reformulating dp analysis

25
© 2008 The Author Journal Compilation © 2008 Blackwell Publishing Ltd Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.x Reformulating the Determiner Phrase Analysis Judy B. Bernstein* William Paterson University of New Jersey Abstract Developed in the 1980s, the Determiner Phrase (DP) analysis stimulated a lot of interest in the internal structure of nominal phrases and in the study of corre- spondences between nominal and clausal structure. Research across a range of languages has uncovered correspondences in areas such as agreement morphology, syntactic movement, and argument structure, some of which are reviewed here. Nominal phrases and clauses also match up in terms of their semantic function: both can serve as predicates or arguments. The issue of exactly what distinguishes a nominal argument from a nominal predicate has received a lot of attention, leading in particular to proposals about the underlying role of the definite article (and determiner elements in general). This article reviews some of these issues and proposals and suggests a reformulation that appeals to the feature ‘person’, a feature found in both the nominal and clausal domains. 1. Introduction Twenty-one years after Abney’s (1987) influential work on the internal structure of nominal phrases, this article reviews some of the original arguments for Abney’s DP (Determiner Phrase) analysis, examines several proposals that grew out of the analysis, and offers a somewhat novel idea for the reformulation of the core property of the functional head D. During the 1980s, important work by Chomsky (1986) developed the idea that verbs project functional structure [i.e., Inflection Phrase (IP), and complementizer phrase (CP)], leaving open the question of if, and then how, this idea could be applied to nouns. Over the course of the decade, other researchers took up the topic of the internal structure of nominal phrases. Abney (1987), Szabolcsi (1983 and later work) and many others demonstrated that nominal phrases display clausal properties and proposed that nouns, like verbs, must also project functional structure. Among the properties of clauses that may also be found in nominals, we can observe: (i) (identical) agreement morphology in nominals and clauses, in some cases licensing ‘pro-drop’ (non-expression of a pronominal subject); (ii) instances of wh-movement in nominals and clauses; and (iii) parallel argu- ment structure in nominals and clauses. I briefly discuss and provide evidence of these properties here (some material drawn from Bernstein 2001).

Upload: ramlohani

Post on 09-Sep-2015

11 views

Category:

Documents


1 download

DESCRIPTION

DP hypothesis

TRANSCRIPT

  • 2008 The AuthorJournal Compilation 2008 Blackwell Publishing Ltd

    Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.x

    Reformulating the Determiner Phrase Analysis

    Judy B. Bernstein*William Paterson University of New Jersey

    AbstractDeveloped in the 1980s, the Determiner Phrase (DP) analysis stimulated a lot ofinterest in the internal structure of nominal phrases and in the study of corre-spondences between nominal and clausal structure. Research across a range oflanguages has uncovered correspondences in areas such as agreement morphology,syntactic movement, and argument structure, some of which are reviewed here.Nominal phrases and clauses also match up in terms of their semantic function:both can serve as predicates or arguments. The issue of exactly what distinguishesa nominal argument from a nominal predicate has received a lot of attention,leading in particular to proposals about the underlying role of the definite article(and determiner elements in general). This article reviews some of these issuesand proposals and suggests a reformulation that appeals to the feature person, afeature found in both the nominal and clausal domains.

    1. Introduction

    Twenty-one years after Abneys (1987) influential work on the internalstructure of nominal phrases, this article reviews some of the originalarguments for Abneys DP (Determiner Phrase) analysis, examines severalproposals that grew out of the analysis, and offers a somewhat novel ideafor the reformulation of the core property of the functional head D.

    During the 1980s, important work by Chomsky (1986) developed theidea that verbs project functional structure [i.e., Inflection Phrase (IP), andcomplementizer phrase (CP)], leaving open the question of if, and thenhow, this idea could be applied to nouns. Over the course of the decade,other researchers took up the topic of the internal structure of nominalphrases. Abney (1987), Szabolcsi (1983 and later work) and many othersdemonstrated that nominal phrases display clausal properties and proposedthat nouns, like verbs, must also project functional structure. Among theproperties of clauses that may also be found in nominals, we can observe:(i) (identical) agreement morphology in nominals and clauses, in somecases licensing pro-drop (non-expression of a pronominal subject); (ii)instances of wh-movement in nominals and clauses; and (iii) parallel argu-ment structure in nominals and clauses. I briefly discuss and provide evidenceof these properties here (some material drawn from Bernstein 2001).

  • 2 Judy B. Bernstein

    2008 The Author Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.xJournal Compilation 2008 Blackwell Publishing Ltd

    Hungarian, a nominative/accusative language, displays identical agree-ment affixes on nouns and verbs. The data in (1) and (2) (drawn fromSzabolcsi 1983 and also discussed in Abney 1987) illustrate the Hungariannominal and clausal agreement patterns. In (1), Case is expressed on thepossessor and the head noun agrees with the possessor in person andnumber. In (2) the sentential subject is also marked for Case and the verbdisplays number and person agreement with the subject.

    Although Turkish nominal agreement morphology is not identical in formto the corresponding verbal agreement morphology, Kornfilt (1984) hasshown for Turkish that both nominal and verbal agreement morphologylicense pro-drop.

    Horrocks and Stavrou (1987) show that in Greek, wh-movementapplies in a parallel fashion in nominal phrases and clauses [data fromHorrocks and Stavrou 1987, their (13) on p. 89]:

    In (3a), the echo question represents the assumed underlying position ofthe wh-word tinos. In (3b), tinos has raised to a position peripheral to the

    (1) a. az n- vendg-e-m (Hungarian)the I-nom guest-poss-1sgmy guest

    b. a te- vendg-e-dthe you-nom guest-poss-2sgyour guest

    c. (a) Mari- vendg-e-(the) Mary-nom guest-poss-3sgMarys guest

    (2) Mari- alud-t- (Hungarian)Mary-nom sleep-past-3sgMary slept.

    (3) a. to vivlio tinos (Greek)the book who-genwhose book

    b. tinos to vivliowho-gen the bookwhose book

    (4) a. ekane ti (Greek)did-3sg whathe did what

    b. ti ekanewhat did-3sgwhat did he do

  • 2008 The Author Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.xJournal Compilation 2008 Blackwell Publishing Ltd

    Reformulating the Determiner Phrase Analysis 3

    nominal phrase to vivlio (the book). In (4), we see the same pattern in theclause: in (4a) we see the underlying position of the wh-word ti (what)and in (4b) the raised position.

    Both nominals and clauses admit internal and external arguments, asillustrated in the examples in (5), which are adapted from Chomsky(1970):

    The sentence in (5a) and the derived nominal in (5b) display both aninternal argument (theme), Carthage, of the lexical head (the verb destroyor the noun destruction), and an external argument (agent), Rome. In bothdomains, the agent subject may bind a reflexive anaphor:

    On the topic of argument structure in nominals, Cinque (1980) demon-strated that only the highest argument in the nominal domain can bepossessivized, and work by many authors since Chomsky has examinedissues related to the argument structure of nominals as well as the natureof the lexical stem (nominal, verbal, or underspecified) in a derivednominal like destruction (see, among many others, Grimshaw 1990; Giorgiand Longobardi 1991; Picallo 1991; Valois 1991; Marantz 1997; Alexiadou2001).

    Analysis of these and other properties within the nominal domain andan emerging recognition of general similarities between nominal phrasesand clauses led to proposals for parallel syntactic structures for the twophrase types. Abney (1987) proposed that DP is the maximal functionalcategory projected by the lexical noun and hosts determiners (e.g., definitearticles). Abney reasoned that determiners in the nominal domain areanalogous to modals in the clausal domain, each lexicalizing a functionalhead projected by the lexical noun or verb (Grimshaw 1991). Abneysresearch on the syntax of DP led to work focusing on DP properties atthe syntaxsemantics interface, such as issues of reference and definiteness(e.g., Vergnaud and Zubizarreta 1992; Longobardi 1994; Zamparelli 1995).Since then, D(P) has also been argued to be relevant to properties such asCase (e.g., Giusti 1995), deixis (e.g., Giusti 1993; Klinge 2008), and specif-icity (e.g., Campbell 1996; Ihsane and Puskas 2001).1

    In this article, I focus on another correspondence between nominalsand clauses, namely, the predicate vs. argument distinction. I re-examinesome of the arguments for D as the functional head relevant to thisdistinction (contrasting it with the parallel distinction in the clausal

    (5) a. Rome destroyed Carthageb. Romes destruction of Carthage.

    (6) a. Romei destroyed itselfi.b. Romesi destruction of itselfi.

  • 4 Judy B. Bernstein

    2008 The Author Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.xJournal Compilation 2008 Blackwell Publishing Ltd

    domain), a proposal I ultimately adopt. I consider whether nominal argu-ments can be distinguished from nominal predicates in terms of interpretivefeatures. I observe, as many others have, that no single interpretive propertyrepresents the fundamental nature of D(P), and I go on to suggest that theinterpretive properties can be derived from another more fundamentalproperty. Specifically, I develop the idea that D (and DP) is best viewedas the locus of person. This property, I argue, underlies whether a givennominal is referential, definite, and/or an argument (or a predicate).Because of space limitations, I will not review here proposals forfunctional projections intervening between NP and DP (e.g., NumberP,GenderP, QuantifierP, and CaseP), but I will occasionally touch on issuesrelevant to these proposals.

    The article is organized as follows. In Section 2, I review some of theoriginal arguments for DP, including its relevance to the predicate/argu-ment distinction. In this regard, I consider the role of D in nominalphrases and identify it, following others before me, as the functional headrelevant to establishing reference and anaphoric dependencies. In Section3, I develop the idea that person is the core property of D and thefeature underlying reference, providing some support for the idea fromconstructions involving personal pronouns, definite articles, and propernames. Section 4 offers some concluding remarks.

    2. Predicates and Arguments

    Abney (1987) and others developed the influential idea of a nominalfunctional projection, DP, analogous to Chomskys (1986) clausal functionalprojection, CP:2

    For a language like English, Abney proposed that determiner elements,such as the definite article in examples like (9a), are the lexical instantiationof the functional head D [see (9b)], in the same way that modals, as tense/aspect markers in English, can be taken as the lexical instantiation of thefunctional head I.3 Following insights from work by Postal (1966), Abneyfurther claimed that English pronouns in examples like (10) also occupy D.

    (7) clauses: [CP [C C [IP [I I [VP [V V ]]]]]](8) nominals: [DP [D D [NP [N N ]]]]

    (9) a. the linguist(s)b. [DP [D the [NP [N linguists ]]]]

    (10) a. we linguistsb. you linguistsc. [DP [D we/you [NP [N linguists ]]]]

  • 2008 The Author Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.xJournal Compilation 2008 Blackwell Publishing Ltd

    Reformulating the Determiner Phrase Analysis 5

    The structural distinction between NP and DP also maps onto a semanticand functional distinction between predicates on the one hand, and argu-ments, those expressions that satisfy the thematic requirements of lexicalheads, nouns, and verbs. Longobardi (1994), building on work by Szabolcsi(1987), Stowell (1989), and others, argued convincingly that an argumentmust be introduced by D. Put another way, D converts a predicate (i.e.,NP) into an argument (i.e., DP; Stowell 1989: 248). The Italian examplesin (11) illustrate the distinction [examples from Longobardi 1994 (19)].

    The determinerless nominal predicate medico of (11a), an NP according tothis line of reasoning, is unable to serve as the head of a relative clause(introduced by che who). Presumably this is because the head of a relativeclause must be an argument and the bare noun medico cannot satisfy thisrequirement. The grammaticality of un medico as the head of a relativeclause in (11b) strongly suggests that this phrase is a (DP) argument.Under this general approach, parametric differences among languageswith respect to obligatory vs. optional appearance of overt lexical materialin D reduce to featural differences of this functional head.

    Still pursuing the idea of parallels between nominals and clauses, we canask if the predicate/argument distinction just examined in the nominaldomain is relevant at the clausal level as well. Recall that under theapproach adopted here, the lexical projection NP corresponds to a pred-icate and the functional projection DP to an argument. Within thedomain of the clause, we might expect that IP corresponds to a (clausal)predicate and CP to a (clausal) argument. This prediction is borne outand can be illustrated straightforwardly by the English examples in (12).

    The matrix sentence in (12a) (a proposition but not an argument) wouldcorrespond to IP. When the same proposition appears as the sententialsubject in (12b), ungrammaticality results. This is because the propositionin this case, as a sentential subject of the verb surprise, is not in theform of an argument but should be. Insertion of the subordinator that(Szabolcsi 1992: 130) converts the IP predicate into a CP argument,resulting in a licit expression in (12c). In a parallel fashion, according toSzabolcsi, in the nominal domain definite articles function as subordinators.

    (11) a. Gianni medico (*che si cura davvero dei suoi pazienti).John is doctor (who really cares for his patients)

    b. Gianni un medico (che si cura davvero dei suoi pazienti).John is a doctor (who really cares for his patients)

    (12) a. [IP Tiffany married Louise].b. *[IP Tiffany married Louise] didnt surprise me.c. [CP That [IP Tiffany married Louise]] didnt surprise me.

  • 6 Judy B. Bernstein

    2008 The Author Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.xJournal Compilation 2008 Blackwell Publishing Ltd

    The approach pursued by Longobardi (following Stowell, Szabolcsi)with respect to the NP/DP distinction and its relation to argumenthoodhas been an influential one, and I adopt it here as well as in previouswork. However, the literature also features another important analysisworthy of consideration. Chierchia (1998) takes the predicate/argumentdistinction to vary cross-linguistically: some languages allow NPs to bearguments (e.g., in English) and in other languages only DPs can bearguments (e.g., in Italian). So whereas a Longobardi-style approach takesthe locus of parametric variation to reside in the feature content offunctional heads like D (an idea about parametric variation going back toBorer 1984), according to Chierchias Nominal Mapping Parameter, thelocus of parametric variation is not with functional (or lexical) heads perse, but rather in the type of phrase that functions as a predicate or anargument in a particular language.4 I come back to these proposals verybriefly at the end of the article (Section 3.4), where I share some specu-lations about how they might be complementary.

    Stowells (1989) and Longobardis (1994) early work on the predicate/argument distinction (NPs are predicates, DPs are arguments) did notcontemplate cases involving non-argument DPs (see Mandelbaum 1994for some relevant discussion). In this regard, vocative expressions, stand-ardly taken to be predicative (see Szabolcsi 1987; Longobardi 1994), areworthy of examination in light of recent evidence that they may in factinvolve DP structure. Consider the following examples:

    Example (13a) with bare plural kids! is expected as a vocative; the absenceof an overt article is consistent with a predicate (NP) structure.5 Corre-spondingly, the impossibility of the kids! in (13b) as a vocative in Englishis consistent with the basic idea that the definite article corresponds to Dand vocatives, as non-arguments, do not involve a DP layer.6 The examplesin (13c,d) with a proper name (Patricia) and a pronoun (you) are moreproblematic, because these expressions may also function as arguments,suggesting that they are DPs.7 The internal structures adopted for theseexamples [recall (10) and see (14)(15)] are consistent with a DP structure:the pronoun you in (13d) occupies D and the proper name Patricia in (13c)is underlying in N and overtly raises to D in certain languages.

    Observing that vocative expressions in some languages may includerelative clauses, demonstratives, and even definite articles, Crisma (1997:31, 135137) in fact proposes that vocatives may have a D position as partof their structure (although the D may not always be lexicalized). Given

    (13) a. Kids! Come here!b. *The kids! Come here!c. Patricia! Come here!d. You (kids)! Come here!

  • 2008 The Author Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.xJournal Compilation 2008 Blackwell Publishing Ltd

    Reformulating the Determiner Phrase Analysis 7

    this broadening of the characterization of vocatives, we are led to theconclusion that at least some non-arguments may include a DP layer. Mustthe Stowell/Szabolcsi intuition distinguishing predicates (as NPs) and argu-ments (as DPs) be abandoned? Not entirely. In fact, a re-characterizationof the generalization still evokes the original intuition. Crisma (1997: 1389;see also Longobardi 2008) partitions entities (here, nominal phrases) intoproperties and individuals (see Carlson 1977; Chierchia 1984). Veryloosely speaking, properties characterize entities in their use as predicates,typically taking the form of adjectives (e.g., tall, hot) and nouns (e.g., gold,John).8 Properties are therefore entities that do not denote individuals, whichtypically correspond to arguments.9 What of vocatives? Crisma reasons that,although not arguments, vocatives share a property with them: both typesof expressions denote individuals.

    Where does this leave us with respect to DP vs. NP and arguments vs.predicates? What emerges from the above considerations, and whatCrisma proposes, is that both arguments and vocatives denote individualsand so involve DP. Properties do not denote individuals and involve NP,never DP. The original insight about arguments involving DPs (not NPs)is preserved.

    Crismas expanded structure for vocatives (as DPs) now makes senseof examples in (13c,d) with proper name and pronoun: in both examples,a DP layer is projected despite the fact that these vocative expressions arenon-arguments. Now, however, the ungrammaticality of an example like(13b) with the definite article receives no immediate explanation,although it could have been excluded on the Longobardi or Szabolcsi/Stowell approach to vocatives as NP predicates. Why would proper namesand personal pronouns, both possible as vocatives and now seen to argu-ably involve DP, pattern differently from a nominal expression with adefinite article, barred in vocative expressions in many languages? Thesolution I propose to distinguish the grammatical (13c,d) from theungrammatical (13b) will require a further consideration, which I take upin Section 3.

    Having reviewed the proposal that DPs correspond to arguments andnow also to vocatives (as individuals), and NPs to properties, I turn nextto an examination of some issues relevant to the identification of theproperty of D(P) responsible for distinguishing individuals from properties.

    2.1 THE ROLE OF D

    Whether all languages distinguish DPs and NPs (under a Longobardi-styleapproach) or only some languages do so (under a Chierchia-styleapproach), it seems reasonable to assume that at least in some languages anominal phrase may not function as an argument, as a type of individual-denoting expression, unless it is a DP. Italian is apparently such a language,as the examples in (14) illustrate (examples from Longobardi 2008):

  • 8 Judy B. Bernstein

    2008 The Author Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.xJournal Compilation 2008 Blackwell Publishing Ltd

    The contrast between the grammatical (14a) and ungrammatical (14c)illustrates that, at least for Italian, the definite article is necessary, presum-ably because antica Roma, as an argument, must be a DP and cannot bewithout a definite article.

    The grammaticality of (14b) is interesting, because the subject Romaantica lacks an overt determiner and yet is grammatical. This suggests thatthe requirement for Italian that D be filled can be satisfied in another way.For examples like (14b), Longoardi (1994 and later work) has developedthe idea that the proper name Roma overtly raises from N to D, the locusof reference. The contrast between the internal structures of lantica Romaand Roma antica would be represented as follows, abstracting away fromthe position of the adjective:

    Example (15a) shows the base position of the proper name to be N and(15b) shows the raised position to be D, after crossing over the adjective.

    The Italian pattern contrasts with the equivalent English facts in (16),where the proper name Rome appears in the lower (N) position in (16a),even though no definite article appears, not the higher (D) one in (16b).

    Longobardi (1994) postulated that the movement of proper names to D,a general requirement of arguments cross-linguistically, is covert inEnglish. A Chierchia-style approach would take the same facts to indicatethat English allows NPs (as well as DPs) to be arguments, unlike Italianarguments that must be DPs.10

    Assuming that the relevance of D to argumenthood is ultimately sup-ported, we can ask exactly what D (e.g., the definite article) contributesto the nominal phrase. For many authors, a nominal phrase is definite byvirtue of the presence of a definite article, regardless of any effect thearticle may have on interpretation. Other authors focus on the possibleinterpretive effects of the definite article (or D more generally), amongthem definiteness, specificity, referentiality, identifiability, and deixis. An

    (14) a. Lantica Roma era una citt potente. (Italian)the ancient Rome was a city powerful

    b. Roma antica era una citt potenteRome ancient was a powerful city

    c. *Antica Roma era una citt potente.

    (15) a. [DP la (antica) [NP Roma ]]b. [DP Romai (antica) [NP ti ]]

    (16) a. Ancient Rome was a powerful city.b. *Rome ancient was a powerful city.

  • 2008 The Author Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.xJournal Compilation 2008 Blackwell Publishing Ltd

    Reformulating the Determiner Phrase Analysis 9

    extensive review of these proposals is beyond the scope of this article (butsee Lyons 1999 for extensive discussion and Alexiadou et al. 2007: 578for a survey of the issues and proposals).11 However, a quick examinationof some simple examples illustrates that no single interpretive propertycharacterizes D, even though all the examples include a definite article.For each example in (17), the accompanying parenthetical indicates themost salient interpretation for the relevant nominal phrase, but not alwaysthe only one [(17g) is from Dchaine and Wiltschko 2002, their (61b)].

    My approach to the definite article essentially follows Vangsnes (1999)idea that determiners (e.g., definite articles) do not have semantic/pragmatic features of their own. Dchaine and Wiltschkos observationsabout the ambiguity of (17g) with a definite article are analogous to themore general observation I make here. Their analysis of the ambiguitydistinguishes two different internal structures, one corresponding to eachof the interpretations. I return to their proposal in Section 3.2 within thecontext of my own proposal about these examples.

    2.2 ESTABLISHING REFERENCE AND REFERENTIAL DEPENDENCIES

    One prominent idea found in the literature is that D is relevant to refer-ence. Building on early Minimalist ideas of Chomsky (1993) and the workof Szabolcsi (1983 and later work) and Stowell (1989), Longobardi (1994,1996) proposed that a reference feature is encoded in D and that argumentnominal phrases must check this feature, subject to parametric variation

    (17) a. Did you go to the beach when you visited Croatia?

    (non-specific)

    b. I protected my eyes from the sun.

    (unique)

    c. The lion is a dangerous animal. (generic)d. I picked up the car from the

    mechanic.(possessive)

    e. Je me suis cass le bras. (French) (inalienable possession)

    I CL-me have broken the arm (Spanish)I broke my arm.

    f. Sal con la mam.12 (possessive)went-out with the motherI went out with my mother.

    g. Jean aime le vin. (French) (generic or definite)

    Jean likes (the) wine.

  • 10 Judy B. Bernstein

    2008 The Author Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.xJournal Compilation 2008 Blackwell Publishing Ltd

    as to whether the feature is checked overtly (in the syntax), as in Italian,or covertly (post-syntactically), as in English.13

    Proper names, as unambiguously referential expressions, are thereforean important object of study. In this regard, the examples from (14) serveto illustrate the well-known analysis Longobardi developed for Italian:proper names raise to D overtly in order to check their [+referential]feature in (14b); otherwise, an overt D element (i.e., the definite article)occupies the D position in (14a).

    Pereltsvaig (2007) adopts the general Stowell/Longobardi idea that D,not N, is relevant to reference. In particular, she develops the idea that Dcarries a referential index, which she takes to be an independent phifeature and so syntactic in nature.14 In order to control the number ofpossible referential indices in the lexicon, corresponding to a potentiallylimitless number of entities, Pereltsvaig invokes processing limitations andsuggests that indices may be reused in different discourses (2007: 6970).15

    If Pereltsvaigs appeal to processing limitations may serve to restrain thenumber of referential indices from reaching infinity, why do the core phifeatures (number, gender, person) never seem to require such restraining?As far as I know, the possible values for these core phi features are quitelimited (e.g., singular, dual, and plural number), even consideringlanguages with a rich array of gender classes.

    I will not follow Pereltsvaigs proposal that a referential index is anindependent phi feature, but instead I will take reference to be derivablefrom the core feature person (see Section 3). As MacLaughlin (1997: 101,footnote 2) points out, person plays a crucial role in syntactic phenomenarelated to reference, such as anaphora, verbal agreement, and the licensingof null arguments. In other words, in my view these syntactic phenomenarely on the feature person, as does the computation of reference.Although I do not adopt Pereltsvaigs proposal about a referential index asan independent phi feature, I do accept the more general idea that D isthe head relevant to interpreting the reference of a nominal phrase.

    Unlike approaches of Longobardi and Pereltsvaig, Baker (2003, see also2008) argues that it is the noun that carries a referential index. Under thisapproach and thinking of the internal structure of the Italian examples in(14) with proper names, the reference (or referential index) of a nominalexpression could plausibly be established via the base position of theproper name, that is, in N. If reference is determined by or in N, however,another explanation is needed for why the D position must be filled inItalian, but apparently not in English. The explanation could still reduceto the individuals vs. properties distinction discussed above, with thequestion left open about the role of D, perhaps playing no role inreference at all.

    In order to tease apart the relevance of D vs. N in establishing referen-tial dependencies, I return to complex nominals, those involving personalpronouns plus nouns. I repeat here the examples and structures from (10):16

  • 2008 The Author Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.xJournal Compilation 2008 Blackwell Publishing Ltd

    Reformulating the Determiner Phrase Analysis 11

    Recall that I adopted (18c) as the representation of the internal structurefor these types of examples. Let us next examine the referential depend-encies admitted between these complex nominals and anaphoric expres-sions. Consider now the examples in (19).

    Here we observe some evidence that, at least for certain languages, Drather than N is the relevant head for establishing referential dependencies:the personal pronoun we in (19a), but not the noun linguists in (19b), maybind a pronominal anaphor.17

    Now although examples like (19) are consistent with Longobardis pro-posal about the relevance of D to reference, they raise another set ofquestions. Until this point, I have focused almost exclusively on definitearticles, because these elements are the prototypical candidates for D. Ifpersonal pronouns also correspond to D recall Postals (1966) intuition we need to re-evaluate the characterization of D, which was developedprimarily around examples with definite articles [recall examples in (17)].

    The idea I develop in the next section in fact crucially relies on exampleslike (18) and (19). Specifically, I will appeal to the person feature of thesepersonal pronouns and suggest that this property characterizes D; I willfurthermore suggest that person is necessary for establishing referentialdependencies.

    3. D Encodes Person

    In this section, I explore and develop the idea that person is the corefeature of D(P), building on recent work by Longobardi (2008) and Bern-stein (2008).18 The intuition is that the individuals (DP) vs. properties(NP) distinction, referential dependencies, and even definiteness rely onperson. Person, which is lexicalized in many languages through personalpronouns (also reflexives and other anaphors) and also is displayed in theclausal domain in the form of verbal agreement marking, identifies anominal phrase as referring to the participants in the conversation (i.e.,first person I, me; second person you) or not (i.e., third person they, themand also lexical nominal phrases). So person, like other core phi featuressuch as number, has a limited number of values: 1st, 2nd, and 3rd person(some authors arguing for 4th, 5th, and 6th).

    (18) a. we linguistsb. you linguistsc. [DP [D we/you [NP [N linguists ]]]]

    (19) a. Wei linguists like to study ourselvesi/ouri own languages.b. *We linguistsi like to study themselvesi/theiri own languages.

  • 12 Judy B. Bernstein

    2008 The Author Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.xJournal Compilation 2008 Blackwell Publishing Ltd

    I begin the section by examining the status of first- and second-personpronouns in predicate and argument expressions, focusing the discussionaround familiar English examples involving the already-seen pronoun-plus-noun construction (we linguists) and then simple personal pronouns(we). For both sorts of pronominals, I argue that the D position is relevantto interpretation (reference) through the grammatical feature person.Next, I look at so-called definite nominal phrases, those displaying adefinite article (e.g., the), a demonstrative (e.g., this), or even a third-personpronoun (e.g., they), all of which I take to share a feature, third person,expressed through a th- morpheme in D. I will capitalize on the homog-enous character of English th- third-person forms (e.g., the, they, and this)and on the fact that they are not always definite in interpretation, as wesaw above in (17) with definite articles. In a section on proper names,I interpret the fact that such elements require or admit a definite articlein many languages as support for Longobardis (1994) idea that they areunderlyingly of category N, like common nouns. Because no semantic orpragmatic information is conveyed by the so-called expletive definitearticle, I speculate that its function is to supply a person feature to DP. Atthe end of the section, I briefly discuss correspondences between theexpression of person in the nominal and clausal domains.

    3.1 PERSONAL PRONOUNS

    Recall Postals intuition about we linguists, that we corresponds to a deter-miner, taken to be D since Abney (1987).19 Examples like (19) furtherillustrated that we, rather than linguists, serves as the antecedent for apossessive or reflexive anaphor, supporting the general hypothesis that Dencodes person and that person is relevant for establishing referential/anaphoric dependencies.

    Related to the Postal examples with pronouns and determiners arethe vocative facts discussed in Section 2. The examples are repeated herein (20).

    Recall that I followed Crismas idea that vocatives may include a DPstructural layer. This provides an analysis of proper names and personalpronouns in these constructions. An example like (20b) with the definitearticle was left unexplained. I appeal here to another property of vocativeexpressions, namely, a required second-person interpretation that is eitherlexically expressed or implicit (see Szabolcsi 1987). I repeat the examples

    (20) a. Kids! Come here!b. *The kids! Come here!c. Patricia! Come here!d. You (kids)! Come here!

  • 2008 The Author Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.xJournal Compilation 2008 Blackwell Publishing Ltd

    Reformulating the Determiner Phrase Analysis 13

    from (20) above in (21), representing in upper case letters (in parentheses)what I take to be unpronounced material.

    I suggest that the common noun and proper name in (21a,c) have animplicit second person pronoun, that this YOU occupies D, and that thecommon noun or proper name occupies N.20 In (21d), the pronoun youwould again occupy D, and N would contain an unpronounced noun. Allthe examples in (21) now have the same internal structure and all involvesecond person, except for (21b). The ungrammaticality of *the kids! as avocative is now accounted for by the absence of second person, whichshould be explicitly or implicitly in D. Moreover, and as I develop in thenext section, the definite article the in D identifies the DP vocative asthird person.

    We have now seen two sorts of constructions involving personal pro-nouns plus nouns, the pronouns taken to occupy D and to supply personand the nouns to occupy N. If these personal pronouns are D elementsin vocative expressions like you guys! and also in arguments like we linguists,what of simple personal pronouns like we, whose referential and argumentstatus are equivalent to those with accompanying nouns (explicit or implicit)?

    Work by Cardinaletti (1994) and Longobardi (1994) illustrates that personalpronouns used in non-argument function as exclamations arguably lack adeterminer layer and correspond to N underlyingly:

    These same pronouns used in argument function must raise to D, judgingby the impossibility of a preceding definite article (23a) or numeral (23b).The internal structure proposed for the grammatical we two (i.e., pronounpreceding the numeral) is provided in (23c).

    In the above discussion, we have seen that in two sorts of expressions Dis the relevant position, underlying or derived, for personal pronouns.

    (21) a. (YOU) kids!b. *The kids!c. (YOU) Patricia!d. You (KIDS)!

    (22) a. poor [NP us]!b. poor [NP me]!c. poor [NP you]!

    (23) a. *the [NP we/us ]b. *two [NP we/us]c. [DP we/usi . . . (two) . . . [NP ti ] ]

  • 14 Judy B. Bernstein

    2008 The Author Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.xJournal Compilation 2008 Blackwell Publishing Ltd

    Why would that be? From the perspective developed here, person-bearingelements are associated with D precisely because person is its core property.I take this property to underlie the referential and anaphoric propertiesof nominal arguments, and now also vocatives, although I have notarticulated the specifics of such an analysis. Important questions remain,such as: What is the mechanism responsible for the raising of personalpronouns to D? Is it a feature of the pronoun itself that compels it toraise or is it a feature of the functional head D? If the latter, does Dhave a probe for person (see Chomsky 2000, 2001 for the idea of func-tional heads as probes for features)? These questions must remain open fornow.

    In the next section, we will see how the basic idea that D encodesperson may generalize to so-called definite expressions, particularly thoseinvolving definite articles but also (other) third-person forms.

    3.2 SO-CALLED DEFINITE EXPRESSIONS

    As we saw in Section 2 [examples in (17)], the presence of the definitearticle does not always yield a definite interpretation, a point made bymany authors including Dchaine and Wiltschko (2002: 4289). Theyobserved that French l- (definite article) forms do not always involve adefinite interpretation and in many cases may yield either a generic inter-pretation or a definite one, as illustrated by the two possible glosses for(24):

    Their approach to pronouns and determiners cross-linguistically distin-guishes three basic types, corresponding either to the syntactic categoryNP (true predicates), DP (true arguments), and P (intermediate status).Under their analysis, le vin would correspond to a DP under the definiteinterpretation of (24) (this particular wine) and to P under the genericinterpretation (wine in general). Whether or not le vin is definite, itfunctions as an argument and so could not correspond to NP under theiranalysis. The internal structures they assign, corresponding to the possibleinterpretations, are the following [their (i) and (ii) in Dchaine and Wilt-schko 2002, footnote 21 on p. 430]:

    (24) Jean aime le vin. (French)Jean likes (the) wine.

    (25) a. [D [ le [NP vin]]]the wine (definite)

    b. [ le [NP vin]]wine (generic)

  • 2008 The Author Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.xJournal Compilation 2008 Blackwell Publishing Ltd

    Reformulating the Determiner Phrase Analysis 15

    According to Dchaine and Wiltschkos analysis, the definite descriptionin (25a) does not derive directly from the presence of the article, butrather from the presence of an unfilled D, which they take to be necessaryfor the relevant interpretation; the generic interpretation corresponds toa structure lacking a DP layer in (25b), but still an argument because ofits P status. Their analysis, which is meant to generalize across Romancelanguages, includes the idea that definite articles (like pronouns) maycorrespond to different functional categories across languages.

    I follow a somewhat different approach to the English th- and Romancel- article and pronominal forms (building on Bernstein 2008; see alsoRadford 1997: 187ff.; Lyons 1999: 27), and treat them as a homogenousgroup. In particular, I take the definite article in languages like Englishand French to consist of a th- or an l- third-person morpheme.21 In thisway, both of the expressions with definite articles in (25) may be argu-ments by virtue of the third-person morpheme, without definiteness as anecessary corollary and without positing additional functional structure todistinguish (25a) from (25b). The same analysis can apply to demonstra-tives, which are not always deictic. The example in (26a) is non-specificand the one in (26b) is deictic (discussed in Bernstein 1997; see Fodorand Sag 1982).

    Thus, a th- demonstrative may be neutral or vague on deixis, just as ath- definite article may be neutral on definiteness. Nevertheless, from myperspective, a nominal phrase with a th- demonstrative will always containa person feature expressed through th-, and will therefore function as anargument.22

    We saw above that English th- forms like definite articles and demon-stratives are not uniformly interpreted as definite and deictic, respectively.I suggested that th- contributes a (third-)person feature, and yet is neutralwith respect to definiteness and deixis. What of third-person personalpronouns like they? Consider the examples in (27).

    We seem to find the same potential ambiguity with these forms. Aspersonal pronouns (containing th-), the forms display a person feature andfunction as arguments (here, nominative subjects). This provides some

    (26) a. This guy (on the subway last night) gave me his seat.

    (indefinite)

    b. This guy (right here) just gave me a dollar!

    (deictic)

    (27) a. They (over there) are Italian. (deictic)b. They say its gong to rain tonight. (impersonal)

  • 16 Judy B. Bernstein

    2008 The Author Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.xJournal Compilation 2008 Blackwell Publishing Ltd

    support for the uniform treatment of the definite articles, demonstratives,and third-person pronouns.

    In light of my proposal that English th- forms encode third person, aquestion arises about the singular h- pronouns in English (e.g., him, her,and (h)it).23 In particular, do these pronouns have an identifiable personmorpheme? My approach to these pronouns draws on Dchaine andWiltschkos observations and their examples of word formation with h-pronouns [e.g., he-goat, she-society, from 2002, (51)]; these ideas aredeveloped more extensively in Bernstein (2008). I take the contributionof the h- forms in the relevant examples to be nothing more than gender:a he-goat is a male goat and a she-society is a female society (women leadersand/or residents). So the idea I pursue is that rather than person, h-encodes gender, which I associate with N, thinking of languages withrobust gender systems (see Ritter 1993 for the idea that gender is a featureof nouns). Dchaine and Wiltschko also take person to be absent from N,although their treatment of English h- pronouns is somewhat different,corresponding to their P.

    The analysis I develop about English th- brings up an important issueabout the nature of third person and how exactly it differs from first/second person, an issue addressed by many authors. Kayne (2000), forexample, building on Benveniste (1966: 228), argues that third persondoes not form a natural class, unlike first and second person. Harley andRitter (2002) follow Forchheimers (1953) idea that first and secondperson are marked compared with unmarked third person (see also Lyons1999: 316). The idea that th- may encode third person (just as m- and y- mayencode first and second person) is apparently at odds with these otherviews about third person. I will maintain that third person does count asa full-fledged person, although its interpretation and syntactic nature dotend to differ from that of first/second person.24

    In terms of interpretation, first- and second-person pronouns refer tospeaker and hearer, respectively. In this sense, they are deictic and theirreference is specific. The interpretation of third-person pronouns, involvingneither speaker nor hearer, is potentially vaguer and as we saw notnecessarily deictic. Unlike first or second person, the vagueness of thirdperson pronouns allows for participation in bound variable anaphora:25

    In contrast to my treatment of third person in the nominal domain, third-person verbal agreement is often less robust than that of first or secondperson. In many Indo-European languages, for example, the third-personverbal form is a bare form, whereas the first- and second-person formsare identifiable via dedicated verbal morphology (see discussion in Lyons

    (28) a. Everyonei likes theiri (own) motherb. *Everyonei likes myi/youri (own) mother.

  • 2008 The Author Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.xJournal Compilation 2008 Blackwell Publishing Ltd

    Reformulating the Determiner Phrase Analysis 17

    1999: 316). In addition and probably related to the interpretational vague-ness of third person, definite articles and third-person clitic pronouns inmany Indo-European languages agree with nouns in number and gender(and sometimes case), unlike first- and second-person forms (see Kayne2000; Nevins 2007).

    3.3 PERSON WITH PROPER NAMES

    Longobardi (1994) (see also Vergnaud and Zubizarreta 1992) shows that propernames in some languages display an optional or obligatory definite article,argued to be expletive, because it contributes nothing to interpretation.Consider the Italian and Catalan examples in (29) and (30), respectively.26

    Longobardi hypothesizes that the function of the expletive definite articlein these cases is to fill in the D position when the proper name itself doesnot raise to D. Recall that for Longobardi the D position is necessaryfor referential expressions like proper names. According to the proposaldeveloped here, the article supplies a person marker, spelling out thethird-person status of the proper name in languages (like Italian) that alsoallow raising of the proper name to D. The expression of the expletivearticle is not limited to Italian and Catalan, but rather seems to be possiblein a number of the worlds languages.

    If the so-called expletive definite articles in the preceding exampleswith proper names may be reinterpreted as person markers, those exampleswould now resemble those with true personal pronouns, such as the following:

    In these examples, the first- and second-person pronouns identify theperson and in turn the reference of the nominal subjects, just as in theexamples we saw with common nouns in (18) from Section 2.2. There-fore, in all these cases in (29)(31), the insertion of a person-bearingelement identifies the person of the nominal subject: the referent of (31a)is first person (involving the speaker) and that of (31b) is second person(involving the hearer).

    (29) a. Maria (Italian)b. la Maria

    Mary(30) a. Pere (Catalan)

    b. en PerePeter

    (31) a. We Marys have to stick together.b. You Marys have to stick together.

  • 18 Judy B. Bernstein

    2008 The Author Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.xJournal Compilation 2008 Blackwell Publishing Ltd

    The examples are also quite clearly consistent with Longobardis idea(recall discussion in Section 2.1) that proper names, like common nouns,start out in N. The examples in (32), like those seen previously, illustratethat it is the person feature of D (not N) that is relevant for the syntacticrelation that establishes anaphoric dependencies.

    Example (32a) shows that a possessive or reflexive pronoun our(selves) islicit when its antecedent is the pronoun we (in D); example (32b) showsthat a third-person anaphor may not have the proper name Marys (in N)as its antecedent. The role of D in establishing anaphoric dependencies in(32a,b) strongly suggests then that it must also be D (hosting the definitearticle) in (32c) that provides the antecedent for the third-person anaphors.

    Recall Longobardis idea (see Section 2.1) that proper names start outin N and raise to D in order to check a referential feature in D. I haveretained Longobardis idea about N-to-D raising and identified Dsrelevant property to be person rather than reference (as Longobardi has inrecent work, see Longobardi 2008). Personal pronouns spell out this prop-erty just as, I claim, definite articles do (even the so-called expletive ones).What about cases of articleless proper names like Mary that arguably raiseto D overtly in Italian but not in English? In particular, does Mary havea person feature? The straightforward answer is yes, but some of the detailsneed to be further articulated.27 The basic idea proposed by Longobardiis that proper names do have a third-person feature but that their move-ment to D is unrelated to their morphology (unlike pronouns and definitearticles); instead, it is related to their interpretation. For this reason, unlikepronouns, proper names do not move to D overtly in all languages.Longobardi develops a parametric approach to person and distinguishesstrong from weak person languages. In the former (like Italian), propernames will raise to D overtly; in the latter (like English), they will raise covertly.

    In summarizing this section, I observe that the proposal that D is theperson head unifies several interrelated ideas about nominal phrases ingeneral and DPs in particular. First, it offers an explanation for whyelements like definite articles and personal pronouns may occupy the samefunctional head, namely, both sorts of elements encode the same gram-matical information, person, whether we adhere to Postals intuition thatpronouns are determiners or follow Lujns (2001) idea that determinersare modified pronouns.28 Second, it identifies a property characterizingsome nominal phrases but not others, and as such could turn out to bethe criterion distinguishing DPs from NPs and consequently arguments

    (32) a . [We Marys]i have to watch ouri backs/ourselvesi.b. *[We Marys]i have to watch theiri backs/themselvesi.c. [The Marys]i (of the world) have to watch theiri backs/

    themselvesi.

  • 2008 The Author Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.xJournal Compilation 2008 Blackwell Publishing Ltd

    Reformulating the Determiner Phrase Analysis 19

    and vocatives (as individuals) from predicates (as properties). Third, itprovides D an underlying phi feature on which to build or derive inter-pretive properties such as reference and even definiteness. Still to beworked out is exactly how such interpretations are derived from the input.

    3.4 PERSON IN NOMINAL PHRASES AND CLAUSES

    To conclude this section on person, I offer some general remarks aboutcorrespondences between person in the nominal domain and person inthe clausal domain. Unlike the other nominal/clausal correspondencesexamined in this article, the person relationship is one of agreement,typically an argument triggering person agreement on the verb.

    To my knowledge, cross-linguistic syntactic work comparing personacross the two domains is limited, although there have been many pro-posals in the literature analyzing person agreement markers in the verbaldomain as incorporated pronouns, at least for certain languages. Siewier-ska (2004: 1227) extensively discusses and compares anaphoric pronounsand person agreement across a broad array of languages, distinguishing amonglanguages exhibiting pronominal agreement markers, syntactic agreementmarkers, and ambiguous agreement markers (Siewierska 2004: 126),observing that the particular characterization of person marking is oftendetermined by a linguists theoretical framework (Siewierska 2004: 125).For present purposes, it is interesting to observe that languages varysignificantly in terms of if and where person is expressed. In the Romancelanguages, for example, person is typically expressed robustly both in theclausal and nominal domains, except that in several Romance languagesthe subject may remain unexpressed. In languages like standard English,there is arguably no longer person marking in the clausal domain (if verbal-s is taken to be a number marker, see Kayne 1989), and yet robust personmarking in the nominal domain.29 In other languages (perhaps Chinese),there is no obvious evidence of person in either the nominal or clausaldomain.30 In addition to being a worthwhile focus of study in its ownright for typological reasons, a comparison of person across the twodomains may prove useful for discovering how grammars of naturallanguage are organized and how they change over time.31

    For my own particular proposal about person, languages (like Japaneseor Chinese) that lack definite articles and also personal pronouns (includingin the form of reflexive anaphors) raise important issues. In particular, ifa language lacks person marking in the nominal domain does it necessarilylack the functional category DP? If so, what category do argumentsbelong to in such a language? Recall that in this article I have pursuedthe idea that the presence or absence of person (encoded in D) underlieswhether a nominal phrase is an argument or not. Another set of questionsarises with languages that lack person marking in the nominal domain andyet display person marking in the clausal domain: what is the trigger for

  • 20 Judy B. Bernstein

    2008 The Author Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.xJournal Compilation 2008 Blackwell Publishing Ltd

    the verbal agreement? Some of these issues are touched on in Longobardi(2008), but they are far from settled. As Longobardi observes, if it can beshown that languages like Japanese allow NPs in argument function, thiswould provide support for Chierchias (1998) approach to argumentscross-linguistically (i.e., they are or can be NPs in some languages andmust be DPs in others).32 If these languages lack DP, how is person thatis relevant to interpretation computed? For those languages like Japanesearguably lacking grammatically encoded person, correlating with absenceof DP, Longobardi (2008) speculates that person is interpreted pragmatically.This parametrization of phi features grammatically encoded in somelanguages, pragmatically interpreted in others should then generalize beyondperson to other phi features, such as number. Further cross-linguistic inves-tigation will determine whether such predictions are borne out.

    4. Conclusion

    The DP analysis has been influential and fueled a lot of research on theinternal structure of nominal phrases. In this short article, I have reviewedsome of the data illustrating cross-linguistic correspondences betweennominal phrases and clauses, data pointing to parallel internal structuresin the nominal and clausal domains. I focused on one particular propertyarguably shared by clausal CP and nominal DP, namely, their status asthematic arguments (matching up with IP and NP predicates, respectively).Directing my attention to DP arguments, I asked if some identifiableproperty of the functional head D is responsible for the argument statusof DP. Following other authors, I observed that elements typically associatedwith D do not have or contribute a consistent interpretation. I concludedthat Ds role is probably not involved directly in interpretation, but mightcertainly be involved derivationally. This general conclusion is consistentwith recent proposals that some interpretive features may not be encodedin the syntax at all, or instead may be hosted in the left periphery of theclause or nominal DP.

    The reformulation I proposed for DP takes D to be a nominal func-tional head encoding person (1st, 2nd, or 3rd), information upon whichreference and anaphora can be established. Examining expressions withpersonal pronouns, determiners (definite articles, demonstratives), andproper names, I conjectured that what the relevant D elements share, orcontribute, is person, information that helps us determine whether theindividual denoted by a nominal phrase is one of the participants in theconversation (first, second person) or not (third person).

    Short Biography

    Judy B. Bernsteins research focuses on morpho-syntactic phenomena, inparticular on micro-parametric variation in Romance languages and more

  • 2008 The Author Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.xJournal Compilation 2008 Blackwell Publishing Ltd

    Reformulating the Determiner Phrase Analysis 21

    recently in varieties of English. She has worked extensively on nominalstructure and has published articles in Language, Lingua, Linguistic Inquiryand Probus. She is currently involved in a collaborative project on themorpho-syntax of Appalachian English. She has been awarded an AmericanCouncil of Learned Societies fellowship, a Fulbright grant, and a NationalScience Foundation grant. Before coming to William Paterson University,where she is Associate Professor of Linguistics, she taught at SyracuseUniversity and before that she was a postdoctoral fellow at the Universityof Southern Maine. She holds a BA in Spanish and Psychology from theUniversity at Albany, an MA in Teaching English as a Second Languagefrom Hunter College, and a PhD in Linguistics from The GraduateCenter of the City University of New York.

    Notes

    * Correspondence address: Judy B. Bernstein, William Paterson University of New Jersey, 300Pompton Rd., Wayne, NJ 07470, USA. Email: [email protected] As far as I know, Giusti is the only author to associate Case with D directly. Recent work byWatanabe (2006) posits a DP-internal CaseP for Japanese, an idea that has appeared over the yearsin work on several languages. Similarly for deixis, Roca (1996) has posited a DP-internalDemonstrativeP, unlike other authors (e.g., Giusti 1993) who take demonstrative elements tobe adjectival (Dryer 1992) and interpreted through D(P), though not corresponding to Dunderlyingly. Klinge (2008) associates D, in particular Germanic th-/- elements, with deixis.According to Ihsane and Puskas (2001), whose analysis builds on the semantic work of En(1991), specificity is encoded in the left periphery of the DP ( la Rizzi 1997 on the leftperiphery of the clause).2 Abneys proposal actually aligned DP with IP, but that would no longer be the relevantanalogy.3 Nowadays, TenseP (TP) would replace IP as the functional projection relevant for tense.4 Longobardi (2008) points out that if a language does not encode phi features syntactically, itmay in fact admit NP arguments, as allowed for within Chierchias system (see Section 3.4).5 The following Italian example is even more telling and supports assuming NP for vocatives,since Italian would bar the same bare plural in argument function: Ragazzi! (Kids!).6 French is unusual in allowing definite articles in vocatives: les enfants! (children!). This factis unexpected under the NP (predicate) approach to vocatives, but also under my proposedapproach to vocatives. Dchaine and Wiltschkos (2002) approach to the definite article inFrench (see discussion in Section 3.2) would offer an account, because for them the definitearticle does not correspond to D in French.7 In argument function, both sorts of expressions appear as subjects of lexical verbs: Patricia metan intelligent woman; You kids talk too much.8 Longobardi (2008) illustrates the notion of properties, typically adjectives or nouns used aspredicates: The food is hot, I am John, This is gold.9 Longobardi shows how the nouns from footnote 8 can also name individuals: I met John, Goldis precious.10 In this regard, the question of English bare plural arguments is crucial: Girls enrolled in sciencecourses. For Chierchia, the subject (girls) would be an NP argument; for Longobardi, the subjectwould involve a null D.11 Lyons (1999) extensively reviews the concept of definiteness from various theoreticalperspectives, analyses ranging across notions of familiarity, identifiability, inclusiveness, uniqueness,and presupposition. As Lyons points out (1999: 274), no one approach is able to account forthe full range of facts; he takes identifiability (i.e., a DPs referent is identifiable to the speakerand hearer) to be the pragmatic property that definiteness encodes (Lyons 1999: 278).

  • 22 Judy B. Bernstein

    2008 The Author Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.xJournal Compilation 2008 Blackwell Publishing Ltd

    12 Francisco Ordez (personal communication) points out that he would only use this whentalking among family members.13 For a DP argument like the lion, Longobardi, building on work by Carlson (1977), appealsto a distinction between kind referring for the generic interpretation and individual referringfor the definite one.14 She distinguishes coindexing from coreference, which she takes to be a semantic/pragmaticnotion, following Fiengo and May (1994; see Pereltsvaig 2007: 678).15 Pereltsvaig (2007: 70) discusses sign languages such as American Sign Language (as describedin MacLaughlin 1997), which assign reference spatially. Unlike MacLaughlin (1997: 59), whotakes these spatial loci to involve person, Pereltsvaig takes the relatively large number of spatialloci in sign languages to support her proposal of referential indices, which would similarlyinvolve a large number.16 A reviewer asks why this construction is restricted to plural. The idea developed in Bernstein(2008) is that these pronoun-plus-noun constructions (for those languages admitting them)require (i) agreement in number between pronoun and noun, and (ii) the expression of personin D. This is satisfied by the examples with first-plural we/us, second plural you, and third plural(vernacular) them: we/us linguists, you linguists, them linguists (see Section 3.2 for the proposalthat th- encodes third person). Although *I linguist displays person through I, it lacks numberagreement between the singular noun linguist and the pronoun I, which Kayne (1989) arguedis unmarked for number. Similarly, Kayne observed that you is grammatically plural (you are vs.*you is), which explains the number mismatch in *you linguist. As for the impossibility ofh- pronouns in this construction (*he/him linguist), see Section 3.2 for the idea that h- does notencode person, but rather gender.17 I expect this to be the case in languages admitting this type of construction. Italian, forexample, displays the same pattern: Noi italiani ammiriamo il nostro presidente (We Italians admireour president.).18 I depart from Lyons (1999: 313) approach, which conflates person and definiteness and startsfrom the observation that personal pronouns and so-called personal determiners (e.g., we lin-guists) come from the same source (Lyons 1999: 310). I also depart from the specifics, but notthe spirit, of Platzack (2004), who takes person to correspond to a functional projection internalto DP.19 Roehrs (2005), on the basis of new data and arguments, concludes, as Postal did, that therelationship between the determiner pronoun and noun is one of complementation and notapposition.20 I suspect that You Patricia! is not pronounceable because of non-agreement (in number)between the pronoun, which is grammatically plural, and the singular proper name. Given theright context, the expression becomes possible when the proper name is plural: You Patricias!21 There have been proposals equating a th- morpheme with definiteness (see Bernstein et al.1999; Dchaine and Wiltschko 2002) and deixis (Klinge 2008).22 I abstract away from the idea that demonstratives, like pronouns, may not correspond to Dunderlyingly (see Giusti 1993; Bernstein 1997). In this way, they would resemble the simplepronouns discussed above.23 Historically, the neutral pronoun it patterned with the other h- forms: her, him, and hit. Theform hit is still heard in conservative varieties of English such as Appalachian English.24 Similarly, Nevins (2007) study of the morphological properties of the person-case constraintled him to conclude that third person, like first and second, is always featurally represented (p.311).25 I opt for their over his/her in this example, because it is more natural sounding, particularlyin colloquial language. This is consistent with the claim that th- (and not h-) encodes thirdperson.26 Catalan reserves this form of the definite article for proper names. With common nouns, themasc. sg. form of the article would be el: el gos the dog (see Longobardi 1994: 656 fordiscussion).27 Recall that Longobardi (2008) distinguishes individuals from properties, only the formerinvolving person specification.28 Historically, many present-day determiners developed from pronouns or demonstratives.

  • 2008 The Author Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.xJournal Compilation 2008 Blackwell Publishing Ltd

    Reformulating the Determiner Phrase Analysis 23

    29 Varieties of English may differ as to which phi feature is encoded in the verbal system. Incurrent work, Raffaella Zanuttini and I contrast what we take to be singular number expressedthrough verbal -s in standard English (following Kayne 1989) to what we propose to be thirdperson expressed with the identical verbal form in Appalachian English.30 I do not mention the Slavic articleless languages here, because they have personal pronouns,which can be used to establish the DP status of nominal phrases. Pereltsvaig (2007) appealsadditionally to differences between post-copular instrumental and nominative nominal phrasesin Russian to argue for the DP status of the latter but not the former. Interestingly, and asexpected given her analysis, post-copular pronouns are also typically nominative (Pereltsvaig2007: 27).31 Siewierska (2004: 261ff.) discusses the diachronic grammaticalization of person markers:independent person marker > weak form > clitic > agglutinative affix > fusional form > [p. 262, (31)].32 Watanabe (2006), in fact, argues that DP is projected in Japanese.

    Works Cited

    Abney, Stephen P. 1987. The English noun phrase in its sentential aspect, dissertation. Cam-bridge, MA: MIT,.

    Alexiadou, Artemis. 2001. Functional structure in nominals: nominalization and ergativity.Amsterdam, The Netherlands: John Benjamins.

    Alexiadou, Artemis, Liliane Haegeman, and Melita Stavrou. 2007. Noun phrase in the generativeperspective. Berlin, Germany: Mouton de Gruyter.

    Baker, Mark. 2003. Lexical categories: verbs, nouns, and adjectives. Cambridge, UK:Cambridge University Press.

    . 2008. The syntax of agreement and concord. Cambridge, UK: Cambridge University Press.Benveniste, Emile. 1966. Problmes de linguistique gnrale. Paris, France: Gallimard.Bernstein, Judy B. 1997. Demonstratives and reinforcers in Romance and Germanic languages.

    Lingua 102.87113.. 2001. The DP hypothesis: identifying clausal properties in the nominal domain.

    Handbook of contemporary syntactic theory, ed. by Mark Baltin and Chris Collins, 53661.Malden, MA: Blackwell Publishers.

    . 2008. English th- forms. Essays on nominal determination, ed. by Alex Klinge andHenrik Mller, 21332. Amsterdam, The Netherlands: John Benjamins.

    Bernstein, Judy B., Wayne Cowart, and Dana McDaniel. 1999. Bare singular effects in genitiveconstructions. Linguistic Inquiry 30.493502.

    Borer, Hagit. 1984. Parametric syntax. Dordrecht, The Netherlands: Foris.Campbell, Richard. 1996. Specificiy operators in SpecDP. Studia Linguistica 50.16188.Cardinaletti, Anna. 1994. On the internal structure of pronominal DPs. The Linguistic Review

    11.195219.Carlson, Greg. 1977. Reference to kinds in English, dissertation. Amherst, MA: University of

    Massachusetts.Chierchia, Gennaro. 1984. Topics in the syntax and semantics of infinitives and gerunds,

    dissertation. Amherst, MA: University of Massachusetts.. 1998. Reference to kinds across languages. Natural Language Semantics 6.339405.Chomsky, Noam. 1970. Remarks on nominalization. Readings in English trnsformtionl grammar,

    ed. by R. Jacobs and P. Rosenbaum, 184221. Waltham, MA: Ginn.. 1986. Barriers. Cambridge, MA: MIT Press.. 1993. A minimalist program for linguistic theory. The view from Building 20: essays in

    linguistics in honor of Sylvain Bromberger, ed. by Kenneth Hale and Samuel Jay Keyser, 152.Cambridge, MA: MIT Press.

    . 2000. Minimalist inquiries: the framework. Step by step: essays on minimalist syntax inhonor of Howard Lasnik, ed. by Roger Martin, David Michaels and Juan Uriagereka, 89155.Cambridge, MA/London, UK: MIT Press.

    . 2001. Derivation by phase. Ken Hale: a life in language, ed. by Michael Kenstowicz, 152.Cambridge, MA/London, UK: MIT Press.

  • 24 Judy B. Bernstein

    2008 The Author Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.xJournal Compilation 2008 Blackwell Publishing Ltd

    Cinque, Guglielmo. 1980. On extraction from NP in Italian. Journal of Italian Linguistics5.4799.

    Crisma, Paola. 1997. Larticolo nella prosa inglese antica e la teoria degli articoli nulli. Padova,Italy: Universit degli Studi di Padova dissertation.

    Dchaine, Rose-Marie, and Martina Wiltschko. 2002. Decomposing pronouns. LinguisticInquiry 33.40942.

    Dryer, Matthew. 1992. The Greenbergian word order correlations. Language 68.81138.En, Murvet. 1991. The semantics of specificity. Linguistic Inquiry 22.125.Fiengo, Robert, and Robert May. 1994. Indices and identity. Cambridge, MA: MIT Press.Fodor, Janet Dean, and Ivan A. Sag. 1982. Referential and quantificational indefinites. Linguistics

    and Philosophy 5.35598.Forchheimer, Paul. 1953. The category of person in language. Berlin, Germany: Walter de

    Gruyter.Giorgi, Alessendra, and Giuseppe Longobardi. 1991. The syntax of Noun Phrases: configuration,

    parameters and empty categories. Cambridge, UK: Cambridge University Press.Giusti, Giuliana. 1993. La sintassi dei determinanti. Padua, Italy: Unipress.. 1995. A unified structural representation of (abstract) Case and articles. Evidence from

    Germanic. Studies in comparative Germnic syntax, ed. by Hubert Haider, Susan Olsen andSten Vikner, 7793. Dordrecht, The Netherlands: Kluwer.

    Grimshaw, Jane. 1990. Argument structure. Cambridge, MA: MIT Press.. 1991. Extended projection. Unpublished manuscript. Brandeis University.Harley, Heidi, and Elizabeth Ritter. 2002. Person and number in pronouns: a feature-geometric

    analysis. Language 78.482526.Horrocks, Geoffrey, and Melita Stavrou. 1987. Bounding theory and Greek syntax: evidence

    for wh-movement in NP. Journal of Linguistics 23.79108.Ihsane, Tabea, and Genoveva Puskas. 2001. Specific is not definite. Generative Grammar in

    Geneva 2.3954.Kayne, Richard S. 1989. Notes on English agreement. CIEFL Bulletin. Hyderabad 1.4167

    (Reprinted in Kayne 2000, 187205).. 2000. Parameters and universals (Chapter 8: Person morphemes and reflexives). New

    York, NY: Oxford University Press.Klinge, Alex. 2008. Stating the case for -root and hw-root determiners. Essays on nominal

    determination, ed. by Alex Klinge and Henrik Mller, 23363. Amsterdam, The Nether-lands: John Benjamins.

    Kornfilt, Jaklin. 1984. Case marking, agreement, and empty categories in Turkish, dissertation.Cambridge, MA: Harvard University.

    Longobardi, Giuseppe. 1994. Reference and proper names: a theory of N-movement in syntaxand logical form. Linguistic Inquiry 25.60965.

    . 1996. The syntax of N-raising: a minimalist theory, OTS Working Papers. ResearchInstitute for Language and Speech. Utrecht, The Netherlands: University of Utrecht.

    . 2008. Reference to individuals, person, and the variety of mapping parameters. Essayson nominal determination, ed. by Alex Klinge and Henrik Mller, 189211. Amsterdam,The Netherlands: John Benjamins.

    Lujn, Marta. 2001. Determiners as modified pronouns. Proceedings of the Chicago LinguisticSociety 36, Volume 1: the main session, ed. by John Boyle, Jung-Hyuck Lee and ArikaOkrent. Chicago, IL: The University of Chicago Press.

    Lyons, Christopher. 1999. Definiteness. Cambridge, UK: Cambridge University Press.MacLaughlin, Dawn. 1997. The structure of determiner phrases: evidence from American Sign

    Language, dissertation. Boston, MA: Boston University.Mandelbaum, Deborah. 1994. Syntactic conditions on saturation, dissertation. New York, NY:

    CUNY.Marantz, Alec. 1997. No escape from syntax: dont try morphological analysis in the privacy

    of your own lexicon. UPenn Working Papers in Linguistics 4(2).20125.Nevins, Andrew. 2007. The representation of third person and its consequences for person-case

    effects. Natural Language and Linguistic Theory 25.273313.Pereltsvaig, Asya. 2007. Copular sentences in Russian. Dordrecht, The Netherlands: Springer.

  • 2008 The Author Language and Linguistics Compass 2 (2008): 10.1111/j.1749-818x.2008.00091.xJournal Compilation 2008 Blackwell Publishing Ltd

    Reformulating the Determiner Phrase Analysis 25

    Picallo, M. Carme. 1991. Nominals and nominalization in Catalan. Probus 3.279316.Platzack, Christer. 2004. Agreement and the person phrase hypothesis. Working papers in

    Scandinavian syntax 73.83112. Lund, Sweden: Department of Scandinavian Languages,Lund University.

    Postal, Paul M. 1966. On so-called pronouns in English. Monograph series on languages andlinguistics 19, ed. by Francis P. Dineen, 177206. Washington, DC: Georgetown UniversityPress.

    Radford, Andrew. 1997. Syntactic theory and the structure of English: a minimalist approach.Cambridge, UK: Cambridge University Press.

    Ritter, Elizabeth. 1993. Wheres gender? Linguistic Inquiry 24.795803.Rizzi, Luigi. 1997. The fine structure of the left periphery. Elements of grammar, ed. by

    Liliane Haegeman, 281337. Dordrecht, The Netherlands: Kluwer Academic Publishers.Roca, Francesc. 1996. La determinacin y la modificacin nominal en espaol. Barcelona,

    Spain: Universitat Autnoma de Barcelona dissertation.Roehrs, Dorian. 2005. Strong pronouns are determiners after all. The function of function

    words and functional categories, ed. by Marcel den Dikken and Christina Tortora, 25185.Amsterdam, The Netherlands: John Benjamins.

    Siewierska, Anna. 2004. Person. Cambridge, UK: Cambridge University Press.Stowell, Timothy. 1989. Subjects, specifiers, and X-bar Theory. Alternative conceptions of

    phrase structure, ed. by Mark Baltin and Tony Kroch, 23262. Chicago, IL: University ofChicago Press.

    Szabolcsi, Anna. 1983. The possessor that ran away from home. The Linguistic Review 3.89102.

    . 1987. Functional categories in the noun phrase. Approaches to Hungarian 2, Theoriesand Analyses, ed. by Istvan Kenesei, 16789. Szeged, Hungary: JATE.

    . 1992. Subordination: articles and complementizers. Approaches to Hungarian 4, ed. byI. Kenesei and Cs. Plh, 12337. Szeged, Hungary: JATE.

    Valois, Daniel. 1991. The internal syntax of DP, dissertation. Los Angeles, CA: UCLA.Vangsnes, ystein Alexander. 1999. The identification of functional architecture, dissertation.

    Bergen, Norway: University of Bergen.Vergnaud, Jean-Roger, and Maria Luisa Zubizarreta. 1992. The definite determiner and the

    inalienable constructions in French and in English. Linguistic Inquiry 23.595652.Watanabe, Akira. 2006. Functional projections of nominals in Japanese: syntax of classifiers.

    Natural Language and Linguistic Theory 24.241306.Zamparelli, Roberto. 1995. Layers in the determiner phrase, dissertation. Rochester, NY:

    University of Rochester.