Copula Cycle

Embed Size (px)

Citation preview

  • 8/13/2019 Copula Cycle

    1/43

    1

    The Copula Cycle

    Terje Lohndal, University of Maryland

    Abstract

    It is well-known that copulas often emerged from demonstratives and pronouns historically.

    The present paper argues that copulas change cyclically. This claim is substantiated through a

    rich number of examples from various languages. Besides the change from demonstratives

    and pronouns to copulas, it is shown that copulas further may develop into affixes.

    Interestingly, there are also languages like Hebrew where the copula first disappears and then

    redevelops at a later stage. This paper describes the stages of the copula cycle, and argues that

    there also are changes that do not fit directly into the major stages. The cycle is understood

    through a formal theory of grammaticalization, where there are cognitive principles that help

    the child to acquire a language.

    1. Introduction*

    It is a well-known fact that demonstratives may develop [] into personal pronouns, which

    themselves may give rise to copulas. (Heine and Kuteva 2002: 109). In the typological

    literature, this process is known as copularization: the grammaticalization process which

    turns full verbs or other non-copular elements into copulas (Hengeveld 1992: 237-256,

    Stassen 1997: 94-99; though see Pustet 2003 for a slightly different terminology). The aim of

    this paper is to argue that there is a systematic cycle where copulas emerge from

    demonstratives and pronouns or from verbs and might then develop further into auxiliaries

    and grammatical markers like affixes. Demonstratives and pronouns may even reappear from

    copulas. In addition to providing a typology of these patterns of change, I also aim at giving atheoretical analysis of these data within the Minimalist Program. Specifically I will approach

    the data from the perspective of a formalist understanding of grammaticalization as in Roberts

    and Roussou (2003) and van Gelderen (2004). This framework, I claim, provides a useful way

    both to classify the changes and to understand the changes in question.

    *Parts of this paper have been presented at the Linguistic Cycles workshop at Arizona State University in April2008. I am grateful to the audience for valuable comments. Thanks also to Werner Abraham, Brian Dillon, JanTerje Faarlund, Elly van Gelderen, David Ingram and David Lightfoot for their useful remarks.

  • 8/13/2019 Copula Cycle

    2/43

    2

    The paper is organized as follows. Section 2 discusses an approach to the syntax of

    copulas. Section 3 discusses grammaticalization and economy, adopting the framework in van

    Gelderen (2004a, b et seq.). The copula cycle is introduced in section 4, and step by step data

    are introduced that motivate the copula cycle. Section 5 concludes the paper.

    2. The syntax of copulas

    How to analyze copulas is a problematic and somewhat controversial question. I will not

    provide an overview here, but simply present and briefly motivate the framework I am

    adopting.

    A common assumption about copulas is articulated by Baker (2003: 40):

    This range of data implies that the copula in English is not involved primarily in the

    dynamics of theta-role assignment, but rather appears when the lexical head of the

    clause cannot bear finite tense and agreement morphology.

    This has been the traditional approach. As has been argued elsewhere, there are many reasons

    to think that this view is wrong (Rothstein 1999, 2001, Lohndal 2006, Lohndal, farli and

    Nygrd 2008). However, the issue is not directly important for the present paper, and what I

    will say in the remainder is largely tangential to this issue. What is important, however, is the

    phrase structure we assume for copulas. In the next section, I will outline an analysis where

    copulas are heads of a predication phrase.

    In a seminal paper, Bowers (1993) proposes a predication phrase, PrP. Pr is a

    functional category where the external argument (the subject) sits in the specifier of Pr. The

    predicate is the complement, and can consist of either a VP, NP, PP or an AP. This can be

    illustrated by way of the following tree structure.

    (1) PrP

    ei

    SUBJECT Pr

    ei

    Pr PREDICATE/XP

  • 8/13/2019 Copula Cycle

    3/43

    3

    This structure accommodates small clauses as well as main clauses. The structure for a small

    clause like (2) is provided in (3).

    (2) Lisa considers Mary crazy.

    (3) [IPLisa considers [PrPMary [PrPr [APcrazy]]]]

    Above the PrP in the clause is the inflectional phrase, IP. To illustrate a more complete

    structure, (4) represents a typical sentence (I have omitted the higher layers).1

    (4) PrP

    ei

    SUBJECT Pr

    ei

    Pr VP

    ei

    V

    ei

    V

    Eide and farli (1999), based on Eide (1996), have argued, pace Bowers (1993), Baker

    (2003) and Mikkelsen (2005), that copulas can lexicalize the head of PrP (see also Adger and

    Ramchand 2003: 336). This has further been corroborated by Lohndal, Nygrd and farli

    (2008), although their overall conclusion is slightly different. It is not possible to repeat all the

    arguments in favor of this conclusion, but let us just look at one important one from Eide and

    farli (1999).

    Eide and farli (1999) note that there is an interesting symmetry between the particle

    somin Norwegian and the copula. First, they argue that the particlesomcan lexicalize the

    predication operator. A few examples supporting this are provided in (5)-(7).

    1As Bowers (1993: 599-600) argues, a main verb moves from V to Pr. Later on many people have used vP as anotation for more or less the same thing as PrP (Chomsky 1995, Kratzer 1996), and some people use PredPinstead of PrP (Adger and Ramchand 2003, Baker 2003). Some have generalized this approach even further such

    that a variety of heads can lexicalize what has been called a relator phrase (den Dikken 2006). There aredifferences between these two approaches, but I will set them aside in this paper as they are not directly relevantfor what follows.

  • 8/13/2019 Copula Cycle

    4/43

    4

    (5) Vi fant [Marit [(*som) naken/*(som) nervevrak]].

    we found Marit as naked / as nervous wreck

    We found Marit nude/as a nervous wreck.

    (6) Vi s [Jon [(*som) rasende/*(som) spkelse]].

    we saw John as furious / as ghost

    We say John being furious/as a ghost.

    (7) Vi returnerte [pakken [(*som) upnet /*(som) flypost]].

    we returned parcel.DEF as unopened / as air mail (Eide and farli 1999: 160)

    We returned the parcel unopened/as air mail.

    An adjectival predicate does not permit the occurrence of the particlesom, whereas a nominal

    small clause predicate forces the presence ofsom. The structures in (5)-(7) all have in

    common that you can always paraphrase them as a matrix sentence, in which case you

    evidently will have to replacesomwith the copula. This is illustrated in (8)-(9).

    (8) Vi fant [Marit [som nervevrak]] ! Vi fant Marit og [Marit [var nervevrak]].

    we found Marit as a nervous wreck !we found Marit and Marit was a nervous

    wreck

    (9) Vi ser [dette [som faktum]] ![Dette [er et faktum]]

    we see this as a fact ! this is a fact (Eide and farli 1999: 165)

    As Eide and farli (1999) note, the bracketed parts of the two structures in each example are

    very similar both semantically and syntactically, and the most striking difference seems to be

    the head of the PrP, viz. the copula orsom. The difference in lexicalization is related to the

    selecting matrix element. Eide and farli hypothesize that the selecting matrix element is averb not selecting for a verbal category whensomoccurs and a tense element when the copula

    occurs. This predicts thatsomand the copula are in complementary distribution, except when

    the head optionally selects a verbal category or a non-verbal category. The latter is precisely

    the case with Norwegian perception verbs. Eide and farli point to (10) and (11) as evidence

    that in such a case, there is no complementary distribution, which is a strong argument in

    favor of treating bothsomand the copula as lexicalizations of the predication operator.

  • 8/13/2019 Copula Cycle

    5/43

    5

    (10) Jeg s [naboen [som spkelse]].

    I saw neighbor.DEFsoma ghost

    I saw the neighbor as a ghost.

    (11) Jeg s [naboen [vre spkelse]].

    I saw neighbor.DEFbe a ghost (Eide and farli 1999: 167)

    I saw the neighbor be a ghost.

    In sum, it seems to be valid to assume the copula to be the head of PrP. This will be important

    for what follows, thus I have taken some space to argue in favor of this position. In the next

    section, I will turn to the specific diachronic framework that I will be using throughout the

    paper.

    3. Economy and grammaticalization

    In recent years, grammaticalization has become an important research topic within generative

    grammar, cf. Longobardi (2001), Roberts and Roussou (2003) and van Gelderen (2004a, b,

    2007, 2008a, b). Within the functionalist paradigm, grammaticalization has been a crucial

    topic for decades, see e.g. Lehmann (1985), Heine, Claudi and Hnnemeyer (1991), Hopper

    and Traugott (1993) and Heine and Kuteva (2002, 2007). The present paper will mainly be

    using the specific proposal laid out by Elly van Gelderen. Although several aspects between

    her theory and Roberts and Roussous are similar, there are also important differences, which

    I cannot deal with here for reasons of space. However, van Gelderen has in recent work

    integrated grammaticalization and the linguistic cycle, and since the topic of this paper bears

    on both issues, I think her framework is especially well-suited for analyzing the diachronic

    source and development of copulas. In the present subsection, I will present the main aspects

    of van Gelderens theory. First I will say something about economy and its role in our internalgrammar, and then I will discuss van Gelderens specific implementation of

    grammaticalization as feature economy.

    3.1. Economy, grammaticalization and features

    Economy has been a pervasive aspect within the history of generative grammar. Here I am

    primarily interested in the economy relating to grammatical derivations, not in the kind of

    economy related to Ockhams razor. With the introduction of the minimalist program inlinguistic theory in the late 1980s (see Chomsky 1995 for a collection of the most central

  • 8/13/2019 Copula Cycle

    6/43

    6

    papers), economy became one of the core principles of the program. One crucial aspect of

    economy is that derivations have to be as cheap as possible, that is, the grammar always

    chooses the least costly derivation. Another aspect is that childrens acquisition of language is

    arguably driven by principles of economy (van Gelderen 2004a, b, 2007a, b, 2008a b,

    Faarlund 2008). Van Gelderen (2004a, b) proposes two specific principles of

    grammaticalization, namely (12) and (13).

    (12) Head Preference Principle (HPP)

    Be a head, rather than a phrase

    (13) Late Merge Principle (LMP)

    Merge as late as possible

    These principles are argued to guide the child during acquisition, which also means that they

    rely on the input. That is, the principles can only work when the input is ambiguous such that

    there is more than one way to analyze the input data.2This relies on the assumption that it is

    more economical to be a head than a phrase as heads are less complex, and the fact that late

    Merge means that movement is avoided. Movement, even if seen as another species of Merge

    (internal Merge, cf. Chomsky 2004), is more costly because it involves a process of copying

    and remerging a lexical item during the derivation. External Merge does not involve copying,

    which makes it cheaper (van Gelderen 2008b). Concretely, HPP means that a speaker will

    prefer to build structures such as (14) rather than (15). I have followed van Gelderen in

    labeling the projection FP and pro stands for a pronoun.

    (14) FP (15) FP

    ru ru

    pro pro Fru

    F .

    Importantly, the HPP is a very specific and non-variant principle, as opposed to e.g. Hawkins

    (2004) efficiency principle Minimize Forms or Stay within Optimality Theory (van

    Gelderen 2008b). The LMP implies that Internal Merge (Move) is less economical than

    2Obviously, this claim is in need of further substantiation which goes beyond the scope of this paper.

  • 8/13/2019 Copula Cycle

    7/43

    7

    External Merge (pace Chomsky 2004), but I will refrain from discussing this further at this

    point. As van Gelderen convincingly has shown, there are real empirical LMP effects, which

    would be problematic to account for if Merge and Move were equally economical.

    The second principle of economy, the LMP, works most clearly in the case of heads.

    Some common examples of LMP at work are listed in (16) (van Gelderen 2008b).

    (16) Preposition to complementizer (e.g.for)

    (17) modals: v> Asp > T

    (18) preposition to aspect (e.g. on)

    (19) vP adverbials to TP/CP adverbials

    (20) Negative objects to negative markers

    One very common change that could be explained by reference to this principle is the well

    known change in lexical verbs becoming modals in English (e.g. Traugott 1972, Lightfoot

    1974, 1979, 1999, 2006, Plank 1984, Roberts 1985, 1993, Kroch 1989, Warner 1993,

    Abraham 2002). In Middle English verbs like can,could,may,might,will,would,shallwere

    full verbs. Lightfoot (2006: 90) summarizes the main differences by way of the following data

    in (21)-(30).

    (21) *He has could understand chapter 4.

    (22) He has understood chapter 4.

    (23) *Canning understand chapter 4,

    (24) Understanding chapter 4,

    (25) *He wanted to can understand.

    (26) He wanted to understand.

    (27) *He will can understand.(28) He will try to understand.

    (29) *He can music.

    (30) He understands music.

    (21)-(30) show that modal auxiliaries differ from ordinary verbs in their distribution. A modal

    does not occur with a perfective (21) or past participle (23), it does not occur in the infinitival

    complement of another verb (25), nor as the complement of another modal (27), and no modalmay occur with a direct object (29). However, when modals were main verbs, these structures

  • 8/13/2019 Copula Cycle

    8/43

    8

    were generally possible. Under the LMP, this change is straightforward.3(31) would be

    preferred compared to (32) because the modal auxiliary is based generated higher in the

    former than in the latter.

    (31) TP (32) TP

    ru ru

    T vP T vP

    might ru ru

    v v

    ru ru

    v v might

    Generally the LMP accounts for changes where a lexical head becomes a functional head, or

    where a functional head becomes a higher functional head, both frequently described in the

    literature on grammaticalization (see e.g. Heine & Kuteva 2002). However, the LMP can also

    account for changes in the position of lexical phrases whereby lexical phrases become base

    generated in the functional domain. Consider the examples in (33)-(34) (van Gelderen 2007a).

    (33) Those who offend actually, are most grievously punished.

    (34) Actually, it is kind of an interesting problem.

    When actuallywas first introduced into the English language from French, it was an

    adjective. In the 15th century it is used as a vP adverb (33). It may have been topicalized

    frequently, leading to a change to a CP adverb (34). This change also shows that the LMP

    works for both phrases and heads.

    Recently, van Gelderen (2007b, 2008b) has reanalyzed the HPP and LMP, collapsing

    them into one overall and more general principle, namely (35).

    3

    Notice, though, that the LMP does not say anything about whenthis change occurred. Obviously, the LMP is aprinciple that interacts with the external data, and unless the external data is such that the principle can kick in, itwont.

  • 8/13/2019 Copula Cycle

    9/43

    9

    (35) Principle of Feature Economy

    Minimize the semantic/interpretable features in the derivation

    Adjunct Specifier Head > affix

    Semantic > [iF] > [uF] > --

    This is among others based on Radfords (2000) argument that interpretable features are

    acquired before uninterpretable features, and the latter ones trigger the grammatical system.

    Assuming that only heads are Probes (Chomsky 2005), (35) means that whenever the HPP is

    at work, we get a (new) Probe. Interestingly, (35) also provides us with a formal definition of

    what grammaticalization really is. It is a change whereby something becomes more

    economical, i.e. where semantic features are reduced (on the assumption that uninterpretable

    features are more economical than interpretable ones). This change in turn has syntactic

    consequences, and the remained of the paper will be dealing with syntactic consequences

    relating to copulas diachronically.

    4. The copula cycle and features

    In this section I will present various stages of what I take to be a copula cycle. First I will

    present the cycle based on general theoretical considerations within the framework laid out in

    section 3. This will function as a guiding heuristic when working our way through the

    examples in the following sections. In section 4.2 I will describe changes where a specifier

    develops into a head, viz. mainly demonstratives or pronouns that become copulas. Section

    4.3 deals with the stage where a full verb has developed to a copula, that is, a head-to-head

    change. This will be shown to be a prime example that the major cycle also has smaller sub-

    stages. Section 4.4 discusses changes where the copula develops into a grammatical marker,

    viz. a suffix. In section 4.5 I sum up the data and argue that the cycle proposed in 4.1 accountsstraightforwardly for the change.

    4.1. The copula cycle

    In the previous section I adopted a principle of feature economy that favors uninterpretable

    features over interpretable features. This predicts that functional elements will start from

    lexical elements, a prediction borne out by a wealth of attested examples in the literature. This

    is also true for copulas, and Pustet (2003: 54) says that the main source for copulas seems tobe verbs and pronouns. Stassen (1997: 92-99) argues that there are three major diachronic

  • 8/13/2019 Copula Cycle

    10/43

    10

    sources for (verbal) copulas. The first source is dynamic verbs covering the whole or parts of

    the semantic domain which includes notions such as do/make/build, happen/occur go/turn

    into/come/become and act(like) (Stassen 1997: 92-93). Examples of languages mentioned

    by Stassen that employ this strategy are the Berber languages Tamazight and Shilha, various

    Tibeto-Burman languages, Lahu, Lisu and Cambodian. The second source is what Stassen

    calls copularization. This is a process where one of the members of the set of locational

    support verbs becomes specialized as the support item for nominal predicates (Stassen 1997:

    94). Commonly it is the semantically most neutral verb that undergoes this change. One

    further step in this process is the gradual loss of locative meaning of the copular verb, thus the

    language shifts towards a lexical differentiation of nominal and locational support items. The

    Tanoan language Kiowa appears to be such a language. The only examples in which the item

    /do:/ can act as a locative support where are sentences where it is accompanied by the particle

    heaway, gone (37). Other locational contexts require the use of a set of different posture

    verbs (38)-(39).4

    (36) Te: Koy-gu ba-do: [Kiowa]

    all Kiowa-UND2PL.UND-COP

    You are all Kiowas

    (37) Po: he: gya-do:

    moon away PL-COP

    There was (temporarily) no moon

    (38) Guy-te -to:

    other 3SG-stay

    There is someone else here

    (39) E:go yi: ol e-cel kicoy-ka

    here two hair 3DUAL-COPin.soup-inThere are two hairs in the soup (Stassen 1997: 94)

    4The abbreviations used in the glosses are as follows: 1, 2, 3 =person; ANIM = animate; ANT = anterior; AUX =auxiliary; C = noun class; CLASS = classifier; COMP = comparative case; COP = copula; DCL = declarative; DEF =definite; DEM = demonstrative; DETR = detransitivizer; DIST = distal; DUAL = dual; ERG = ergative case; ESS =essive case; EX = existential; FEM= feminine; FIN = finite; FM = focus-marker; FOCUS = focus/topic particle; FUT= future; GEN = genitive case; HAB = habitual; IMP = imperative; INAN = inanimate; INSTR = instrument; LOC =locative case; MASC= masculine;NM = nominalizer;NOM= nominative;NPL = nonplural marker; NPST = non-

    past; PAT = patient; PERF = perfective; PL = plural; PROG = progressive; PROX = proximal; PRS = present; PST =past; PX = proximity marker; REL = relative marker; S = part of verb stem; SBJ = subject; SEQ = sequentialconverb; SG= singular; SUPER = superessive case; UND = undergoer-marker.

  • 8/13/2019 Copula Cycle

    11/43

    11

    Other languages where copularization happens are the Yuman languages (Mojave, Cocopa,

    Yavapai) and in North Carolina Cherokee (Stassen 1997: 94). The third source is the process

    of verbalization, which can be viewed as the final stage of the grammaticalization process by

    which pronouns and discourse particles are reanalysed as abstract linking morphemes in

    predicate nominal sentences (Stassen 1997: 95). This process is especially frequent in many

    African languages, e.g. Temne, Gola, Zande, Gbeya, Zulu and Shona, to mention but a few

    (see Stassen 1997: 96 for a more comprehensive overview).

    In the present paper, I will offer a different suggestion. I will propose a typology based

    on structural positions. On the basis of the framework presented in section 3, I will try to

    explain the cyclic nature of the changes in terms of feature economy. Note, though, that my

    proposal is not incompatible with Stassens; it is another way of looking at the issues, and,

    hopefully also a way to contribute with some different and further insights. Basically, I will

    argue that there is a general copula cycle which looks like (40). Both the structural positions

    and the feature contents of the lexical items are present (cf. Chomsky 2000 et seq.).

    (40) demonstrative/pronoun > copula > grammatical marker

    specifier > head > affix

    iF > uF > --

    Several interesting questions emerge from (40). What kind of specifier position do

    demonstratives occupy? How do prepositions that develop into copulas fit into (40)? Are there

    sub-changes internal to the cycle in (40)? Below I will address all of these questions. Section

    4.2 discusses specifiers becoming heads, and I will argue that in most cases the relevant

    specifier is the specifier of a predication phrase. This specifier then turns into the head of the

    predication phrase (cf. the discussion in section 2). In section 4.3 I discuss the stage where the

    copula is a head, and argues that changes take place within this stage of the cycle. Section 4.4discusses heads becoming affixes, that is, grammatical markers. Section 5 provides a short

    summary.

    4.2. Specifier to head

    This section will discuss a number of examples showing that specifiers may develop into

    copulas. The main part of the section will be devoted to demonstratives and pronouns

    becoming copulas, but I will also somewhat briefly discuss existentials and prepositionsundergoing the same change.

  • 8/13/2019 Copula Cycle

    12/43

    12

    4.2.1. Demonstratives and pronouns

    There are several cases where a demonstrative or a pronoun has developed into a copula.5The

    most famous case is the one described in Li and Thompson (1977), and their paper will also

    play an important role in this section. However, towards the end I will also show that there are

    several other attested cases of this change. I also think that the theoretical framework adopted

    in this paper is a good alternative compared to functional theories of grammaticalization when

    it comes to analyzing this change. As Devitt (1994: 143-144) points out, the change from

    demonstratives and pronouns to copulas is somewhat problematic for grammaticization

    theory if semantic generalization is assumed to be the key process in the development of

    grammatical morphology. As will hopefully become evident, I think an advantage of the

    formal theory of grammaticalization adopted here is that it is able to deal straightforwardly

    with this change.

    The most widely know case of a pronoun becoming a copula is the Mandarinsh,

    discussed in Li and Thompsons (1977) seminal paper on the development of copula

    morphemes.6In Archaic Chinese (11th -3rd B.C.), equational sentences did generally not

    have a copula. This is shown in (41).

    (41) Wng-Ti w zh" y" [Archaic Chinese]

    Wang-Tai outstanding person DCL

    Wang-Tai is an outstanding person (Li and Thompson 1977: 421)

    In Modern Mandarin, the copulashregularly occurs in equational sentences:

    (42) ni-ge rn sh xush#ng [Modern Mandarin]

    that-CLASSman COPstudentThat man is a student (Li and Thompson 1977: 422)

    5In some cases, it may be hard to determine whether a lexical item is a copula or a nominal element (cf. Dprez2003). However, I will not discuss any such cases in this paper.6Diessel (1999: 143-147) argues against Li and Thompsons (1977) analysis. He claims that the developments ofnonverbal copulas from third person pronouns and from demonstratives follow two different pathways.

    However, I will argue that in both cases we are dealing with a case where a specifier becomes a head, whichjustifies subsuming the possible different pathways in the present paper. Possibly there might though be twoseparate sub-stages here.

  • 8/13/2019 Copula Cycle

    13/43

    13

    This modern copula was a demonstrative in Archaic Chinese, as shown in the following

    example:

    (43) f$-z% zh y sh b&ng y" [Archaic Chinese]

    Confucius arrive at this nation DCL

    Confucius arrived at this nation (Li and Thompson 1977: 423)

    However,shalso occurred in equational sentences in Archaic Chinese as an anaphoric

    demonstrative pronoun:

    (44) j y q sh#ng yo y q s' sh hu y"[Archaic Chinese]

    already wish him live also wish him die this indecision DCL

    wishing him to live while wishing him to die, that is indecision

    (Li and Thompson 1977: 424)

    Li and Thompson say that it seems fairly clear that the use ofshas a copula was productive

    by the late Han period (1st -2nd century A.D.). In the vernacular literature, one finds

    sentences such as (45) showingshas a copula.

    (45) c' bi w sh h d"ng?

    this white thing COPwhat kind

    What kind of stuff is this white thing? (Li and Thompson 1977: 425)

    The important thing, however, is that it is the use ofshas an anaphoric demonstrative

    pronoun that gave rise to the modern copula construction: These topic-comment

    constructions, then, set the stage for reanalysis: the topic-comment construction without acopula became a subject predicate construction with the anaphoric demonstrative pronounsh

    being reanalyzed as a copula (Li and Thompson 1977: 425). Importantly, Li and Thompson

    argue that there are some ambiguous data that favor this analysis. The data in (46)-(47) could

    in fact well be interpreted as subject-predicate equational sentences. They are given a topic-

    comment interpretation because there is no sentence in Archaic Chinese whereshfunctions

    solely as a copula verb.

  • 8/13/2019 Copula Cycle

    14/43

    14

    (46) zh( r sh%zh%, sh b rn y" [Archaic Chinese]

    know then use him, this not kind DECL

    To use him knowing (that he would rebel), that was unkind.

    (47) j y q sh#ng yo y q s%, sh hu y"

    already wish him live also wish him die, this indecision DCL

    Wishing him to live while wishing him to die, that is indecision.

    (Li and Thompson 1977: 424)

    Since (46)-(47) are open to multiple analyses, this made it very likely that children analyzed

    these strings differently at a later stage, given that there is sufficient representation of the

    structures in the input (Lightfoot 1999). Ambiguity in the input (the primary linguistic data) is

    a common source for language change (and often very little input is required; see Westergaard

    2008), thus it is not incidental that the demonstrative pronoun develops into a copula when we

    assume the principles of economy given in section 2. Let me now show how this can be

    analyzed formally.

    We have seen that a demonstrative can develop into a copula in Chinese. I will suggest

    that the demonstrative is in SpecPrP, and that through the HPP it is analyzed as head of the

    PrP. Using the sentences in (48) and (49) (repeated from above for expository convenience),

    there is a change from (50) to (51).

    (48) Wng-Ti w zh" y"

    Wang-Tai outstanding person DCL

    Wang-Tai is an outstanding person (=(41))

    (49) ni-ge rn sh xush#ng

    that-CLASSman COPstudent

    That man is a student (=(42))

    (50) PrP

    ei

    Wng-Ti Pr

    ei

    Pr w zh"

  • 8/13/2019 Copula Cycle

    15/43

    15

    (51) PrP

    ei

    nige rn Pr

    ei

    Pr xush#ng

    sh

    In order to see this change even clearer, consider the part in bold in (52) (=(46)).

    (52) zh( r sh%zh%, sh b rn y" [Archaic Chinese]

    know then use him, this not kind DECL

    To use him knowing (that he would rebel), that was unkind.

    The stage whereshwas a demonstrative is given in (53), and the new stage where it is

    analyzed as a copula is given in (54). The structures are simplified by not showing any of the

    higher functional projections.

    (53) PrP

    ei

    sh Pr

    ei

    Pr b rn

    (54) PrP

    ei

    6 Pr

    zh(r sh%zh% ei

    Pr b rn

    sh

    These examples show how a formal theory can be used to explain the changes in question.

    However, I would like to point out an important issue at this point. Van Gelderen (2004b: 73ff.) has invoked exactly this specifier-to-head change to explain how demonstratives may

  • 8/13/2019 Copula Cycle

    16/43

    16

    develop into complementizers. I think the fact that demonstratives become both

    complementizers and copulas show a crucial point: it is of utmost importance in which

    specifieran element is. If an element is in SpecCP, it may become a complementizer; if it is in

    SpecPrP, it may become a copula. Consequently, an element in SpecCP is predicted not to

    develop into a copula. The latter change has to the best of my knowledge not been attested.

    Notice that this is not a novel proposal about the framework I am adopting, but I think the

    data that have been discussed underline this fact even further. Let us now look at more cases

    of this change across various languages.

    Li and Thompson claim that analogous processes like the Chinese one have taken

    place in Hebrew, Palestinian Arabic and Wappo. What is particularly interesting for our

    purposes is that they argue that there are traits of full cycles in both Hebrew and Palestinian

    Arabic. I will use the former language to illustrate the process.

    In Hebrew, the triliteral verbal copula, h-y-yis not used in the present tense. Instead,

    one finds that equational sentences contain a demonstrative pronoun or a personal pronoun (if

    the subject is non-sentential):

    (55) [)e nitnaged lo] ze m*guxax [Hebrew ]

    that we.will.oppose.to him this absurd

    It would be absurd for us to oppose him.

    (56) mo)e (ze) student )eli

    Moshe that student my

    Moshe is a student of mine

    (57) ata (hu) ha-ganav

    you he DEF-thief

    You are the thief. (Li and Thompson 1977: 427)

    For most speakers, according to Berman and Grosu (1976), the humorpheme is optional in

    sentences containing pronominal subjects but obligatory in sentences with full DP subjects:

    (58) David hu ha-ganav [Hebrew]

    David he DEF-thief

    David is the thief

    (59) *David ha-ganavDavid THE-thief (Li and Thompson 1977: 428)

  • 8/13/2019 Copula Cycle

    17/43

    17

    There are several other restrictions on where huappears (Li and Thompson 1977: 428-429).

    These suggest, Li and Thompson argue, that pronouns are being reanalyzed as copulas in

    Modern Hebrew. Recent research has also supported this view: Colloquial Hebrew uses the

    overt copula with ever increasing frequency (Katz 1996: 87). Two arguments are advanced

    in favor of this claim. First, the morphemeszeand huare also able to function as pronouns:

    (60) ze me)une [Hebrew]

    Thats strange

    (61) hu ohev et-Rivka

    he loves ACC-Rivka

    He loves Rivka (Li and Thompson 1977: 429)

    Sentences such as (62) have completed the transition from topic-comment to subject-predicate

    equational sentences according to Li and Thompson (1977: 429-431).

    (62) David hu ha-ganav [Hebrew]

    David he DEF-thief

    David is the thief (Li and Thompson 1977: 429)

    I will not present the arguments in favor of this view. Instead I will direct attention to another

    phenomenon, discussed by Katz (1996). The following example shows that the copula does

    not agree in Person with its subject. The subject is second person, but the form of the copula

    is nevertheless the invariant hu.

    (63) Ata hu ha-i) [Hebrew]thou.MASCCOPDEF-man

    You are the man (Katz 1996: 90)

    Katz argues that this mismatch in persons shows that huis a copula and not merely a deictic

    marker: The loss of the category of person is one property which leads us to conclude that

    the erstwhile pronouns have grammaticalized in such utterances, and are for all intents and

    purposes full fledged copulas (Katz 1996: 90; cf. Devitt 1994: 140).

  • 8/13/2019 Copula Cycle

    18/43

    18

    An interesting important point is nevertheless that we have seen a development where

    a copula disappears (h-y-y) and a new copula hudevelops from a pronoun. As Li and

    Thompson (1977: 438) say, Thus, it seems that the emergence and decline of the copula may

    be seen as a diachronic cycle. The cycle is clearly very complex. This is further supported by

    data in Katz (1996) who presents evidence that copulas may become pronouns.

    Katz (1996: 118-133) argues that the Turkish third person singular pronoun o(n)

    developed from the third person singular present of the copula verb. In pre-nineteenth century

    texts, the usual form of the third person absolutive was ol, and not oas it is today. The olroot

    is also found in another instance, namely in the verb olmakto exist. Katz says that olwas

    also used as the equivalent of the copula in the early period. As evidence for the directionality

    ol> o, she points towards the fact that in earlier texts, both were used interchangeably. In

    modern texts, ois the only form. Regarding the verb olmakto exist, it is clear that this verb

    shares the common root with the demonstrative pronoun olsince -makis the infinitival

    ending. Katz (1996: 122-123) also presents two main arguments for why the pronoun

    developed from the copula and not the other way around. First, it is only in the very earliest

    texts that the pronoun has a copula function. This seems to indicate that the copula function

    emerged earlier, and that the pronoun is the new form. Second, early Turkish oral texts used

    independent pronouns sparsely whereas Modern Turkish uses them amply. Here too, the

    pronoun form seems to be the innovative one. Based on this, it seems clear that the verbal

    form oldeveloped into the pronoun o(n) (see Katz 1996: 120 on the nasal).

    Going back to demonstratives becoming copulas, another interesting case is Panare.

    Panare has three different copulas, as shown in (64)-(66).

    (64) maestro kj eapa [Panare]

    teacher COP.ANIM.PROXPanare

    The Panare is (PROX) a.teacher(65) maestro nj eapa

    teacher COP.ANIM.DISTa.teacher

    The Panare is (DIST) a teacher.

    (66) echipen mn manko

    fruit INANmango

    Mango is a fruit (Gildea 1993: 55)

  • 8/13/2019 Copula Cycle

    19/43

    19

    These examples show that the different copula depends on whether the third person subject is

    animate or inanimate and whether they are inside or outside the speakers sphere of

    perception. First and second person subjects in Panare have no copula:

    (67) maestro yu/amen [Panare]

    teacher 1SG/2SG

    I am/You are a teacher (Gildea 1993: 54)

    Gildea (1993: 56) says that: It is clear on morphological grounds that the copulas kj, nj and

    mn are derived from nonverbal elements in that verbs take tense/aspect suffixes and prefixes

    which agree with the subject for person, while these copulas take no affixes and they vary

    suppletively to agree with the subject for animacy. The nonverbal forms which most

    resemble these copulas are the pronouns knANIM.INVISIBLE, njwho? and mn

    INAN.INVISIBLE. Gildea argues that the nonverbal copulas and these pronouns are derived

    from the same historical set of pronouns, viz. demonstratives. There are two reasons for this

    claim. First, demonstrative pronouns are not used with first or second persons, which fits well

    with the fact that nonverbal copulas neither are used with these persons (67). Second, the

    deictic opposition is common for demonstrative pronouns, cf. English thisand that. As

    Gildea (1993: 58) points out, there is one difference between Panare and the examples

    discussed by Li and Thompson (1977). In the latter case, all word orders are subject-predicate,

    whereas in Panare the order is predicate-subject. He proposes that in Panare we have left

    dislocation (instead of right dislocation as in Chinese):A teacher [is] he, John. Thus the

    demonstrative was required as the subject, and a new analysis happened. It seems likely that

    left dislocation has worked as a kind of trigger in Panare, just as right dislocation has in

    Chinese. That is, both types of dislocations provided a natural domain for the emergence of

    the copula.7What happens further, and which shows that the demonstrative has separated from the

    subject semantically is that the deixis expressed by the copula may differ from that expressed

    by the subject. An example is shown in (68).

    7It may be that prosody plays a role in these changes, but that question goes beyond the scope of this study.

  • 8/13/2019 Copula Cycle

    20/43

    20

    (68) maestro nj mj [Panare]

    teacher ANIM.DIST ANIM.VISIBLE

    This (here) guy was a teacher.

    [he (PROX) is (DIST) a teacher] (Gildea 1993: 61)

    The subject in (68) is mjANIM.VISIBLE, so the spatial dexis is proximal (i.e. the person is

    right here); the copula is njANIM.DIST, so the temporal deixis is distal (i.e. not right now).

    Distal temporal deixis is most commonly interpreted as past. Based on this, Gildea argues that

    the historical demonstrative pronouns have become synchronic tense markers. In the final

    step, these copulas develop into auxiliaries. The following examples show how kjand njare

    used as verbal auxiliaries for aspect-marked verbs ( is a non-agreeing intransitive prefix).

    (69) pmanpj kj Toman

    A.DETR.hit.IMPERF.IANIM.PROXToman

    Tom is falling (and landing with an impact)

    (70) ?pmanpj nj Toman

    A.DETR.hit.IMPERF.IANIM.DISTToman

    Tom is falling (DIST) (Gildea 1993: 65)

    The only difference between these clauses is the choice of auxiliary. In (69), the auxiliary is

    kj, and as the translation shows, the clause is in the present tense. (70) has the auxiliary nj,

    and although the clause is translated into present tense, it is understood to be occurring at a

    spatially distal location (Gildea 1993: 65). Gildea (1993: 67-68) then shows that these

    auxiliaries share syntactic properties with prototypical tense markers in Panare; see his work

    for discussion.

    Homonymies are also indicative of grammaticalization processes. In the languageKenya Luo, the third person pronoun can optionally be used as a copula, as shown in (71).

    (71) dhk +n br [Kenya Luo]

    woman COPgoodness

    woman she is goodness (Tucker 1993: 308)

    (72) dhk ber

    woman goodnesswoman she is goodness (Tucker 1993: 308)

  • 8/13/2019 Copula Cycle

    21/43

    21

    In Lango, a language closely related to Kenya Luo, the grammaticalization has moved further.

    The original pronominal function is not transparent any more, and the copula !noccurs not

    only with third person singular subjects:

    (73) n ,n dktl [Lango]

    1SG COP 1SG.doctor.HAB

    I am the doctor (Noonan 1992: 146)

    Again we see a process where pronouns develop into copulas. This is also the case in Lakota.

    In this language the copula hc"ais etymologically based on the demonstrative pronoun h

    this and the element c"asuch. Neither hnor c"aare compatible with verbal inflectional

    categories: hc"a, however, can be inflected like a stative intransitive verb (Pustet 2003: 56):

    (74) wa)cu he-m-cha [Lakota]

    white.man S-1SG.PAT-COP

    I am a white man (Pustet 2003: 68)

    (75) Phalni he-m-cha

    Pawnee.Indian S-1SG.PAT.COP

    I am a Pawnee Indian (Pustet 2003: 69)

    Lastly, let us have a look at one more case study. McWhorter (1997) (see also

    McWhorter 1995) discusses whether copulas in English-based Atlantic creoles emerged due

    to influence from West African copula systems or whether copulas are the result of

    independent developments. Some data relevant to this issue are provided in (76)-(77). (76)-

    (77) show that Ewe uses one copula in an equative context and another copula in a locative

    context.

    (76) L -nye t-mel [Ewe]

    crocodile he-COPaquatic.ANIMAL

    The crocodile is an animal that lives in the water

  • 8/13/2019 Copula Cycle

    22/43

    22

    (77) -l x- me

    he-COPhouse in

    He is in the house (McWhorter 1997: 243)

    The Sranan creole has a similar pattern, as shown in (78)-(79).

    (78) Mi na datra [Sranan]

    I COPdoctor

    I am a doctor

    (79) A de na ini Sranan Kondre

    he COP LOC inside Surinam

    He is in Surinam (McWhorter 1997: 243)

    McWhorther argues against the hypothesis that the copula systems in creoles are the result of

    influence from the West African languages (though see Migge 2002 for a critical discussion).

    On the basis of data from Arends (1989), he shows that a zero copula is possible in the oldest

    Sranan documents (the example is from 1770).

    (80) Mi blibi joe wan bon mattie fo dem. [Sranan]

    I believe you a good friend for them

    I believe youre a good friend of theirs (McWhorter 1997: 244)

    Arends argues that the copula evolved from the demonstrative dathat:

    (81) Adjabre, da Djutongo [Sranan]

    adjabre that Jews languageAdjabre is Saramaccan (McWhorter 1997: 244)

    The modern copula emerged when what McWhorter calls a resumptive to a preceding topic is

    analyzed as a copula. That is, the demonstrative dais analyzed as a copula at a later stage.

    McWhorter also discusses the development of the locative copula de. He discusses

    several ways of analyzing this issue (1997: 247-250), which I will not go into here. Instead, I

    will just give his suggestion as to how this copula emerged. In contemporary Saramaccan, the

  • 8/13/2019 Copula Cycle

    23/43

    23

    deictic adverb deis often optionally inserted into sentences in order to lend deictic emphasis.

    Thus using deshows that the utterance is expressive.

    (82) Niti fa mi de a Winiki d!, niti mi ji ta [Saramaccan]

    never since I COP LOCWinikii there never I hear talk

    Never since Ive been there at Winikii have I heard that

    (83) D Gaam d Kfi g lku d! d. ku suwki d!

    THEchief RELKofi go see there COPwith sickness there

    The chief who Kofi went to look at is sick (McWhorter 1997: 250)

    McWhorter argues that it is from this usage that dein Modern Saramaccan is derived. At first

    there was no expression of the locative copula, just as we have seen for the equative copula:

    (84) D wmi a wsu [Saramaccan]

    THEman LOChouse

    The man is at home (McWhorter 1997: 250)

    McWhorter says that however, it would have been a common expressive strategy to insert an

    expressive deictic debetween subject and the predicate, similar to todays usage (1997: 250):

    (85) D wmi de a wsu [Saramaccan]

    THEman there LOChouse

    The man is there at home (McWhorter 1997: 250)

    This deictic marker was now ripe to be analyzed differently by new speakers as these speakers

    already had heard a number of Ewe sentences where the copula was obligatory:

    (86) l x- me [Ewe]

    he COPhouse in

    He is in the house (McWhorter 1997: 251)

    Thus, deis no longer an adverbial but a copula:

    (87) D wmi d a wsu [Saramaccan]

  • 8/13/2019 Copula Cycle

    24/43

    24

    THEman COP LOChouse

    The man is at home (McWhorter 1997: 251)

    Hence, the transition to a copula is complete.

    In sum, we have seen a number of examples where demonstratives and/or pronouns

    develop into copulas. I have claimed that this is an instance of a preference in acquisition for

    analyzing elements as heads instead of specifiers, following van Gelderen (2004a, b et seq.).

    In the next section I show another case of this change, namely copulas emerging from

    existentials.

    4.2.2. Existentials

    Another pathway where copulas emerge is from existentials. This is not a widely attested

    change, but data from Chalcatongo Mixtec show that it exists: Mixtec also allows a

    construction in which the existential is used with an adjective. This construction []

    apparently has the same meaning as the normal copula + adjective construction (Macaulay

    1996: 87). To see this clearly, observe that the normal copula in Mixtec is ka.

    (88) x! ka ll [Chalcatongo Mixtec]

    Juan COPsmall

    Jus is small/short (Macaulay 1996: 130)

    The existential marker in the language is !oothere is:

    (89) in kax wa" /oo tenn [Chalcatongo Mixtec]

    insides box DEFEX tomato

    in the box there are tomatoes (Macaulay 1996: 129)

    The existentiale can also be used as a copula, which arguably is the new function.

    (90) k#s#/ x [Chalcatongo Mixtec]

    pot COP new

    the pot is new/there is a new pot (lit. as for the pot, it exists new)

    (Macaulay 1996: 130)

  • 8/13/2019 Copula Cycle

    25/43

    25

    These two cases show that existentials may develop into copulas. I will argue that this change

    is just another specifier to head change. That is, simplified structures of (89) and (90) would

    look like (91) and (92).

    (91) PrP

    ei

    /oo Pr

    ei

    Pr tenn

    (92) PrP

    ei

    k#s# Pr

    ei

    Pr x

    /

    This presupposes a certain view on existentials. It seems reasonable to analyze these

    existentials on a par with expletives like English there, which means that they are phrases. I

    will assume, following the argumentation in Richards (2004), Richards and Biberauer (2005)

    and Biberauer and Richards (2006) that expletives are merged in what corresponds to SpecPrP

    in my framework (cf. also farli 2005 within a somewhat different perspective). Given that,

    the analysis should be straightforward.

    4.2.3. Summary

    In this section I have discussed a number of examples of what I have analyzed as a change

    from an element being a specifier to an element becoming a head. The main cases have been

    demonstratives and pronouns that develop into copulas, but we have also seen that existentials

    may undergo a similar change. In the next section, I will discuss cases where a head develops

    into a different head.

  • 8/13/2019 Copula Cycle

    26/43

    26

    4.3. Head-to-head

    There are a number of cases where full verbs develop into copulas. However, there are also

    cases where copulas develop into auxiliaries. Both of these changes do not involve change of

    projection type both are heads but I will argue that the nature of the heads change. In this

    section I will provide examples of these changes and an analysis.

    4.3.1 Full verbs

    It is well-attested in the literature that full verbs may develop into functional elements. A

    famous case is the development of English full verbs into modals (see the references in

    section 3). A similar change is where full verbs become copulas. This has even happened in

    English; the past tense forms of become from wesmeaning to stay, to remain (Devitt 1994:

    128). In this section I will give some further examples and an analysis.

    In Alyawarra, copula constructions either use the verb animato sit or ayntimato

    lie. These verbs have still retained their basic postural sense. This is shown in (93)-(94).

    (93) an-il-ayant-a nhilantiya [Alyawarra]

    sit-LIG-AUX-IMPhere.still

    Sit here for a while ; stay here

    (94) ayinga alingkarr-a an-ina

    I.NOMtired-NOMsit-PAST.CONT

    I was tired; I sat feeling tired. (Devitt 1994: 129)

    Similar cases are found in other languages. The simplex verb ci-in Chantyal has extended it

    meaning beyond the expression of a basic posture sense. The original use is shown in (95),

    and the new copula meaning in (96)-(97).

    (95) Ram kurci-phyara0ci-I [Chantyal]

    Ram chair-SUPER sit-PERF

    Ram sat on the chair (Noonan and Grunow-Hrsta 2002: 82)

    (96) cater n* k$i n* mast*r ci-wa p*ri-m b$i-i

    then FOCUSyou FOCUS teacher sit-NM happen-NPSTsay-PERF

    Your have to be the teacher, they said from that time on

    (97) m*%ya k$i-ye $*nuwar lu%-i ci-si-mlove you-GENappearance shine-ANTsit-ANT-NPST

  • 8/13/2019 Copula Cycle

    27/43

    27

    (My) love, your appearance seems to shine (Noonan and Grunow-Hrsta 2002: 87)

    There is also a copula, mu(with the allomorphs mand mu), which derives historically from

    a verb meaning sit/stay. This copula has developed into an auxiliary verb and a non-past

    suffix on verbs (cf. Turunen 2006 for a similar change in Erzya). (98)-(99) show the auxiliary

    and (100)-(101) show the affix.

    (98) na kam-ri t$o-wa-b$*nda p*yle, Ram kam la-wa mu [Chantyal]

    I work-LOCarrive-NM-COMP before Ram work do-NM COP.NPST

    Even before I get to work, Ram will be working

    (99) c* din n* b*nnu n* b$*r*-ysi-wa mu-wa $in-si-m

    that day FOCUSgun FOCUSfill-ANT-NOMCOP-NM COP-ANT-NPST

    That day, it turned out that the gun was loaded

    (Noonan and Grunow-Hrsta 2002: 89)

    (100) na-s*sar-mu

    I-ERGkill-NPST

    Ill kill it

    (101) *mrika-ne-ma-thy la-si-r* ci-m

    America-person-PL-ESSdo-ANT-SEQsit-NPST

    He sits down like an American (Noonan and Grunow-Hrsta 2002: 88)

    Interestingly, we see almost a full cycle here. First we have a full verb, then a copula, then an

    auxiliary and finally an affix. Below we will see further instances of these separate stages, but

    it is noticeable that we seem to find all of them gathered in one language. This also speaks in

    favor of the present theory as we would expect such cycles to emerge in individual languages.

    Munro (1977) argues on the basis of data from Mojave that existential constructions

    also should be included in the possible sources for copulas (see also Hengeveld 1992: 254-

    255; and see e.g. Devitt 1994: 138 for a dissenting view on the direction of the change). The

    discussion centers around sentences like (102).

    (102) John k!a&'ide:-1-ido-p1 [Mojave]

    John doctor-SBJ 3SG-COP-PERF

    John is a doctor (Munro 1977: 445)

  • 8/13/2019 Copula Cycle

    28/43

    28

    It is problematic to analyze a sentence like (102). The subject isJohn, although k#a$%ide:

    doctor is grammatically marked as the subject. Munro (1977: 450) proposes that the subject

    actually isJohn k#a$%ide:John (is) doctor, i.e a predication embedded by the verb ido. That

    is, the paraphrase is better given as (103).

    (103) There is [John (is) doctor]

    Under this analysis, the subject is the embedded non-verbal predication, and the existential

    verb may be expected to show agreement with this (inherently third person singular). The

    following examples show that this prediction is borne out.

    (104) Man('n(e-1 -ido-p1 [Mojave]

    2SG 1SG-SBJ 3SG-COP-PF

    Youre me.

    (105) 'in(ep k!a&'ide:-1-ido-p1

    1SG doctor-SBJ 3SG-COP-PF

    Im a doctor. (Munro 1977: 452-453)

    However, if the verb in (102) were a regular copula, we expect that it would agree with the

    subject of the embedded non-verbal predication. This is indeed found:

    (106) 'in(ep k!a&'ide:-1'-ido-p1 [Mojave]

    1SG doctor-SBJ 1SG-COP-PF

    Im a doctor (Munro 1977: 452)

    As Hengeveld (1992: 255) argues: The fact that [(102)] and [(106)] exist side by side may be

    taken as a sign of copularization of the existential verb used in a reality construction.

    Regarding the analysis of these cases, I claim that this change is fully compatible with

    the perspective adopted in this paper. Whereas full verbs project a VP and moves from V to

    Pr (copies indicated by strikethrough below), the copula is merged directly in Pr. In other

    words, instead of the structure in (107), we have the structure in (108).

  • 8/13/2019 Copula Cycle

    29/43

    29

    (107) PrP

    ei

    SUBJECT Pr

    ei

    Pr VP

    verb ei

    V

    ei

    V

    verb

    (108) PrP

    ei

    SUBJECT Pr

    ei

    Pr PREDICATE

    copula

    In (108) the copula is directly merged in the PrP, thus there is no VP below it. Alternatively,

    there might be a VP, but then this VP is lexicalized by a full verb (see Pustet 2003 for

    languages where this is the case), and there is no movement from V to Pr. In other words, this

    change is a consequence of the Late Merge Principle. The verb goes from a lexical element to

    a functional element. If argument structure is part of syntax through features (Hornstein 1999,

    2001), these features will be lost when a full verb changes into a copula. In the next section,

    we will look at auxiliaries that develop from copulas.

    4.3.2. Auxiliaries

    Copulas that have developed from full verbs can also develop further. As Kuteva (2001: 67)

    mentions, a copula derived from a bodily posture verb can develop further into an auxiliary.8

    The Latin postural verbstarestand developed into the copula estarin Spanish, meaning to

    8This shows that semantic restrictions are important to the cycle as well. However, I will not try to deal with thespecific role of semantics in the present paper.

  • 8/13/2019 Copula Cycle

    30/43

    30

    be (somewhere, or temporarily) (Comrie 1976: 102). In Modern Spanish, this verb has

    developed into a progressive auxiliary (109). We find the same structure in Italian (110).

    (109) estoy cantando [Spanish]

    I.stand singing

    I am singing

    (110) sto cantando [Italian]

    I.stand singing

    I am singing (Comrie 1976: 102)

    Similar developments are attested in Imonda (Kuteva 2001: 65):

    (111) ag-iani sabla ed-ia ekuk lh-ual-fna [Imonda]

    women-NPLtwo PX-LOCdistance stand-DUAL-PROG

    The two women were standing there in the distance.

    (112) pilin ed-ia fa-hd-lh-f.

    plate PX-LOCCLASS-put.up-COP-PRS

    The plate is up there. (Heine and Kuteva 2002: 282)

    Though, as Kuteva points out, this is by no means a universal auxiliation process (2001: 67).

    However, we see traits of it in English, cf. (113).

    (113) The car is arriving.

    Devitt (1994) argues that the directionality in these cases is one where copulas develop into

    auxiliaries. He notes a strong tendency for forms that serve as an auxiliary for progressive orcontinuous aspect to also appear as copulas in locative complements. This is shown for two

    languages in (114)-(117).

    (114) amu e idu tani manji manamu [Kui]

    1P(IN) that house in COP.PRS.PRTCOP.1P(IN)

    We are staying in that house

    (115) anu lakai mai1S sacrifice.PRS.PRTCOP.1S

  • 8/13/2019 Copula Cycle

    31/43

    31

    (116) nga di der al [Nung]

    1S come PART COP PRES

    I am (in the act of) coming

    (117) magma hpe chum hka ma al

    chief MASChouse in NEG COP PRES

    The chief is not in his house (Devitt 1994: 174)

    According to Devitt, in these cases the auxiliaries have developed from copulas. I will now

    show how these cases can be analyzed.

    An auxiliary is syntactically different compared to a copula in that the auxiliary is

    commonly taken to be directly merged in the middle field, the IP domain. Once again, this is a

    Late Merge effect. The change can be illustrated by comparing the somewhat simplified trees

    in (118) and (119).

    (118) IP

    ei

    I

    ei

    I PrP

    copula ei

    Pr

    ei

    P

    copula

  • 8/13/2019 Copula Cycle

    32/43

    32

    (119) IP

    ei

    I

    ei

    I PrP

    auxiliary ei

    Pr

    ei

    P VP

    verb 5

    verb

    In (118) the copula is merged in PrP and moved to IP (in order to get inflected for tense and

    other agreement properties). In (119), the auxiliary is directly merged in IP, and there is a

    different verb in the argument domain of the clause. The next section discusses a slightly

    different case where a head develops into a new head, namely adpositions becoming copulas.

    4.3.3. AdpositionsAdpositions may be taken as another possible source of copulas. In Gbeya, the copula n!"

    resembles the preposition n!"in several respects (Samarin 1966: 76). The copula n!"is

    peculiar because it is used in combination with other copular elements, and seems to occur

    exclusively with the verbs &andya, the singular and plural verbs to be (Samarin 1966: 77).

    (120) wa y n)2 wey [Gbeya]

    3PL COP.PL COPmanthey are men

    (121) ta *2 n) bu

    tree COP.SG. COPwhite

    the tree is white (Samarin 1966: 77)

    Frajzyngier (1986) discusses Chadic languages and argues that a grammaticalization from

    preposition to copula has happened in several of these languages. He calls the copulas in theselanguages locative copulas. A few examples are provided in (122)-(124).

  • 8/13/2019 Copula Cycle

    33/43

    33

    (122) kn g m [Bolewa]

    buffalo COP in water

    A buffalo is in the water

    (123) kn g m s

    buffalo COPin water not

    The buffalo is not in the water

    (124) s+-n k rw

    shirt-1SG COPon tree

    My shirt is on the tree (Frajzyngier 1986: 372)

    (122)-(124) are all ungrammatical if either or the preposition following is deleted. These

    examples can be categorized as belonging to the type (125). There is another type as well

    (126), which makes it possible to test whether we are dealing with a copula in (122)-(124), or

    whether we are dealing with two prepositions in a row.

    (125) X is in/at Y

    (126) X VERB Z in/at Y

    If is a preposition, we would expect it to occur in (125) as well as in (126). However, if is

    a copula, we would not expect it to occur in (126). The relevant data are provided in (127)-

    (129), which show that the copula disappears in this configuration.

    (127) zet s+ g k rw [Bolewa]

    put shirt in on tree

    Put the shirt on a tre

    (128) sn zu s+-n g g ngrk

    3SGput shirt-3SG in in bag

    He put his shirt into a bag (Frajzyngier 1986: 372)

    (129) mm m g klb g g sr-n

    person DEMwith bottle in in hand-3MASC

    That man has a bottle in his hand (Frajzyngier 1986: 373)

  • 8/13/2019 Copula Cycle

    34/43

    34

    Frajzyngier also discusses the direction of the change: Did locative copulas emerge from

    prepositions or did prepositions emerge from locative copulas? He discards the latter option

    based on the fact that prepositions are far more frequent than copulas, thus it makes more

    sense that in some languages a copula emerged than to say that such a copula disappeared

    from most of the languages.

    The question is now how we are to analyze this change within the current framework.

    On standard assumptions, both prepositions and copulas are heads. Adopting the proposal in

    Pesetsky and Torrego (2004: 518), we can assume that prepositions have the following feature

    content (I have modified the notation slightly).

    (130) PP

    ei

    P DP

    uphi iphi

    iCase uCase

    In (130), uphi stands for unvalued phi-features whereas iphi stands for valued phi-features.

    Pesetsky and Torrego (2001, 2004) argue that Case is best understood in terms of a displaced

    Tense feature, but I have just put Case instead of T for expository convenience. The

    unvalued phi-feature makes prepositions into a probe, which is necessary given the Agree

    framework in Chomsky (2000, 2001, 2004, 2005, 2007). The Case feature is a valued feature

    on P on this approach. Assuming the Economy of Features principle, we can argue that the

    learner will try to get rid of this feature if possible. When this Case feature is removed, the

    feature content is identical to copulas. A possible problem here is that the feature content also

    resembles complementizers, and a very similar story has been invoked in order to explain how

    prepositions may become complementizers (van Gelderen 2008b). At present it is not entirely

    clear what the main difference is. Judging from the data above, it seems that those

    prepositions that become copulas have a limited distribution. Specifically, they tend to occur

    in the middle field, which is not the case for many of those prepositions that become

    complementizers, at least not in English (van Gelderen 2008b). This might provide the learner

    with a cue, that is, there need to be a certain structural relationship in order for a preposition

    to become a copula in order for the Head Preference Principle to be involved in relation to the

    PredP. This is also related to the discussion regarding whether a demonstrative develops into a

  • 8/13/2019 Copula Cycle

    35/43

    35

    complementizer or a copula, cf. section 4.2.1. At present I am not able to formulate very

    specific criteria for this relationship, but hopefully future research will elucidate this.

    Heine and Kuteva (2002: 100) also quote a few cases of a locative copula developing

    into a locative preposition. One such example is lebe at in Ewe with becomes a preposition

    meaning at:

    (131) agbal)3 le kpl*3 dz. [Ewe]

    Book.def be.at table.def on

    The book is on the table.

    (132) me kp*2l*2ri le m*2dz.

    1sg see lorry at street top

    I saw a lorry on the street. (Heine and Kuteva 2002: 100)

    Heine (1993) and Heine and Kuteva (2002: 101) view this as an instance of

    grammaticalization. In the present framework, I would view such examples as cases of

    renewal. That is, a cycle starts over again, with the new preposition having a potential to

    redevelop into a copula, given the adequate external influence such that one of the economy

    principles is able to work.

    In this section we have seen a few examples of prepositions that change into copulas.

    In the next main section, I discuss cases where a head develops into a grammatical marker.

    4.4. Head to affix

    So far we have seen changes where a specifier becomes a head and where a head becomes a

    different head. In this section, I will look at cases where heads develop into affixes.

    Typologically, copulas as affixes are not very frequent: In the majority of cases, copulas are

    free morphemes (Pustet 2003: 39).

    The copula niin Yoruba is homophonous with the general focus marker in the

    language (Stassen 1997: 96):

    (133) Oni,wo ni mi [Yoruba]

    merchant COP 1SG

    I am a merchant (Ashiwaju 1968: 28)

  • 8/13/2019 Copula Cycle

    36/43

    36

    (134) Ll)2 ni w*2n d

    in.the.evening FMthey arrived

    It was in the evening that they arrived (Bamgb-se 1966: 37)

    Verbal copulas may evolve into affixes. Modern Turkish has two different copular affixes,

    though they occur in a very restricted set of contexts. One of them is the palatal glidey.

    Notice that it can only be used after non-consonants (135):

    (135) (ben) sat-c--y--m [Turkish]

    1SG seller-COP-1SG

    I am a seller (Kornfilt 1997: 77)

    (136) (ben) 4retmen-im

    1SG teacher-1SG

    I am a teacher (Kornfilt 1997: 78)

    As Kornfilt argues,yis an insertion rule restricted to morpheme boundaries in copula

    sentences. If it were a general phonological rule ofy-insertion between vowels, even restricted

    to morpheme boundaries, we would expect (137) to be grammatical.

    (137) *kitab-- oku-du-y-um [Turkish]

    book-ACCread-PAST-Y-1SG (Kornfilt 1997: 78)

    There is a diachronic fact supportingyas a copula. Until the early years of the Republic (that

    is, in the early 20th century), tense and agreement markers in copula sentences were attached

    to a free morpheme iin the past tenses (138)-(139).

    (138) (ben) sat-c-i-di-m [Turkish]

    1SG seller COP-PST-1SG

    I was a seller

    (139) (ben) 4retmen i-di-m

    1SG teacher COP-PST-1SG

    I was a teacher (Kornfilt 1997: 80)

  • 8/13/2019 Copula Cycle

    37/43

    37

    These inflected copular forms then developed to postclitics, and during this change, the

    copular iturned into the glidey. Still one does find the free inflected copula in written

    Turkish, according to Kornfilt, which nicely illustrates that there was such a previous stage

    and that the transition has not yet made it entirely into writing. Thus, the reduced forms are:

    (140) (ben) sat-c--y-di-m [Turkish]

    1SG seller-COP-PST-1SG

    I was a seller

    (141) (ben) 4retmen-di-m

    1SG teacher-PST-1SG

    I was a teacher (Kornfilt 1997: 79)

    There are other analyses ofyas a copula; Jansky (1954: 36) argues thatyis a linking

    consonant that appears in specific phonological contexts. However, I think Kornfilt has

    provided convincing evidence that this is not an adequate analysis and thatyindeed is a true

    copula.

    The other copula element in Turkish is the suffix DIr, a suppletive form for the

    copula (Kornfilt 1997: 81). The historical source of DIris the verb to stand (Devitt 1994:

    128, Pustet 2003: 58). It is used in the third person present only, and it is always optional, but

    this is the only place where it is optional (Kornfilt 1997: 82).

    (142) bu cadde-dir [Turkish]

    this road-COP.3PRS

    this is a road

    (143) bu cadde

    this roadthis is a road (Pustet 2003: 58)

    Again, we see a copula emerging from a previous full verb.

    4.5. Summary

    In this section I have provided a number of examples in support of a copula cycle. The cycle,

    I have argued, looks like (144).

  • 8/13/2019 Copula Cycle

    38/43

    38

    (144) demonstrative/pronoun > copula > grammatical marker

    specifier > head > affix

    iF > uF > --

    I have argued that the major transitions are those where a specifier becomes a head and a head

    becomes an affix. However, I have also claimed that there are minor stages within each main

    stage in (144). In particular, a verb may develop into a copula. Since a verb is already a head,

    this is not a category change. Instead, I have argued that verbs and copulas are merged in

    different positions, viz. the VP and the PrP.

    5. Conclusion

    The aim of this paper has been twofold. On the one hand, I have wanted to describe the major

    patterns in which copulas emerge and disappear, what I have called the copula cycle. On the

    other hand, I have tried to show that a formal theory of grammaticalization is a good

    framework in order to account for and explain these changes. When demonstratives and

    pronouns become heads, this is a change from the specifier of a predication phrase to a head

    of the same phrase. In terms of features, it means that interpretable features are replaced by

    uninterpretable features. I have also shown how full verbs become copulas, and subsequently

    how copulas may turn into auxiliaries or grammatical markers. In sum, I have argued in favor

    of a copula cycle involving three major stages, but also that these stages in many cases have a

    more fine-grained internal cartography.

    References

    farli, Tor A. 2005. The Basic Nexus. Ms., Norwegian University of Science andTechnology.

    Abraham, Werner. 2002. Modal verbs: epistemics in German and English.Modality and its

    interaction with the verbal system, Sjef Barbiers, Frits Beukema and Wim van der

    Wurff (eds.), 19-50. Amsterdam: John Benjamins.

    Adger, David and Gillian Ramchand. 2003. Predication and Equation.Linguistic Inquiry34:

    325-359.

    Ashiwaju, Michael. 1968.Lehrbuch der Yoruba-Sprache. Leipzig: VEB Verlag.Bamgbos5e, Ay-. 1966.A grammar of Yoruba. Cambridge: Cambridge University Press.

  • 8/13/2019 Copula Cycle

    39/43

    39

    Baker, Mark C. 2003.Lexical Categories. Cambridge: Cambridge University Press.

    Bhat, D. N. S. 1994. The Adjectival Category. Amsterdam: John Benjamins.

    Biberauer, Theresa and Marc Richards. 2006. True Optionality: When the grammar doesnt

    mind.Minimalist Essays, Cedric Boeckx (ed.), 35-67. Amsterdam: John Benjamins.

    Bowers, John. 1993. The syntax of predication.Linguistic Inquiry24: 591-656.

    Bowers, John. 2001. Predication. The Handbook of Contemporary Syntactic Theory, Mark

    Baltin and Chris Collins (eds.), 299-333. Malden: Blackwell.

    Chomsky, Noam. 1995. The Minimalist Program. Cambridge: MIT Press.

    Chomsky, Noam. 2000. Minimalist Inquires. Step by Step. Essays on Minimalist Syntax in

    Honor of Howard Lasnik, Roger Martin, David Michaels and Juan Uriagereka (eds.),

    89-155. Cambridge, MA: The MIT Press.

    Chomsky, Noam. 2001. Derivation by Phase.Ken Hale: A Life in Language, Michael

    Kenstowicz (ed.), 1-52. Cambridge, MA: The MIT Press.

    Chomsky, Noam. 2004. Beyond Explanatory Adequacy. Structures and Beyond: The

    Cartography of Syntactic Structure, Volume 3, Adriana Belletti (ed.), 104-131.

    Oxford: Oxford University Press.

    Chomsky, Noam. 2005. On Phases. Ms., MIT [To appear in Foundational Issues in

    Linguistic Theory, Robert Freidin, Carlos P. Otero & Maria-Luisa Zubizaretta (eds.).

    Cambridge, MA: The MIT Press.]

    Chomsky, Noam. 2007. Approaching UG from Below.Interfaces + Recursion

    = Langauge? Chomskys Minimalism and the View from Syntax-Semantics, Hans M.

    Grtner & Uli Sauerland (eds.), 1-30. Berlin: Mouton de Guyter.

    Comrie, Bernard. 1976.Aspect: an introduction to the study of verbal aspect and related

    problems. Cambridge: Cambridge University Press.

    den Dikken, Marcel. 2006.Relators and Linkers. The Syntax of Predication, Predicate

    Inversion, and Copulas. Cambridge, MA: The MIT Press.

    Dprez, Viviane. 2003. Haitian Creole Se: A Copula, a Pronoun, Both or Neither? On the

    Double Life of a Functional Head,Recent Development in Creole Studies, Dany

    Adone (ed.), 135-173. Tbingen: Max Niemeyer.

    Devitt, Daniel. 1994. Copula Constructions in Crosslinguistic and Diachronic Perspective.

    Doctoral dissertation, SUNY Buffalo.

    Diessel, Holger. 1999.Demonstratives: form, function, and grammaticalization. Amsterdam:

    John Benjamins.Eid, Mushira. 1983. The Copula Function of Pronouns.Lingua59: 197-207.

  • 8/13/2019 Copula Cycle

    40/43

    40

    Eide, Kristin Melum. 1996. Som-predikativer. Cand.philol. thesis, Norwegian University of

    Science and Technology.

    Eide, Kristin M. and Tor A. farli. 1999. The Syntactic Disguises of the Predication

    Operator. Studia Linguistica53: 155-181.

    Faarlund, Jan Terje. 2008. A mentalist interpretation of grammaticalization theory.

    Grammatical Change and Linguistic Theory: the Rosendal Papers, ed. by Thorhallur

    Eythrsson, 221-244. Amsterdam: John Benjamins.

    Frajzyngier, Zygmunt. 1986. From preposition to copula.Proceedings of the Twelfth Annual

    Meeting of the Berkeley Linguistics Society, Vassiliki Nikiforidou, Mary VanClay,

    Mary Niepokui and Deborah Feder (eds.), 371-386. Berkeley: Berkeley Linguistics

    Society.

    van Gelderen, Elly. 2004a. Economy, Innovation and Prescriptivism: From Spec to Head and

    Head to Head.Journal of Comparative Germanic Linguistics7: 59-98.

    van Gelderen, Elly. 2004b. Grammaticalization as Economy. Amsterdam: John Benjamins.

    van Gelderen, Elly. 2007a. Linguistic Cycles. Ms., Arizona State University.

    van Gelderen, Elly. 2007b. The Definiteness Cycle in Germanic.Journal of Germanic

    Linguistics19: 275-308.

    van Gelderen, Elly. 2008a. Linguistic cycles and Economy Principle: The role of Universal

    Grammar in language change. Grammatical Change and Linguistic Theory. The

    Rosendal Papers, Thrhallur Eythrsson (ed.), 245-264. Amsterdam: John Benjamins.

    van Gelderen, Elly. 2008b. Where did Late Merge go? Grammaticalization as Feature

    Economy. Studia Linguistica62: 287-300.

    Gildea, Spike. 1993. The Development of Tense Markers from Demonstrative Pronouns in

    Panare (Cariban). Studies in Language17: 73-73.

    Heine, Bernd. 1993.Auxiliaries: Cognitive Forces and Grammaticalization. New York:

    Oxford University Press.Heine, Bernd, Ulrike Claudi and Friederike Hnnemeyer. 1993. Grammaticalization: a

    conceptual framework. Chicago: University of Chicago Press.

    Heine, Bernd and Tania Kuteva. 2002. World Lexicon of Grammaticalization. Cambridge:

    Cambridge University Press.

    Heine, Bernd and Tania Kuteva. 2007. The genesis of grammar: a reconstruction. Oxford:

    Oxford University Press.

    Hengeveld, Kees. 1992.Non-verbal Predication: Theory, Typology, Diachrony. Berlin:Mouton de Gruyter.

  • 8/13/2019 Copula Cycle

    41/43

    41

    Hopper, Paul J. and Elisabeth Closs Traugott. 1993. Grammaticalization. Cambridge:

    Cambridge University Press.

    Jansky, Herbert. 1954.Lehrbuch der Trkishen Sprache. Wiesbaden: Harrassowitz.

    Katz, Aya. 1996. Cyclical Grammaticalization and the Cognitive Link Between Pronoun and

    Copula. Doctoral dissertation, Rice University.

    Kornfilt, Jaklin. 1997. Turkish. London: Routledge.

    Kroch, Anthony. 1989. Reflexes of grammar in patterns of language change.Journal of

    Language Variation and Change 13: 199-244.

    Kuteva, Tania. 2001.Auxiliation: An Enquiry into the Nature of Grammaticalization. Oxford:

    Oxford University Press.

    Lehmann, Christian. 1985. Grammaticalization: synchronic variation and diachronic change.

    Lingua e Stile20: 303-318.

    Li, Charles and Sandra A. Thompson. 1977. A mechanism for the development of copula

    morphemes.Mechanisms of Syntactic Change, Charles N. Li (ed.), 419-444. Austin:

    University of Texas Press.

    Lightfoot, David. 1974. The diachronic analysis of English modals.Historical linguistics, J.

    M. Anderson & C. M. Jones (eds.), 219-249. Amsterdam: North Holland.

    Lightfoot, David. 1979.Principles of Diachronic Change. Cambridge: Cambridge University

    Press.

    Lightfoot, David. 1999. The Development of Language. Acquisition, Change and Evolution.

    Malden: Blackwell Publishers.

    Lightfoot, David. 2006.How New Languages Emerge. Cambridge: Cambridge University

    Press.

    Lohndal, Terje, Mari Nygrd and Tor A. farli. 2008. The structure of copular clauses in

    Norwegian. Working Papers in Scandinavian Syntax82: 23-41.

    Longobardi, Guiseppe. 2001. Formal syntax, diachronic minimalism, and etymology: thehistory of French chez.Linguistic Inquiry32: 275-302.

    Macaulay, Monica. 1996.A Grammar of Chalcatongo Mixtec. Berkeley: University of

    California Press.

    McWhorter, John H. 1995. Looking into the Void: Zero Copula in the Creole Mesolect.

    American Speech70: 339-360.

    McWhorter, John H. 1997. Lost in transmission: A case for the independent emergence of

    the copula in Atlantic creoles. The Structure and Status of Pidgins and Creoles, ArthurK. Spears and Donald Winford (eds.), 241-21. Amsterdam: John Benjamins.

  • 8/13/2019 Copula Cycle

    42/43

    42

    Migge, Bettina. 2002. The origin of the copulas (d/n)aand dein the Eastern Maroon Creole.

    Diachronica19: 81-133.

    Mikkelsen, Line. 2005. Copular Clauses. Amsterdam: John Benjamins.

    Munro, Pamela. 1977. From existential to copula: the history of Yuman BE.Mechanisms of

    Syntactic Change, Charles N. Li (ed.), 445-490. Austin: University of Texas Press.

    Noonan, Michael. 1992.A Grammar of Lango. Berlin: Mouton de Gruyter.

    Noonan, Michael and Karen Grunow-Hrsta. 2002. Posture-verbs in two Tibeto-Burman

    languages of Nepal. The Linguistics of Sitting, Standing and Lying, John Newman

    (ed.), 79-101. Amsterdam: John Benjamins.

    Pesetsky, David and Esther Torrego. 2001. T-to-C Movement: Causes and Consequences.Ken

    Hale: A Life in Language, Michael Kenstowicz (ed.), 355-426. Cambridge, MA:

    The MIT Press.

    Pesetsky, David and Esther Torrego. 2004. Tense, Case, and the Nature of Syntactic

    Categories. The Syntax of Time, Jacqueline Guron and Jacqueline Lecarme (ed.), 495-

    537. Cambridge, MA: The MIT Press.

    Plank, Frans. 1984. The modals story retold. Studies in Language8: 305-364.

    Pustet, Regina. 2003. Copulas: Universals in the Categorization of the Lexicon. Oxford:

    Oxford University Press.

    Radford, Andrew. 2000. Children in search of perfection: Towards a minimalist model of

    acquisition.Essex Research Reports in Linguistics34.

    Richards, Marc D. 2004. Object Shift and Scrambling in North and West Germanic: A Case

    Study in Symmetrical Syntax. Doctoral dissertation, University of Cambridge.

    Richards, Marc and Theresa Biberauer. 2005. ExplainingExpl. The Function of Function

    Words and Functional Categories, Marcel den Dikken and Christina M. Tortora

    (eds.), 115-153. Amsterdam: John Benjamins.

    Roberts, Ian. 1985. Agreement patterns and the development of the English modal auxiliaries.Natural Language and Linguistic Theory3: 21-58.

    Roberts, Ian. 1993. Verbs and Diachronic Syntax. Dordrecht: Kluwer.

    Roberts, Ian and Anna Roussou. 1999. A formal approach to grammaticalization.

    Linguistics37: 1011-1041.

    Roberts, Ian and Anna Roussou. 2003. Syntactic change: a minimalist approach to

    grammaticalization. Cambridge: Cambridge University Press.

    Rothstein, Susan. 1999. Fine-grained structure in the eventuality domain: the semantics ofpredicative adjective phrases and be.Natural Language Semantics7: 347-420.

  • 8/13/2019 Copula Cycle

    43/43

    Rothstein, Susan. 2001.Predicates and their subjects. Dordrecht: Kluwer.

    Samarin, William J. 1966. The Gbeya Language. Grammar, Texts, and Vocabularies.

    Berkeley & Los Angeles: University of California Press.

    Stassen, Leon. 1997.Intransitive Predication. Oxford: Oxford University Press.

    Traugott, Elisabeth C. 1972.A History of English Syntax: A Transformational Approach to

    the History of English Sentence Structure. New York: Holt, Rinehart & Winston.

    Tucker, A. N.1993.A Grammar of Kenya Luo (Dholuo). Kln: Kppe.

    Turunen, Rigina. 2006. Complex Morphosyntactic Features of Nominal Predicates in Erzya.

    SKY Journal of Linguistics19: 173-187.

    Warner, Anthony R. 1993.English Auxiliaries: Structure and History. Cambridge:

    Cambridge University Press.

    Westergaard, Marit R. 2008. Acquisition and Change: On the Robustness of the Triggering

    Experience for Word Order Cues.Lingua118: 1841-1863.