to the literature on clitics vs.phrasal (edge) affixes. Moreover, the introduction mentions that mostlanguages referred to in the volume have more than one way of expressingpossession, noting in particular West Flemish and Urdu.

The first paper in the volume, by Cynthia L. Allen, ''Dealing withpostmodified possessors in early English: Split and group genitives'', takes adiachronic look at two post-modified possessor constructions in thedevelopment of written English: the group genitive construction (1) and thesplit genitive construction (2).

(1) the king of France's daughter

(2) the king's daughter of France

Allen documents the rise of the group genitive in the late Middle Englishperiod and its sudden favouring over the split genitive near the start of theEarly Modern English period. For the period where both strategies wereavailable, she provides evidence for the claim that the complexity of theinvolved possessor phrases plays a role: it turns out that the group genitivehas always been used predominantly where possessor phrases were maximallysimple and involved only the possessor N premodified by a determiner or apossessive and postmodified by the simplest possible PP. The split genitive,on the other hand, was found where possessor phrases had more premodifyingmaterial than just a determiner (titles, adjectives, etc.). Allen ties this toprocessing factors: the group genitive was favoured when the possessor wasshort and simple and thus did not require much effort to create or parse theresulting structure.

The second paper, ''Variation in the form and function of the possessivemorpheme in Late Middle and Early Modern English'' by Teo Juvonen, supplementsthe preceding paper as it surveys the use of different strategies in themorphological marking of the possessive in a corpus of Late Middle and EarlyModern English. The morphological strategies discussed by Juvonen are:''s''-ending as in (3), ''s''-less ending as in (4) and separated genitive asin (5).

(3) the Kynges brother

(4) wyth onye of my maister councell.

(5) to yower worly worschyppe and herte ys desyre.

The paper focuses on the different possessive encoding strategies, and howthey were used across the different genres. Juvonen affirms that the groupgenitive had become the dominant form by the end of the 15th century (asdiscussed by Allen); he gives the split genitive example in (6), and mentionsthat by the latter half of the 17th century there seems to be increasinguncertainty about the status of the ''of''-phrase: does it modify thepossessor ''a Bishops'' or the possessum ''son''?

(6) it was Gerelius a Bishops son of Suedeland

Based on this observation, Juvonen suggests that the group genitive replacedthe split genitive because of grammaticalization and semantic bleaching of thepreposition ''of'': he claims that this is what caused the ''of''-phrase to belinked more closely to its head, thus disabling the split genitive.

The third paper, ''The great regression: Genitive variability in Late ModernEnglish news texts'' by Benedikt Szmrecsanyi, again looks at possessiveencoding strategies in English, this time focusing on the''s''-genitive/''of''-genitive alternation in Late Modern English news textsas in (7) vs. (8). The paper documents the collapse in the frequency of the''s''-genitive in the early 19th century and its subsequent recovery, aimingto explain the resulting v-shaped pattern.

(7) the president's speech

(8) the speech of the president

Szmrecsanyi views any subtle changes in the conditioning factors as evidenceof a change in the genitive choice grammar of English, showing why the slumpin the ''s''-genitive frequencies around the first half of the 19th centurywas so much more severe than expected: this has to do with a change in thestatus of the factor ''possessor animacy''. While the sheer input frequenciesof human possessors dropped, the animacy constraint (due to whichhuman/animate possessors favour the ''s''-genitive) was relaxed - an actualchange in the grammar of genitive choice speeding up the decrease in''s''-genitives.

Szmrecsanyi notes that dropping the selectional restriction concerning animacycan be seen as a sign of grammaticalization, while the increasing sensitivityof the ''s''-genitive towards the factors ''possessor thematicity'' and''possessum length'' can be seen as a development towards a freer ''choice ofitems according to communicative intents'' (Lehmann 1995: 164) and thus asdegrammaticalization. The fact that the ''s''-genitive increasingly attractedownership relations can be explained using the ''paradigmatic integrity''parameter, again by Lehmann (1995): lexical-semantic features are added to the''s''-genitive, again indicating degrammaticalization.

The fourth paper, by Catherine O'Connor, Joan Maling and Barbora Skarabelatitled ''Nominal categories and the expression of possession: Across-linguistic study of probabilistic tendencies and categoricalconstraints,'' presents a cross-linguistic study of the Monolexemic PossessorConstruction (MLP). They compare the stochastic patterns of prenominalpossessives in English to the MLP found in a variety of languages (Germanic,Slavic, Romance). The choice of a prenominal possessive over a postnominal onein English correlates strongly with the features animacy, weight and discoursestatus, but is not categorical in nature. This is not the case with the MLP;in this construction, the possessor occurs immediately left to the possessum,and the possessor may not be longer than a single word, as in the Czechexample in (9). If the full name of the possessor is to be expressed, thefull phrase adnominal genitive has to be used as in (9c).

(9) a.Milan-ova knihaMilan-POSS.ADJ book'Milan's book'

a'Kunder-ova knihaKundera-POSS.ADJ book'Kundera's book'

b.*Koupila jsem Milan-ovu Kunder-ovu knih-ubuy.PAST.1.SG.FEM be.PRES.1SG Milan-POSS.ADJ.ACC Kundera-POSS.ADJ.ACC book-ACC'I bought Milan Kundera's book.'

c.Koupila jsem knih-u Milan-a Kunder-ybuy.PAST.1SG.FEM be.PRES.1SG book-ACC Milan-GEN Kundera-GEN'I bought (a/the) book of Milan Kundera.'

Evidence from Czech, Russian, Icelandic, German, etc. is adduced showing thatoptimal weight, discourse status and (less strictly) animacy are allgrammaticalized in the MLP. With respect to discourse status, across-linguistically valid accessibility hierarchy emerges as in (10). If alanguage has an MLP, it will allow it with pronouns; if a language allows itwith e.g. kinship terms, it will also allow it with any element occurring tothe left of the kinship terms in (10).

(10) Monolexemic Possessor Accessibility Hierarchy:Pronoun >> Proper Noun >> Kinship Term >> Common NounMost accessible <-----------------------> Least accessible

The scale in (10) implies that in a given context, pragmatic decisions musttake place to resolve the possessor in an MLP; in particular, the questionarises how a possessor is resolved if there are e.g. multiple pronounsavailable. To address this question, the paper further includes a discussionof whether the categorical restriction is at work in terms of pragmaticcommunicative decisions, or whether it just constitutes a frozen remnant ofthe stochastic tendencies observed e.g. in English; by citing elicitationexperiments with native speakers of Czech, the authors confirm that thecategorical restrictions reflect an active discourse pragmatic requirement.

In ''Expression of possession in English: The significance of the rightedge,'' Kersti Börjars, David Denison, Grzegorz Krajewski and Alan Scottreturn to the topic of ''s''-genitive/''of''-genitive alternation in English.The focus of the paper is on the categorization of the ''s''-genitive as aclitic or an affix. The authors are especially interested in the right edgecriterion, which is key evidence for the ''s''-genitive's status as a clitic:the item's ability to occur at the right edge even in cases where thepossessor is postmodified, as in (11).

(11) the man in the car's wallet

The authors discuss two new variables, length of premodifying sequence as wellas length of postmodifying sequence, to see whether it makes a differencewhere the weight of the possessor is located, before or after the head. Itturns out that the effect of premodification is weaker than that ofpostmodification, so that the latter decreases the odds of the ''s''-genitivemore strongly (unfortunately, the data the authors work with is too sparse toexamine any further the effects of the actual length of the postmodification).The so-called split possessive is argued to be a strategy for avoidingstandard ''s''-genitives where the possessor contains postmodification, andthe data shows a clear correlation between the presence of a split and thelength of the postmodification.

The sixth paper, ''A cognitive analysis of ‘John's hat’'' by Richard A.Hudson, presents a cognitive analysis of the English ''s''-genitive, couchedwithin Hudson's Word Grammar framework (Hudson, 2010). A string such as''John's hat'' spawns two different syntactic analyses in the mind of aspeaker of English. Under the first analysis, the morpheme ''{z}'' behaveslike a suffix and is a direct descendant of the Old English inflected genitivecase; here, the string ''John's'' behaves like a single word which doubles infunction as a determiner. Under the second analysis, the same morpheme ''{z}''behaves like a clitic giving rise to the group genitive.

Hudson claims that each of these analyses has advantages and disadvantages fora learner of English. The suffix analysis involves a straightforwardmorphology/syntax mapping, but the possessor phrase must receive a complexanalysis, doubling in syntactic classification as a (possessive) pronoun and anoun (common or proper). Hudson acknowledges that this analysis seems moreintuitive in cases where we have simple (e.g., proper noun/one word)possessors, like ''John's hat''. When the possessor is complex, and the''s''-genitive is not adjacent to the possessor phrase head, the groupgenitive is the only analysis available. Here, Hudson suggests a simplemapping at the syntax-semantics and morphology-syntax interfaces, at the costof the special morphology involving a clitic.

Hudson further discusses the competition for the ''s''-genitive and the''of''-genitive, and argues that the variation is due to a processing effect:people prefer the ''s''-genitive with short possessors, since they put the''landmark'' relation first. If the distance between the head of the possessorand the possessum gets too large, processing benefits dictate the''of''-genitive.

John Payne's paper ''The oblique genitive in English'' deals with theconstruction in (12), the ''oblique genitive'' (OG), also referred to as the''double genitive''. Payne notes that the construction has previously beenanalysed as a variant of the ''s''-genitive, as a variant of the''of''-genitive, and an equivalent of the partitive.

(12) a friend of the Prime Minister's

Payne compares the OG to all of these correspondents in turn. In short, the OGis much more semantically restricted than the ''s''-genitive and involves aquite different pattern in the selection of determiners; the ''of''-genitivedoes not quite stand in complementary distribution with the OG either, andpatterns differently with respect to weight; and finally, the partitive alwaysinvolves anti-uniqueness, while the OG does not always do so.

The choice between the OG and the ''s''-genitive is argued to be largely amatter of information structure: in the ''s''-genitive, the referent isidentified by first processing the genitive NP, which provides an ''anchor''(Fraurud 1990) for the identification, while in the OG, the function of thatgenitive NP anchor is reduced, and processing happens largely by contextualanchors.

In the eighth paper, ''The marker of the English ‘Group Genitive’ is a specialclitic, not an inflection,'' Stephen R. Anderson develops a formal account ofthe possessive marker ‘s’ in English. Anderson establishes the feature [POSS](realized by the ''s''-genitive) as a feature which is marked on the phrasallevel (in his view: on a possessor DP residing in the specifier position of ahigher DP), then discusses two different accounts of phrasal properties. Oneis the account that he is in favour of, namely to treat the group genitive asa ''special clitic'' (Zwicky 1977). Under this account, rules modify thephonological makeup of phrases by introducing affix-like phonological content(i.e. clitics or particles) at a certain point within the phrase. The otheraccount, called ''EDGE inflection'', as put forward by e.g. Nevis (1986) andZwicky (1987), treats the group genitive as a special inflectional patternapplied at the edges of words.

Anderson mentions that both accounts produce the right facts for the EnglishGroup Genitive, but the theoretical implications and mechanisms are different:One involves a clitic as a single marker of the [POSS] feature at the edge ofthe phrase, the other realizes the feature (through intermediate constituents)on a single grammatical word, as an affix. He demonstrates that there arecases where one analysis is favourable over the other. Anderson establishesthree diagnostics for distinguishing clitics from edge inflection: selectionof certain parts of speech is more likely to apply to affixes; lexical gaps aswell as idiosyncratic shapes are more likely to occur with affixes.

Phonologically, Anderson argues that the possessive ''/z/'' is adjoined to thefinal syllable (instead of being incorporated into it). Here, possessive''/z/'' is no different from plural ''/z/'', which is also adjoined. This way,Anderson can nicely account for the data in (13) vs. (14), by saying that twoinstances of adjoined ''/z/'' are collapsed into one in (13), while in (14) weonly have a single instance of adjoined ''/z/''.

(13) anyone who likes kids' (*kids's) ideas

(14) the fuzz's old cars; at Buzz's

Liliane Haegeman discusses two kinds of prenominal possessor patterns in adialect of Dutch, in ''Two prenominal possessors in West Flemish''. The papershows that, while several other works propose a unitary account for the twopatterns, they show different syntactic features and thus cannot have anidentical syntax. The first pattern is shown in (15), referred to by Haegemanas the doubling construction (DC); the second pattern, called the ''senconstruction'' (SC), is shown in (16). In (15), the DP possessor ''Valère'' isdoubled by the possessive pronoun ''zenen'', and the latter can also occur onits own; when this is the case, as in (17), the properties of the pronoun arethe same as in the DC.

(15)(Valère) zen-en hoed(Valère) his-MSG hat'Valère's hat'

(16)Valère sen hoedValère sen hoed'Valère's hat'

(17)zen-en hoedhis-MSG hat'his hat'

The possessive pronoun displays double agreement, matching both the possessor(person, gender in the singular, number) as well as the possessum (gender,number). In the SC, ''sen'' does not agree with either. The author presentsabundant evidence against a unified approach to the two constructions,including the agreement patterns, reciprocal possessors, and adjacencyeffects.

In another descriptive paper, titled ''A Mozart sonata and the ‘Palme murder’:The structure and uses of proper-name compounds in Swedish'', MariaKoptjevskaja-Tamm describes Swedish nominal compounds where the first nominalconstitutes a personal proper name (proper name compounds; PNC). Thesecompounds are (almost) synonymous with other possessive nominals.Koptjevskaja-Tamm asks what influences the choice between the constructions,but also discusses the similarities and differences between PNCs and commonnoun compounds (CNCs).

Koptjevskaja-Tamm provides a detailed discussion of the uses of PNCs: PNCs maybe used as proper names as well as common nouns; they compete with possessiveNPs for naming streets, churches and other entities. The author notes that theheaviness of the proper noun might play a role, so that longer proper nounsappear mostly with genitives, while shorter ones appear mostly within PNCs.The connection to processing seems obvious (see also the papers by Allen,Hudson, and Payne), but as the exceptions to the rule are numerous,Koptjevskaja-Tamm notes that this must be a tendency only. The differencebetween identifying particular instances (possessive NPs) and typifiedinstances (PNCs) is general and productive in Swedish (and other Germaniclanguages, one might add).

Koptjevskaja-Tamm approaches two theoretical questions at the end of thepaper: 1) whether the issue of (non-) referentiality is relevant for theoccurrence of proper names within compounds, and 2) whether the distinctionbetween instance specification and type specification is relevant for choosingbetween PNCs and the corresponding ''s''-genitives. In answering thesequestions, she concludes that instead of treating PNCs as a singleconstruction, it might be more fruitful to split the construction apart intoseveral distinct patterns.

The last paper, ''Possessive clitics and ‘ezafe’ in Urdu'' by Tina Bögel andMiriam Butt, considers the ''ezafe'' (18), a loan construction from Persian.The paper discusses its formal properties and syntactic distribution, andprovides an analysis couched within Lexical-Functional Grammar (LFG).

(18)sahib=e takhtowner.M.SG=Ez throne.M.SG'the owner of the throne'

The authors mention that the ''ezafe'' construction has been discussed byothers (e.g., Samvelian 2007), some of whom identify it as a clitic, others asan affix and a part of nominal morphology. In Persian as well as in Urdu, the''ezafe'' construction displays a head-initial pattern, and modifiers appearto the right, which is exceptional in both languages. Moreover, the ''ezafe''always forms a unit prosodically with the head noun to its left, while at thesame time licensing modifiers to the right; syntactic function and prosodicrealization thus differ. Bögel and Butt discuss the account of Samvelian(2007), who argues that the Persian ''ezafe'' is a phrasal affix. UnlikeAnderson (this volume, 2005), who refers to phrasal affixes as ''specialclitics'', Samvelian analyses ''ezafe'' as part of word-level morphology, andnot as introduced post-lexically; Samvelian's main evidence comes from otherphrasal affixes which seem to be in complementary distribution with the''ezafe'', and thus must be generated on the same level (by the HaplologyCriterion).

Bögel and Butt challenge Samvelian's account and argue that different groupsof phrasal affixes can belong to different classes, and that the HaplologyCriterion must not hold in the morphological component, but may apply in thephonological/prosodic part of the grammar. They argue that the Urdu ''ezafe''behaves like a clitic in many respects: e.g., it is separable from its hostusing parentheticals, it can take scope over noun conjunction, it does notdisplay morphophonological idiosyncrasies. It also has some non-clitic-likeproperties; for example, it displays a high degree of lexical selection, onlyoccurring with nouns of Persian origin as its head. The authors conclude thatUrdu ''ezafe'' should be analysed as a clitic (a phrasal affix). Their LFGanalysis involves separate modules of grammar, taking into account itsmis-alignment: while it is a functional head selecting a modifier to itsright, prosodically it attaches to the word on its left.


Shortcomings of the volume as a whole are of a formal nature. Some examples insome of the papers lack glosses, which I am not sure is an error on theauthors' or on the publisher's part. The numbering of the examples is also offin some cases. In addition, some cited references are not included in thebibliography at the end of the volume (I have found at least 4 such instancesacross all the papers).

There is also some variation in the volume regarding the terminology ofclitics. Anderson uses the term ''special clitic'' for the English groupgenitive in the sense coined by Zwicky (1977). Anderson in his earlier workused the term ''phrasal affix'' (Anderson 1992) which turns out to beequivalent to Zwicky's ''special clitic''. Now, however, he uses Zwicky'sterm, which is why the term ''phrasal affixes'' does not feature in his paperin this volume. Bögel & Butt in their paper, referring to Anderson's work, usehis earlier terminology and talk about ''phrasal affixes''.

The paper by Börjars et al. may in part provide the answer for such issues.Anderson as well as Bögel and Butt in their papers acknowledge that themarkers they analyse (English group genitive and Urdu ''ezafe'') display mixedproperties of affixes and clitics, but both papers analyse the markers asclitics. A dichotomy ''affix'' vs. ''clitic'' may turn out to be anoversimplification that does not do justice to the mixed properties of suchitems, and Börjars et al. instead suggest a scale of grammatical categorieswith a ''clitic end'' and an ''affix end''. While this is an interestingproposal, the exact makeup of the proposed scale is left for further research.

The volume provides an interesting perspective on possessive alternations,which is the key theme of several papers. Throughout the papers, the featuresanimacy, weight and topicality/discourse status crop up, and clearcorrelations are established between these features and the choice of aparticular possessive encoding strategy. The question arises whether thesetendencies form part of the grammar, or whether they belong in a separatecomponent capturing language use; O'Connor et al. answer this by looking atlanguages where those factors are implicated in categorical distinctionsbetween separate constructions, and thus clearly form part of the grammar. Itcan therefore be concluded that the statistical patterns displayed e.g. byEnglish actually form part of the grammar and need to be represented in amodel of grammar. A question I would add is in how far the features animacy,weight and topicality hold up in a cross-linguistic study of possessivepatterns, or whether there are more features that involve categoricaldistinctions and/or statistical preferences in other languages.

All in all, the volume is essential reading for any linguist interested in themorphosyntactic realization of possession. While the overall focus is clearlyon English, this is not necessarily a negative: it enables the volume toapproach the various issues in English from several distinct angles, whilemaintaining a manageable set of data. Empirical-statistical, cognitive andtheoretical-explanatory accounts add up to render a rather complete picture ofthe English possessive constructions from a synchronic as well as from adiachronic perspective. In addition, descriptive papers team up to provideinsights into other languages' possessive structures as well.


Anderson, Stephen R. 1992. A-Morphous Morphology. Cambridge: CUP.

Anderson, Stephen R. 2005. Aspects of the Theory of Clitics. Oxford: OUP.

Fraurud, Kari. 1990. Definiteness and the processing of noun phrases innatural discourse. Journal of Semantics 7: 395-433.

Hudson, Richard. 2010. Word Grammar and Cognition. Cambridge: CUP.

Lehmann, Christian. 1995. Thoughts on Grammaticalization. Munich: Lincom.

Nevis, Joel A. 1986. Finnish Particle Clitics and General Clitic Theory[Working Papers in Linguistics 33]. Columbus OH: Dept. of Linguistics, TheOhio State University.

Samvelian, Pollet. 2007. A (phrasal) affix analysis of the Persian Ezafe.Journal of Linguistics 43: 605-645.

Zwicky, Arnold M. 1977. On Clitics. Bloomington IN: Indiana UniversityLinguistics Club.

Zwicky, Arnold M. 1987. Suppressing the Z's. Journal of Linguistics 23:133-148.


Sebastian Sulger is a Ph.D. student at the Department of Linguistics,University of Konstanz. He is interested in case, argument structure, andgrammar interfaces (morphology-syntax, syntax-semantics), as well as areaswithin computational linguistics. He has published papers on possession inHindi/Urdu, nominal argument structure, and copula constructions. Hisinterests in computational linguistics include grammar development,treebanking, and multiword expressions.

Page Updated: 19-Sep-2013