Syntactic Alternations in Englishv1

Embed Size (px)

Citation preview

  • 7/23/2019 Syntactic Alternations in Englishv1

    1/149

    Chapter 1..........................................................................................................................................40.1. Introduction...............................................................................................................................41.0. Background and Preliminaries. Theta Relations in the Lexicon and in Sntax........................!

    1.1. Lexical entries.......................................................................................................................!1.". #rgument structure................................................................................................................!

    1.$. %&'ar Theor.........................................................................................................................(1.4. )xternal argument *s. Internal argument............................................................................1"1.4.1. The Thematic +imension.............................................................................................1$1.4.". The #spectual +imension.............................................................................................14

    1.!. Lexical Sntax....................................................................................................................."0".0. The Lexical #rgument Structure o, Transiti*e -er's.............................................................."$$.0. n +e,ining the /otions o, Con,lation and Incorporation....................................................."4.0. # /ote on Con,lation o, Prepositions......................................................................................$$!.0. Intransiti*e -er's.....................................................................................................................$

    !.1. The Lexical #rgument Structure o, 2nergati*e -er's........................................................$!.". n Cognate '3ects.............................................................................................................4"

    .0. The Lexical #rgument Structure o, 2naccusati*e -er's........................................................44.1. 5ore on the Lexical #rgument Structure o, 2naccusati*es and the Lack o, #ccusati*eCase #ssignment........................................................................................................................4.". 2naccusati*it explained in terms o, theta&role ,eatures 6Reinhart7 "0008.........................!$

    .0. The Lexical #rgument Structure o, )rgati*e -er's................................................................!.1. +e&ad3ecti*al *er's..............................................................................................................!(

    (.0. The Lexical #rgument Structure o, Location and Locatum -er's and Their Properties....... .(.1. Patient&5anner -er's and #gent&5anner -er's.................................................................0(.". 5anner o, 5otion -er's.....................................................................................................$

    Chapter "........................................................................................................................................!0.1. Introduction.............................................................................................................................!

    1.0. The +omain o, the +ati*e #lternation...................................................................................(".0. The +omain o, the Bene,acti*e #lternation...........................................................................(1$.0. Semantic properties o, indirect o'3ects in the dou'le o'3ect construction and theprepositional o'3ect construction....................................................................................................($

    $.1. #nimateness o, the I........................................................................................................($$.". The #,,ectedness Constraint...............................................................................................(4$.$. The )xistence Presupposition.............................................................................................($.4. Su'3ect as Cause9#gent......................................................................................................($.!. 5ore on the I as delimiter and other delimiting expressions. 5ore on the + asmeasurer o, the e*ent. Preliminar sntactic conse:uences......................................................($.. 2T#;7 linking issues and possi'le sntactic structures o, the t

  • 7/23/2019 Syntactic Alternations in Englishv1

    2/149

    4.1. C&command asmmetries in the dou'le o'3ect construction.............................................10$4.". Case&assignment7 passi*i=ations and nominali=ations in the t 61((8 analsis o, the prepositional dati*e construction and the dou'le o'3ectconstruction..................................................................................................................................110.1. Pesetsk>s 61!8 analsis o, the prepositional dati*e construction and the dou'le o'3ect

    construction..................................................................................................................................11(.1. ;ale and ?eser>s 6"00$8 analsis o, the t mo*ement o, the I in the dou'le o'3ect construction and its Scope@ree=ing e,,ects. 5orphological properties o, the t mo*ements o, the 'ona ,ide +........................................................................1$(.". )monds and stler>s 6"00!8 account o, #>&mo*ement restrictions on the promoted I. .1$(.$. ur account o, #> mo*ement restrictions on the promoted I. Ailliams> 618 BlockingPrinciple and its e,ects..............................................................................................................140

    Re,erences....................................................................................................................................144

    FOREWORD

    This book has grown out of the material gathered for an elective course taught to the fourth

    students of English at the University of Craiova, University of Bucharest and Spiru aret!

    University in Bucharest since "##$%

    Chapter & addresses the thorny problem of syntactic alternations in English focusing on the

    argument structure and its changes of transitive, unergative and ergative verbs in the compelling

    le'ical(synta' framework proposed by ale and )eyse!s series of breakthrough articles published

    in &**$, &**+, &***, "### and "##$%

    Chapter " discusses the syntactic alternations given rise to by detransitive verbs in the same

    syntactic framework ale and )eyser, "##$- as well as in two other syntactic frameworks

    .esetsky, &**/ and 0illiams, &**1-%

    Syntactic 2lternations in English! targets mainly 32 students and students who prepare their

    graduation papers% The reader is encouraged and trained to e'plore the reasons why an analysis

    is to be preferred to another and to uncover the rationale that motivates the e'planatory

    superiority of a syntactic analysis over another%

    4 do hope the book will help my students become familiar with the methodological procedure of

    sentiently learning to argue in favour of or against a certain syntactic hypothesis% 4 thank them for

    their supportive feedback and intellectual curiosity they manifested when attending the course%

    3y special thanks go to my former teachers and good friends .ersi Baciu, .usi Cornilescu and

    5omnica 6erban for having read the manuscript at various stages of elaboration, for their

    academic advice and observations and for their having taught me to find delight in what 4 do%

    "

  • 7/23/2019 Syntactic Alternations in Englishv1

    3/149

    SYNTACTIC ALTERNATIONS IN ENGLISH

    $

  • 7/23/2019 Syntactic Alternations in Englishv1

    4/149

    CANONICAL VS. DERIVED STRUCTURES

    Chapter 1

    0.1. Introd!t"on

    4n this chapter we discuss several alternations in the synta' of English verbs, with specialreference to alternations in transitivity, which are not reflected by corresponding alternations inmorphological form%By syntactic alternations we understand alternations such as the following7

    &- a% 8ohn broke the vaseb% The vase broke

    "- a% e narrowed his eyes and frowned

    b.is eyes narrowed

    $- a% The cold turned the leaves redb% The leaves turned red

    The problem raised by these alternations concerns the important issue of what makes it possible

    for the same le'ical verb(entry to reali9e in so different syntactic configurations% 4n addressing

    le'ical properties, linguists have focused their attention on two distinct :uestions%

    ;ne is the :uestion of mapping (linking)of the thematic specification of the le'ical verb(entry to

    syntactic structure, i%e% which theta role should reali9e in which argument position% That is the

    concern of principles like Chomsky!s &*+&- Theta Criterion!, Baker!s &*++- Uniform Theta

    2ssignment ypothesis! UT2- or

  • 7/23/2019 Syntactic Alternations in Englishv1

    5/149

    prominence in argument structure mapping to syntactic structure% 2ll these principles are revised

    below% Basically, such principles are, first of all, set to determine the number of arguments that

    need to be selected from the le'icon for each selection of the verb and, ne't, mapping=linking

    principles guide the order of merging inserting- these arguments into the derivation%

    The other :uestion concerns the structure of the lexicon itself:do the verbs that appear to have

    different thematic structures correspond to one or more le'ical entries> Until recently, the idea

    that the two syntactic reali9ations of the verbs above correspond to two distinct le'ical verbs,

    listed as independent entries, has been :uite pervasive% ?evertheless, studies on the le'icon

    have shown that there are generali9ations that relate apparently distinct items and these

    generali9ations cannot be @ust an accident%

    Aollowing the system proposed by ale and )eyser &**$, &**+, &***, "###-, the purpose of

    the present analysis is to identify the formal conditions that trigger syntactic alternations mainly

    based on the operations of conflationand incorporation!-% To do this we begin with an aside7 a

    discussion of the notion of le'ical entry, of the notion of le'ical argument structure of predicates

    and the latter!s syntactic pro@ection in the elementary terms of (bar theory%

    1.0. #a!$%rond and &re'"("nar"e). Theta Re'at"on) "n the Le*"!on and "n S+nta*.

    1.1. Le*"!a' entr"e)

    2 le'ical entry e'presses the linguistic knowledge that a speaker of a given language possessesin relation to it% e'ical items include a collection of properties% Take, for instance, the le'ical

    items book and run% Some of these properties cover aspects of phonology and morphology

    which are language specific, and thus idiosyncratic- and a representation of le'ical items!

    semantic properties% e'ical entries also list their properties with respect to the categorial

    features they have7 for instance, the fact that the le'ical item bookhas the categorial feature D?

    and that the le'ical item runhas the categorial feature DF%

    1.,. Ar%(ent )tr!tre

    2 sentence always contains a predicate an e'pression denoting an activity or an event- and at

    least one argument an e'pression denoting the participant in the activity or the event-% Aor

    e'ample, in sentences such as the following7

    G- a% DThe dog died

    !

  • 7/23/2019 Syntactic Alternations in Englishv1

    6/149

    b% DEverybody laughedc% DThe policeman arrestedDthe suspectd% D8ohn fearsDghosts

    the italici9ed verbs are predicates and the bracketed e'pressions are their arguments%

    Specifically, all sentences in G- are propositions they describe the semantic content of the

    clause-% Aor instance, the proposition in Gc- consists of the predicate arrest and its two

    arguments, the policemanand thesuspects% The two arguments represent the two participants in

    the act of arresting and the predicate is the e'pression that describes the activity in which they

    are engaged% The description of the set of arguments associated with a predicate provides the

    argument structureof that predicate%

    owever, saying that a verb like arrest! takes two arguments fails to account for the important

    fact that the two arguments play different semanticroles in relation to the act of arrest% 4t fails to

    account for the fact that the policeman is the person who performs the act and that the suspect is

    the person who suffers the conse:uences of the act% 2rrest! is different from fear!, still a

    transitive verb, whose arguments bear the semantic roles of E'periencer and Theme% ence,

    any ade:uate account of argument structure should provide a description of the semantic roles

    that each argument plays with respect to the predicate%

    The argument ?.s are associated with the predicate in a particular conceptual relationship and

    they bear particular semanticor thematic rolessuch asAgent, Instrument, Patient, Locationetc%-

    that depend on the meaning of the verb% Semantic roles or theta(roles are the

    semantic=conceptual facet of the structure of events function of their participants% The

    combination of thematic roles that may be associated with a given predicate form the thematic

    structure or the theta-gridor the argument structureof a predicate% Theta Theorythe theory of

    thematic relations- is concerned with the description of the le'ical structure of a predicate,

    function of the semantic interpretation of its argument ?.s% Theta(roles are generally referred to

    by labels%

    Het, there is no agreement about how many such specific thematic roles there are and what their

    labels are% Some types are, however, generally distinguished7

    /- a% 2

  • 7/23/2019 Syntactic Alternations in Englishv1

    7/149

    by the predicate%

    /- e% BE?EAECT4FE=BE?EA4C42IH7 the entity that benefits from the action e'pressed

    by the predicate%

    /- f%

  • 7/23/2019 Syntactic Alternations in Englishv1

    8/149

    fact, they ac:uire substance only in relation to the predicate that re:uires them, as they depend

    on the meaning of the respective verb% 2rguments are identified according to the semantic,

    rather than the syntactic relation they bear to the predicate%

    4n the le'icon, erbs are viewed as more elementary predicates linked by various relations%

    KThey Dthe verbs are composed of a number of primitive elements that recur in the definitions of

    many verbs% Similarities in the meanings of verbs can be captured by attributing shared

    elements to their decompositions% Ferbs fall into classes and are therefore e'pected to have

    shared properties by virtue of these common elementsK Iappaport and evin &*++--% The

    assumption behind Iappaport and evin!s &*++- predicate decomposition is that at some level

    of representation the meaning of verbs has internal structure%

    4n order to make e'plicit the aspects of meaning relevant to predicate decomposition, Iappaport

    and evin &*++- employ the notion of lexical conceptual structures CSs-% These structures

    provide specifications of the verbLs meaning and the arguments of the verb are indicated in the

    representations as variables% L!"s should be ie#ed as the lexical part of the erb$s meaning%

    Any changes in argument structure of a erb #ill engendersemantic changes at the leel of

    L!"% The verb itself decomposes in more elementary predicates linked by various relations%

    These elementary predicates, such as BEC;3E, C2USE, BE, may occur over a large number

    of verbs that are semantically related% The argument places i%e% the participants in the event- of

    these predicative constituents are held by variables i%e% instead of ?.s- and are no longer

    identified by theta(role labels% The arguments are thus variables occurring in substructures of

    CSs% ere are a few e'amples7

    1- a% .UT7 D' does smth cause Dy come Dto be at 9' does something which causes y to come to be at 9-

    b% )47 D' does smth cause Dy come Dto be not alive

    The variables in CSs are ultimately mapped onto positions in synta'% ?otions like 2gent,

    Theme, ocation etc% are not primitives but are defined in terms of positions #ithin L!"s% Aor

    e'ample, LThemeL the variable y- is defined in a- as an entity that comes to be at a ;C2T4;?

    9- or in b- as an entity that comes to be in a ST2TE% 4n the same way, the sub@ect of the verb5; in 1a, b- is either an 2gent or an 4nstrument and the specification as to DM uman will

    distinguish between these two roles e%g% %ohn killed &ary= The bullet killed %ohn-% Thus, theta(

    roles are definednotions7 they are @ust convenient descriptive labels that have no theoretical

    status or significance% owever, we shall continue to identify variables by theta(role labels for

    convenience, but it should be kept in mind that they are means of referring to variables%

    (

  • 7/23/2019 Syntactic Alternations in Englishv1

    9/149

    4n grammar, we speak of lexical categories i%e% the parts of speech7 noun, verb, ad@ective,

    preposition- and syntactic categoriesi%e% ?., F., etc%-% ere we briefly define only the nominal

    and the verbal parts of speech%

    2 part of speech is nominalif it can be the obligatory constituent, that is the argumenti%e% a

    participant in an event- of a predicate% 2 nominal part of speech can be marked for gender,

    number and case% Aor e'ample, ?. is a nominal category in English as it is marked for gender,

    number and case7 e%g% loe himi%e% ?.7 masculine, singular, accusative case-%

    2 part of speech is erbal if it has the ability to license an argument% icensing an argument

    means assigning it a semantic interpretation as already e'plained above% Aor e'ample, kill is a

    verbal category because it licenses two arguments7 an 2gent who does the killing- and a .atient

    who is the sufferer of killing- e%g% 'rutus killed !aesar-N gie is a verbal category because it

    licenses three arguments an 2gent, a Theme Othe thing given- and a

  • 7/23/2019 Syntactic Alternations in Englishv1

    10/149

    0ith

  • 7/23/2019 Syntactic Alternations in Englishv1

    11/149

    0e notice that all branching nodes are binary branching% Such a structure is more constrained it

    does not allow ternary branching- and it has been proved that in terms of language ac:uisition

    fewer choices = branches automatically mean more speed in the construction of the core

    grammar of the language ac:uired%

    Spec is an optional constituent and is normally involved in spelling out the reference of the head

    i%e% in the sense that it makes it more e'plicit-% Aor e'ample ery interested in architecture

    where the heads are Linterested!, in! and architecture!- pro@ects as in &$-7

    &$- 2.

    Spec2. 2!

    5eg. 2# ..

    very interested .!

    .# ?.

    in ?!

    ?#

    architecture

    0e notice that the tree above gives us the .hrase Structure template of phrases as cognitive

    units% Thus, a phrase in

  • 7/23/2019 Syntactic Alternations in Englishv1

    12/149

    the !ha"n '"n$relation as in the e'ample below7

    &/- 8ohniwas Darrested ti

    4n the present analysis, we adopt a more recent variant of (bar structure, namely the #are &hra)e

    Str!tremodel put forth by Chomsky &**1, The 3inimalist .rogram- as it is adopted by ale

    and )eyser &**+, &***, "###- on whose work our approach is heavily based% The new format of

    (bar structure looks as in &J- below7

    &J- 2 B

    2 category that does not pro@ect any further is a ma'imal pro@ection i%e%, the upper -, and one

    that is not a pro@ection at all is a minimal pro@ection the head -% The intermediary i%e%, the one

    that dominates D B- does not count as it is invisible at the interface sound(meaning% 4n such a

    model, there are no bar levels and no distinction between le'ical items and heads! pro@ected from

    them, i%e%, an item can be both an #or an . Chomsky &**1-% ?ote, however, that in such a

    configuration, categories are still defined on the basis of the properties of le'ical items% Thus a

    complement B- is the uni:ue sister of the head -% 2 specifier 2- is the uni:ue sister of the first

    branching pro@ection of the head -N the intermediary pro@ection - dominating D B is the

    head!s first branching pro@ection%

    2nother relational property of the structure in &J- is that it conforms to )ayne!s principle of

    *nambiguous Proection &*+G- which states that branching is necessarily binary% 2 head

    determines an unambiguous pro@ection of its phrasal category and an unambiguous

    arrangement of its arguments Spec and complement-% The structure in &J- is unambiguous

    because the sister relation holds unambiguously between the head and its complement B and

    between Spec 2 and , the node that dominates D B% The c(command relation is likewise

    unambiguous7 the specifier 2 asymmetrically c(commands the complement B%

    The notion of c(command was introduced in linguistic vocabulary by Ieinhart &*1J- in a study

    on anaphoric relations i%e% relations between various types of pronouns and their antecedents-%

    C(command e'presses the structural relationship between two nodes and which are in the

    same constituent and where is higher in the tree than or at least not lower than 7

    C/!o((and Ieinhart! definition-

    1"

  • 7/23/2019 Syntactic Alternations in Englishv1

    13/149

    c(commands iff

    every branching node dominating also dominates

    &1-

    2 -

    B -

    4n &J- and &1-, 2 - unambiguously c(commands B - since the first branching node over 2

    also dominates B%

    C(command is an asymmetrical relation in which one term is higher and dominated by the

    relevant first branching or ma'imal pro@ection node, while the second term can be situated far

    down on the tree%

    Theta(Theory also interfaces with the structural notion of government%

  • 7/23/2019 Syntactic Alternations in Englishv1

    14/149

    with an argument position of the verb-% 0hen all the positions in the (grid of a verb have been

    discharged, the verbLs (grid has been saturated% Theta(marking observes the following

    configuration re:uirement7

    &*- Theta(marking should take place under government by the theta(marking head,therefore, within some pro@ection of the head%

    4n an ordinary transitive configuration such as %ohn kissed &ary, the 5; is directly theta(marked

    and theta(governed- by F#, so condition &*- is met%

    1.2. E*terna' ar%(ent 3). Interna' ar%(ent

    2rgument structure interfaces with two other kinds of representations

  • 7/23/2019 Syntactic Alternations in Englishv1

    15/149

    The central roles in the hierarchy have all to do with the idea of location, the E'periencer

    included7 the E'periencer is a Kpersonal locationK, the locus of a psychological process cf% 2%

    Cornilescu, &**/-% The higher position of the E'periencer in the hierarchy as compared to the

    other locative thematic roles is due to its always being DM.ersonal% The last position in the

    hierarchy occupied by the Theme is highly debatable% There are linguists Baker &*+*-, arson

    &*++-- who represent the Theme as higher than the

  • 7/23/2019 Syntactic Alternations in Englishv1

    16/149

    ""- frighten ' y-- E'p Theme Thunder Theme, sub@ect- frightens them E'p, ob@ect-% er behaviour appalled him%

    This shows that more than the thematic hierarchy is at stake in deciding syntactic relations%

    There must be some other principle that supplements the thematic hierarchy% 4n

  • 7/23/2019 Syntactic Alternations in Englishv1

    17/149

    arrie- or protracted=durative events also called accomplishments makea chair, build a house,

    run a mile, dra# a picture, dig a hole, kill, paint a portrait)

    These verb(types differ in their conceptual comple'ity which can be described in terms of

    semantic components in the CS of a predicateN their semantic comple'ity is based on their

    temporal structures set on perceptual and cognitive grounds7 whether or not they contain stages

    in their development in time, whether or not they have natural or arbitrary endpoints, whether or

    not they are agentive=non(agentive%

    "tates

    States have been called staticor non-dynamicsituations which contrast with non(states which

    are dynamic Bergman &*+$-, .inkster &*+G--% States do not change throughout their duration,

    that is, all stages or phases of a state are the same%

    Aor e'ample if Tom is ill from Tuesday through Ariday, then at any time during that interval Tom

    is illis true and likewise over any subinterval during that interval%

    Thematically, states are compatible with Themes and ocation as in e%g% The roomocation-

    reeks of tobaccoTheme-, E'periencers e%g% the sub@ects of loe, hate, etc%- but not with 2gents

    because states are not controllableN notice their non(occurrence in imperatives e%g% Q/no# the

    ans#er- or with agentive adverbs e%g% Q%ohn #illingly . deliberately kno#sthe ans#er-%

    Actiities or Processes

    The term process is viewed to uncover both activities, associated with human sub@ects he ran,

    she read for an hour- and activities which are not cases of human agency the ball moed, it

    sno#ed for 0 hours-%

    Unlike state predications, process and event- predications are dynamic% 5ynamism refers to

    these predications! property to change over time% Clearly, activities occur with the thematic role

    2gent as the typical sub@ect of the operator 5; is the 2gent Ioss &*1"--% Ioss also suggested

    that KAillmoreLs notion of 2gent might be replaced by the notion Kpossible sub@ect of 5;K% 4n fact,

    the sub@ect of activities may be any thematic role that can be viewed as a causal factor e%g%

    2gent, 4nstrument, Source, Cause-%

    +ents

    1

  • 7/23/2019 Syntactic Alternations in Englishv1

    18/149

    Events are divided up into accomplishments #rite a letter, run a #hile- and achievements

    recognise, find a penny- on the basis of whether they contain an activity phase or not%

    2ccomplishments are made up of an activity phase and an achievement phase which is a point7

    if he #rote a letterthen, we can truthfully say that he finished (#riting) the letter%

    2n achievement is all culmination% 2lthough an achievement is possibly preceded by some

    activity, the verb refers only to the achievement phase, not to the preceding activity% Aor

    e'ample, when achievements have a tentative sense7 it took an hour for her to find the bookthis

    is rather different from a similar accomplishment7 it took an hour for her to dra# a picture% The

    act, which results in finding, is searching not finding, but that which results in having drawn a

    picture isdrawing a picture%

    Both accomplishments and achievements involve a product, upshot or outcome #rite a letter,

    dra# a circle= #in the race, find a penny-%

    Unlike processes, events also involve definite change of states% 2n event such as the ball

    reached the bottom of the slope is true when the ball changes location to a well defined place

    the bottom of the slope- and thus, is true of a single, uni:ue interval but not of any subinterval

    or superintervals of the interval%

    2nother way of characteri9ing events is as either duratie ( accomplishments i%e% they take

    place over an interval of time- orpunctual( achievements i%e% they occur at a point in time-% 2

    further distinction between achievements and accomplishments is that the former involve single

    change of state achievements- while the latter involve comple' change of states

    accomplishments- 5owty &*1*--%

    Using the method of le'ical decomposition in a model ( theoretical framework, 5owty &*1*-

    analyses achievements such as7 notice, realie, ignite non(agentive- and kill, point out

    (something to somebody)agentive- as single change of states because they have a semantic

    structure that consists of the operator BEC;3E plus the embedded clause which is a stative

    predication-%

    They are instantaneous events7 BEC;3E is true at t&iff is true at t&and false at ti(&7

    8ohn lost the book BEC;3E 8ohn does not know the location of the book- The soup cooled BEC;3E the soup is cool-

    2ccomplishments such as flo# from x to y, dissole non(agentive-, build a house, #alk a #hile

    agentive- are comple' change of states because they conceptuali9e not only the change of

    state but also the causing factor% 2s such, their semantic structure CS- involves two operators7

    1(

  • 7/23/2019 Syntactic Alternations in Englishv1

    19/149

    C2USE and BEC;3E plus a stative embedded clause7 C2USE DBEC;3E 7

    8ohn painted a picture% D8ohn paints C2USE DBEC;3E Da picture e'ists% 8ohn cooled the soup% D8ohn does something C2USE D BEC;3E Dthe soup is cool%

    0e conclude that the analysis of durative events in terms of the bi(sentential predicate C2USE

    a causing activity with a causing agent and a result state- makes up the CS of these

    predicates%

    0e have seen that a prototypical event has a comple' structure made up of two aspectual

    subparts7 it consists of an activity part and a resulting state% This event structure of the entire

    sentence sets up another hierarchy, namely, an aspectual hierarchyof the thematic roles% The

    aspectual hierarchy interacts with the thematic hierarchy of the predicate in the sentence%

  • 7/23/2019 Syntactic Alternations in Englishv1

    20/149

    2gent E'periencer

  • 7/23/2019 Syntactic Alternations in Englishv1

    21/149

    higher on the dimension, are theta marked and pro@ect last, outside the compound% The same

    holds for psychological verbs such as fearand frightenand their verb(compounds7

    "J- a% 3an fears

  • 7/23/2019 Syntactic Alternations in Englishv1

    22/149

    The other arguments of the verb are reali9ed inside the pro@ection of the verb and they are

    referred to as internal?. arguments% Aurther on, if a ?. argument is assigned its theta(role

    directly by the verb, that ?. will be the direct?. argument of the verb% 2ny ?. whose theta(role

    is assigned by a preposition is referred to as indirect ?. argument 3arant9 &*+G--%

    &%/% 2s with Iappaport and evin &*++- and

  • 7/23/2019 Syntactic Alternations in Englishv1

    23/149

    The notion of the .ro@ection .rinciple includes not only the subcategorisation properties of

    le'ical items i%e%, the range of complements a given item takes as its subcategorisation

    information- but also their thematic semantic- properties% Thus, a more ade:uate formulation of

    this principle should be restated as7

    .ro@ection .rinciple generalised-

    Syntactic representations must be pro@ected from the e'icon, in that they observe the le'icalproperties of the items they contain%

    4t has been assumed that the term lexical properties3subsumes all relevant conte'tual properties,

    including subcategorisation and thematic properties cf% Iadford &*++-% The .ro@ection .rinciple

    would then tell us, for e'ample, that a verb such as murder, which is specified in the e'icon as

    permitting only an 2gent sub@ect, cannot be inserted into a structure in which it has a Theme or a

  • 7/23/2019 Syntactic Alternations in Englishv1

    24/149

    a matter of fact, the former derivation is more economical it contains only one step- while the

    latter is less economical it contains two steps-% owever the theory of economy says that

    shorter derivations block longer ones% There is another, deeper reason why %ohncannot be

    directly inserted in DSpec,4.N this reason lies in Theta Theory% ?otice that if %ohnwere directly

    inserted in DSpec,4., this argument would lack a theta role and the verb likewould fail to assign

    its e'ternal theta role% Such a derivation is uninterpretable and crashes% 2n argument without a

    theta role violates both the Theta !riterion and the principle of 4ull Interpretation Chomsky

    &*+J-, causing the derivation to crash% Theta !riterionsays that every argument must be theta

    marked and The .rinciple of 4ull Interpretation states that elements e%g% arguments- that

    contribute to the semantic interpretation of a structure must be present in the syntactic

    representation of that structure-% Thus, the shortest derivation condition entails that a violation

    of the Theta Criterion causes the derivation to crash, by failure to satisfy Aull 4nterpretation%

    Theta theory i%e%, assignment of semantic roles- does not interact with the syntactic theory of

    movement as conceived in Chomsky!s &inimalist Program&**1, for instance% 3orphological

    features such as Case and agreement, which have to be checked in a syntactic checking

    domain, drive movement or displacement of elements%

    Sentential synta' operates 5EI4F2T4;?2H and involves successive operations that lead to

    convergence of the interface levels7 sound the articulatory(perceptual system, called

    Phonological 4orm- and meaning the conceptual(intentional system, called Logical 4orm-% The

    displacement property reflects the disparity R in fact complementarity R between morphology

    checking of features- and Theta theory Chomsky &**17"""-% 3ovement and theta theories

    mutually supply each other!s lack%

    4n this chapter, our purpose is twofold% Airst, we focus our attention on E4C2 SH?T2 l(

    synta'- i%e% on the argument reali9ation of verbs and their syntactic pro@ections- and try to

    e'plain when and under what formal conditions changes in the argument structure of a verb

    bring about syntactic alternations based on the operations of conflation! and incorporation!-%

    Second, we want to arrive at devising the lexical argument structureof transitive and intransitive

    verbs, i%e% the linking=merging process of their arguments when the representation of a predicate

    argument structure is conceived of as itself a synta'%ale and )eyser!s findings in l(synta' have paved the way to Chomsky!s new structures of F.s

    in the sentence architectureN these F. structures which we will briefly mention- are adopted in

    the 3inimalist .rogram &**1-%

    0e also mention the analysis of the argument structures of intransitive and transitive predicates

    in the old

  • 7/23/2019 Syntactic Alternations in Englishv1

    25/149

    understand the refinements and the e'planatory power of later approaches to le'ical argument

    structure%

    ,.0. The Le*"!a' Ar%(ent Str!tre o6 Tran)"t"3e Ver)

    Iemember that at the level of sentential(synta' s(synta'- transitive verbs are dyadic predicates

    that include an 2gent and one internal argument as a rule, bearing the role of Theme-% Since

    the 2gent en@oys the most prominent position on both the thematic dimension 2gent- and the

    aspectual dimension Cause-, it is always an e'ternal argument and will be pro@ected as 5(

    structure sub@ect

  • 7/23/2019 Syntactic Alternations in Englishv1

    26/149

    2gent

  • 7/23/2019 Syntactic Alternations in Englishv1

    27/149

    should be determined entirely by the relations it bears with the pro@ecting head F% Specifically,

    the 5. in :uestion bears the Spec relation within the matri' higher- F whose head takes a

    complement F that is also a pro@ection of the category F%

    The logical conse:uence of the syntactic and semantic facts above is that the Spec of transitive

    verbs cannot bear @ust any theta role but, unavoidably, that of AGENT an argument that

    designates the doer of the action event &- in :uestion-% 4mportantly, it also follows that theta(

    roles are not primitive notionsN they are instead derivative of le'ical syntactic relations and they

    are limited in this manner%

    4f the sub@ect of a transitive verb is an !e'ternal sub@ect!, then, how is it related to the pro@ecting

    verb the lower verb head-> 4n a sense, the answer comes straightforwardly7 the e'ternal

    argument is related to its verb through predication% The :uestion that arises is this7 of which out

    of the two heads F is the e'ternal sub@ect part of the le'ical argument structure> et us make the

    notion of predication more precise% The predication structure we have to consider here is the

    sister pair F(F"configuration%

    The e'ternal sub@ect is not present in the le'ical argument structure of the lower head F in the

    sense in which the ob@ect TE3E is present in its le'ical argument structure% 4n sentential

    synta', Spec 5. raises to Spec 4. due to a general principle of grammar which re:uires that all

    sentences must have sub@ects the E*tended &ro5e!t"on &r"n!"p'e- in order to receive Case in

    accordance with English morphology% The syntactic structure assigned to the sentence will

    include the full range of functional categories and pro@ections involved in the formation of an

    interpretable sentence% ?ot all sub@ects are e'ternal! in this sense% 2nd, conse:uently, not all

    sub@ects are 2gents%

    et us now analyse the properties of the complementselected by the verb i%e%, the direct ob@ect

    of transitive verbs-%

    Transitive verbal e'pressions like those represented in $/-7

    $/- a% The cowboys make trouble b% 3ary baked a cake c% The bitch has puppies

    d% 0e built a house

    are erb-complement constructionswhose semantics involve creation! or production! evin

    &**$-% Their sub@ects, e'ternal arguments, are agentive! in conformity with the widely accepted

    hierarchy of thematic roles

  • 7/23/2019 Syntactic Alternations in Englishv1

    28/149

    Their complementstrouble, a cake, puppies, a house- are full 5.s that function as grammatical

    ob@ects in the basic sentential syntactic use of such verb phrases% That is, the argument

    structure configuration pro@ected by the head verbs above contains @ust one argument, i%e% an

    "nterna' ar%(ent, their complement% The complement relation is defined as the uni:ue sister

    to the head as in $J- below7

    $J- F

    F 5.

    make trouble

    4f this structure is embedded in the structure $G- above we obtain $1-, the full structure of a

    transitive verb7

    $1- F&

    5. F

    F F"

    F 5.

    Thus, the specific property of the transitive verbs that head the argument structure of sentences

    in $/- is that they take a complementas an internal argument the ob@ect 5.s in the e'amples

    cited- but the structure these heads pro@ect does not include a specifierof their own they only

    contain an e'ternal argument with the syntactic function of sub@ect-%

    2 complement position is created by the 7er%eoperation with a head i%e% the head and its

    complement form one constituent-% The merge operation, which leads to the formation of

    syntactic structures, shows that categories are combined in a pair(wise fashion to form larger

    categories to be defined in more technical terms below-% ;ne conse:uence of this is that

    phrases have an intrinsically binary structure, since the merge operation combines two

    categories together and never combines three or four at one go in virtue of the Unambiguous

    .ro@ection-%The above analysis of transitive verbs shows that they are comple', causative double(F.

    structures% The higher F. is interpreted by Chomsky The &inimalist Program &**1- as an

    abstract causative '"%ht 3erv( i%e% a null verb with much the same causative interpretation as

    a verb like makeso that 5e built a househas a similar interpretation to 5e made the house to

    come to be built-% This causative light verb is affi'al in natureN for e'ample the suffi' Ren in a

    "(

  • 7/23/2019 Syntactic Alternations in Englishv1

    29/149

    verb like saddenis an affi'al light verb that can combine with the ad@ective sadwhich raises

    and ad@oins to - to form the causative verb saddenmeaning make sad!, cause to become

    sad!-% 0hen hosts no affi' the lower head verb raises to ad@oin to it producing a structure that

    can be paraphrased literally as 5emade 6 build a house-% The e'ternal argument of cannot

    be lower than DSpec % 4f it is in DSpec , then, R F. configuration can be taken to e'press the

    agentie . causatierole of the e'ternal argument cf% Chomsky &**1-% Thus, the e'ternal=agent

    role is a property of the R F. configuration and a specifier bearing this role is therefore a

    necessary part of the configuration% The internal arguments of the verb occupy the positions of

    specifier and complement of F%

    4n the 3inimalist framework, the template of a transitive verb that can also take an 4; as an

    internal argument is7

    $+- 4!

    4# .

    Su !

    # F.

    4;- F!

    F# 5;

    The argument structure attributed to the mono(transitive verbs above can be contrasted with the

    configurations pro@ected by de/tran)"t"3e 3er) of the type put, get, pound, dripfollowed by

    prepositions as in7

    $*- a% 8ohn put the books on the shelfb% 3ary got the cows into the corral

    c% Christopher pounded nails into the walld% .eter dripped paint on the floor

    Aor the moment being we are concerned @ust with the structure headed by the prepositions on

    and into-% .repositions are prototypically inter(relational! in the sense that they specify an

    interrelation spatial, temporal or other- between two entities or two events or two

    circumstances-% .repositional heads have the inherent and fundamental le'ical property that

    they pro@ect a structure containing both a complement 5. the shelf, the corral, the #all, the

    floor- and a specifier 5. the books, the co#s, nails, paint-% 2s usual, the complement the

    shelf- is the uni:ue sister of the head while the specifier the books- is the uni:ue sister of the

    "

  • 7/23/2019 Syntactic Alternations in Englishv1

    30/149

    structure formed by the head and the complement7

    G#- .

    5. .

    the books . 5.

    on the shelf

    0e notice that it is the structural configuration defined by the head . which pro@ects t#o internal

    argument positions% The presence of a specifier argument with de(transitive verbs is the

    essential structural difference between the le'ical configuration of G#- and that of the mono(

    transitive verbs discussed above representation $J-%

    2s already said, while the mono(transitive verbs of $J-, sharing the structure of $1-, have a

    sub@ect, that sub@ect is an e'ternal argument, in the le'ical configuration% The evidence for this

    le'ical difference is straightforward% So, the structure presented in G#- can, in its entirety

    specifier, head, complement-, appear as a complement of a verbal head within a le'ical

    pro@ection% This property is evinced by a large number of transitive verbs of location! likeputand

    alike heput the books on the shelf, he fit the horse #ith shoes-7

    G&- F

    F .

    put 5. .

    the books . 5.

    on the shelf

    The argument structure of the le'ical item put is a comple' configuration consisting of a .(

    pro@ection embedded as a complement within a F(pro@ection% The specifier within the embedded

    .(pro@ection will appear as the grammatical ob@ect of the verb in sentential synta' i%e% it is c(

    commanded and governed byputwhich will assign it structural 2ccusative case in the active

    voice-% Crucially, the specifier of the embedded .(pro@ection in G&- is within the structuralconfiguration associated with the le'ical entry of the verb% e'ically, it is properly an internal

    argument% 4ts status is :uite different from the sub@ect argument of verbs like make, bakeetc in

    $1-% The sub@ects of these latter verbs are e'ternal arguments and at no point in the derivation

    do they occupy an internal position comparable to that occupied by the specifier the books in

    G&-%

    $0

  • 7/23/2019 Syntactic Alternations in Englishv1

    31/149

    4n the structure G&- there are two heads F and .% The semantic type of F is event, and that of .

    is inter(relationship% 4f these two basic semantic notions combine, they will do so unambiguously

    and the most salient meaning! attached to such a structure is change.That is, some entity

    represented by Spec 5. the sub@ect! of .- comes to be involved in an inter(relation with an

    entity corresponding to the 5. ob@ect the shelf- of .% Thus, the head . provides the place entity

    shelf- that must enter into inter(relationship with another entity% This other entity must be

    present in the syntactic structure as it is in the semantic component of the verb put% Therefore,

    the Spec 5. the books- is re:uired in F as a necessity in the le'ical syntactic pro@ection of F

    here% The internal sub@ect! of a predicate of change3is called, as already said, athemeor 8an

    affectedpatient9% 4t again follows that semantic roles are derivative of the le'ical semantic

    relations%

    -.0. On De6"n"n% the Not"on) o6 Con6'at"on and In!orporat"on

    Aollowing ale and )eyser "###- we operate a distinction between the process of

    CONFLATION Talmy &*+/- and the process of INCOR&ORATIONBaker &*++-% 0e shall try

    to show that, in English, de-nominal erbs such as laugh, corral, cale- are derived by the

    process of conflation while de-adectial erbssuch as narro#, thicken, clear- are derived by

    the process of incorporation% et us first define the two notions%

    ale and )eyser &***- make the following assumptions about argument structure7

    Argument structure is defined in reference to t#o possible relations bet#een a head and itsarguments, namely the head-complement relation and the head-specifier relation

    2s already shown, for a given construction such as G"- below, a complement B- is the uni:ue

    sister of the head -% 2 specifier 2- is the uni:ue sister of the first branching pro@ection of the

    head -N the intermediary pro@ection - dominating D B is the head!s first branching

    pro@ection%

    G"-

    2

    B

    2 given head may enter into a relation with either B or 2, with both or with neither of them%

    $1

  • 7/23/2019 Syntactic Alternations in Englishv1

    32/149

    *nergatie erbsof the type laughare derived by a process of CONFLATION, which involves a

    bare nominal root and a phonologically empty verb, as illustrated below7

    G$- F F

    F ? F ?

    laugh calve

    This process is morpho(phonologically motivated and is taken as occurring concomitant with the

    operation 7ERGE%

    The operation that takes a pair of syntactic ob@ects S; i, S;@- and replaces them by a new

    combined syntactic ob@ect S;i@is called 7er%e Chomsky &**1-%

    7er%e is an operation that forms larger units out of those already constructed% 2pplied to two

    ob@ects and , 3erge forms the new ob@ect ), eliminating the two items and % 3erge

    applies in the simplest possible form7 at the root% ) contains and which are its constituents%

    The newly formed syntactic ob@ect ) bears the label that identifies its type% 4n ), either or

    pro@ects and becomes the head of ), thus identifying it% 4f the head that pro@ects is nominal, the

    ma'imal pro@ection ) will be a nominal phraseN if the head that pro@ects is verbal, the ma'imal

    pro@ection ) will be a verbal phrase% Thus, the head that pro@ects becomes the label of the

    comple' formed% 3inimal and ma'imal pro@ections are not identified by any special markingN

    they are e'clusively identified by the structure in which they appear% 4t is the relational properties

    of categories, not properties inherent to them that identify these types of pro@ections Chomsky

    &**1-%

    4n our former e'ample that accounts for the derivation of the unergative verb laugh G#-, the

    nominal is the complement of the empty verb% The process of conflation fuses the two items into

    a single word% 2t conflation, the verb is no longer empty, as it bears the overt phonological

    matri' of the noun% ?otice also that in G$- it is the newly formed verb that is the head that

    pro@ects and, thus, ) i%e%, F- is a ma'imal verbal pro@ection as it does no longer pro@ect-%

    Aor conflation to take place, ? must be governed by the F to which it transfers its phonological

    matri' i%e%, they must be sisters at the base-% 4n conflation, the syntactic relation between the

    two heads involved is one of 7strict complementation3ale W )eyser "###-% 0e do not discuss

    the morphological process that is responsible for the changes in the form of the noun calfinto its

    conflated verbal form cale-%

    0ith respect to the tree diagrams above, we have to be aware that they are @ust abstract

    representations of the relations involved in the corresponding le'ical items where conflation is

    $"

  • 7/23/2019 Syntactic Alternations in Englishv1

    33/149

    involved% owever, there is no level of syntactic representation at which the above structures

    actually appear under this form% That follows from the assumption that conflation is a

    concomitant of 3erge R the conflated structure appears instead%

    2s the structure in G$- shows, the verb laugh is fundamentally transitive also sharing with

    canonical transitive verbs the property of not pro@ecting a specifier% The result of conflation is the

    single word laughthat functions in sentential synta' as a regular intransitie (unergatie) erb

    Conflation is also involved in the derivation of English 'o!at"on and 'o!at( verbs

    e'emplified in GG- and G/-, respectively7

    GG- bag, bottle, box, cage, can, corral, garage, ail, sheleG/- bandage, bar, bell, blindfold, clothe, curtain, fund, saddle

    Aor e'ample7

    GJ- a% 4 shelved the books b% She saddled the horse

    These verbs are )+nthet"! !onterpart)of the ana'+t"!put, fitwhose structure was shown in

    $+- above% The only difference between these verbs and verbs such as put is that their

    preposition is not overtly present and the upper head occupied overtly in a structure withput- is

    also empty% The structure of these synthetic verbs is shown in G1-7

    G1- F

    F .

    5. .

    the books . ?

    the horse shelf

    saddle

    The diagram above is a description of the properties of the head the empty .- which makes up

    the le'ical item% 2s in the regular way, the head . has the syntactic property that it takes a

    complement i%e%, shelf, saddle- and it pro@ects a specifier i%e%, the books, the horse-% But in

    these cases the head . is empty% Therefore, it must conflate with its complement% The upper

    verb F is also empty and, thus, it necessarily conflates with its complement . itself a product of

    conflation-% 4n this way, the verbal head ac:uires phonological substance, resulting in the verbs

    $$

  • 7/23/2019 Syntactic Alternations in Englishv1

    34/149

    sheleand saddle%

    ?otice that strict complementation! is again observed, and the relation between F and . is

    again local as F governs its complement .% 0e have to do with a head(head local relation, as

    the 5.s in the structure are Specs%

    The conclusion we draw is that de-nominal erb formation is the result of the operation

    Conflation that is concomitant with 3erge, and the condition of local head(head relation is met%

    5e(nominal verb formation is notsimply a process of category changeas demonstrated in what

    follows% 4f the strict head(head! condition is not observed, conflation gives ill(formed results%

    Thus, government is a necessary condition for conflation but not a sufficient one%

    et us e'emplify this by considering the following locatum and location verbs7

    G+- a% ucy corralled the calves cf%, put the calves in the corral-

    b% 3ary rosined the rope cf%, treated the rope with rosin-vs%

    G*- a% Qucy calved in the corral b% Q3ary roped with rosin

    The structural configuration of G+- is that associated with location! and locatum! verbs and

    conflation proceeds in the usual way7

    /#- F

    F .

    corral 5. .

    rosin the calves . ?

    the rope

    4n the case of G*-, the structure is superficially the same, as indicated in /&-7

    /&- F

    F . ? .

    calf . 5.

    rope in the corral

    with rosin

    $4

  • 7/23/2019 Syntactic Alternations in Englishv1

    35/149

    The ill(formedness of the sentences in G*- follows from the fact that the element that

    undergoes conflation is in the "pecifier position of the P-proection, not in the head of that

    pro@ection% Conflation of a Spec is impossible although the verb c(commands the relevant ?%

    Thus, in the case of conflation government is a necessary condition but something stronger

    must also be observed i%e%, the strict head(head local relation-%

    The same restriction to conflation operating on a noun in Spec position e'plains the non(

    e'istence of the location3verbs in /"a, b- and the locatum! verbs in /$a, b- below7

    /"- a% Qe booked on the shelf cf%, e put the books on the shelf = he shelved the books-

    b% Q0e appled the bo' cf%, 0e put apples in the bo' = we bo'ed the apples-

    /$- a% QThey housed with a roof cf%, They fitted a house with a roof = they roofed a house-

    b% QThey water with poison cf%, They contaminated water with poison = they poisoned

    water-

    The restriction on conflation from a Spec position e'plains certain gaps in the English le'icon

    i%e% the impossibility of verbs such as those in G*- and /", /$- on the intended reading%

    2.0. A Note on Con6'at"on o6 &repo)"t"on)

    There is independent evidence that some prepositions and perhaps the category as a whole-occupy a special position among le'ical items, different from the other le'ical items that name

    eventualities F, 2- or entities ?-%

    ;n the one hand, it is known that prepositions evince 7semantic poerty3a property that sets

    them apart from verbs, nouns and ad@ectives i%e% their descriptive content is very poor-%

    .repositions are relational, e'pressing, for e'ample, the motional or locational relation between

    some entity a figure!- and another entity a place!-% There is, therefore, an element of

    conte'tual dependency in their semantics% 2 preposition is interpreted by virtue of the

    construction in which it appears and it shares a feature with the functional category Case% 0hilesome prepositions have semantic content! like under, aboe, etc%- other prepositions can be

    said to be essentially empty semantically, e'pressing @ust a relation e%g% of in most of its

    functions, and to, at, for in many of their functions-% .repositions can fuse with ad@acent

    functional heads in nominal e'tended pro@ections typically with 5-, a process that eliminates

    any vestige of the original, unfused shapes of the individual components% Aor e'ample, the

    $!

  • 7/23/2019 Syntactic Alternations in Englishv1

    36/149

    Arench preposition fuses with the masculine articles to give auand aux, a fact which strongly

    supports the idea that these latter entities e'ist in the Arench morphology as such and that they

    are related to Case%

    ;n the other hand, prepositions are a closed class of lexical items @ust like functional

    categories% The most conspicuous e'ample is the functional category of Case to which they

    have some affinity7 as it is well known prepositions are the historical source of Case in many

    languages%

    3oreover, the study of certain language impairments also provides evidence that members of .

    category e'hibit a behaviour that places them together with functional categories% Aor e'ample,

    agrammatism is characteri9ed by the widespread omission of function words and affi'es and

    the greater retention of content words and studies of the production of prepositions by

    agrammatic patients indicate that different types of prepositions show greater or lesser

    susceptibility to omission )aplan &**", cf% ale and )eyser "###-%

    .repositions ac:uire their phonological inde' in two ways7 i- through Conflation of nominal

    heads- as in /G- below and ii- through Focabulary insertion directly as in on the table, in the

    house, at the moies-%

    /G- F

    F .

    bottle 5. .

    the wine . ?

    ;n this assumption, the failure of . to conflate see //- can be e'plained as in /J-7

    //- a% Q8ohn inned the calves cf%, 8ohn got in the calves in the milkpen-b% QShe onned the horse cf%, She got on the horse-

    /J- F

    F .

    5. .

    the calves . ?

    in

    Since the preposition has no phonological signature=matri' it cannot pass a phonological

    signature to FN hence the impossibility of the starred forms in //- above%

    ?otice, however, that in /J- the head(complement relation does hold between . and F7. is the

    $

  • 7/23/2019 Syntactic Alternations in Englishv1

    37/149

    head of the complement that F takes% 2ccordingly, depending on language specific factors, .

    can conflate with F whenever . has enough content to substantiate conflation and to pass its

    phonological matri' onto the defective verb% That is what happens in the case of a small class of

    prepositions that have verbal uses in English, most likely derived by conflation7

    /1- a% e downed a second glass of te:uila b% They upped the prices

    G%&% 4n contrast, INCOR&ORATION, a process very similar to conflation, observes the c(

    command condition between the items involved but it is not sub@ect to strict head(head local

    relation!% The process of In!orporat"onproved to be remarkably e'planatory in many syntactic

    processes characteristic to polysynthetic agglutinative- languages Baker &*++-%

    Consider, for e'ample, ?oun 4ncorporation in opi in a sentence such as /+- Baker &*++, ale

    W )eyser "###-7

    /+- ?u! pu(t ki(yta4 that(2CC house(2FE

    4 have that as a house!

    The sentence in /+- has the following structure7

    /*- F

    5. F

    nu 5. F

    4! 5 ? (y(

    pu( kii( have!

    that! house!

    4n this case, by the process of incorporation, kii( house! a nominal head- moves and

    incorporates into another head R7y(, verbal this time, giving the denominal possessive verb ki-7y(

    have a house! while the determinerpu( the 2CC form of- that! remains stranded% Conflation

    would be impossible in this case because ? is not a sister of the matri' verbN rather 5. is the

    strict complement of F in accordance with the principle of strict complementation! and the

    3erge operation%

    The noun kiihouse! is internal to 5. and, thus, 5 intervenes between the noun and the target

    verb 87y- have!% 4t follows that the derived ?(F compound ki-7y-7 have house! is derived by

    $

  • 7/23/2019 Syntactic Alternations in Englishv1

    38/149

    incorporation, vi9% by head (o3e(ent, e'tracting the noun from 5. and ad@oining it to the

    governing verb, as shown in J#-7

    J#- F

    5. F nu! 5. F 4! 5 t?i ?i F pu( kii( (y( that! house! have

    2s e'pected, in incorporation, nuclear relations are not changed7 the target F remains the head

    of the construction% 2s the target verb in opi is a suffi' it must attach to an incorporated

    nominal root%

    The peculiarity of this verb is that it pro@ects a Specifier from inside which only the noun moves

    by the process of incorporation% 4n English, this property of pro@ecting a Spec is shared by the

    categories . and 2%

    0e conclude that conflationis not a movement operationN it is sub@ect to the operation of merge

    at the root% 4n contrast, incorporationdoe) involve movement in the above case from a specifier

    position, and it is constrained only by government%

    Theprocess of !onflationdoes not leave a tracebehind the conflated item as conflation is not a

    movement rule% 4t is @ust the substantiation of defective phonological features at 3erge%

    Conflation leaves the entire structure unchanged with respect to synta' and semantic structure%

    Theprocess of Incorporation, a morph(syntactic process, necessarily leaves traces behind the

    moved items as far as it is a movement operation see the e'ample from opi above-%

    4ncorporation is formed by means of the head moement3variant of &oe

    4f these assumptions are correct with respect to derivations that concern both conflation and

    incorporation in that they involve syntactic processes and relations such as c(command,

    government, movement- defined over le'ical ob@ects, then the formation of the le'ical items in

    :uestion is sub@ect to principles that are operative in synta'% 4t follows that the structures over

    which le'ical derivations are defined are true syntactic ob@ects%

    4.0. Intran)"t"3e Ver)

    ;f late, researchers have become more and more aware of a clear lack of homogeneity in the

    behaviour of intransitive verbs .erlmutter, &*1+-, oekstra &*+*-, Bur9io &*+J-,

  • 7/23/2019 Syntactic Alternations in Englishv1

    39/149

    &**#-, ale and )eyser &**$, &**+, &***, "###-% 4t has turned up that superficially intransitive

    verbs divide, according to their distributional properties, into two classes7 one class of verbs is

    usually called the ner%at"3eclass of intransitive verbs pure intransitives- and the other the

    na!!)at"3eor er%at"3e-class of intransitive verbs%

    4.1. The Le*"!a' Ar%(ent Str!tre o6 Uner%at"3e Ver)

    Unergative verbs are basically monadic and their single argument is the 2gent i%e% e'ternal

    argument- of the activity7 talk, shout, dance, hide, run, flo#, ump, #alkTheir uni:ue argument

    is a sub@ect in the 5(structure as well as in the S(structure7

    unergative 4. 5(Structure=S(structure-

    verbs ?. 4L

    4o F.

    F#

    Aor e'ample7

    J&- 8ohn talked about the Eli9abethan dramaJ"- 8ohn shouted at his wife

    2n unusually large number of English verbs give the appearance of being related to nouns R

    e%g% dance, laugh, bottleand saddleis both a noun and a verb, and shele, sheathe, sheae,

    enslae and imprisonare verbs which are clearly related to nouns, in one way or another% Ferbs

    of this sort are :uite generally held to be deno("na'9( they are verbs derived from nouns!%

    They are intransitive verbs of the ner%at"3e type designating activities%

    2 more e'tensive list of denominal verbs is offered below cf% ale and )eyser &**+-7

    belch, burp, cough, cra#l, cry, dance, gallop, gleam, glitter, glo#, hop, ump, laugh, leap, limp,

    nap, run, scream, shout, sleep, skip, sneee, sob, somersault, sparkle, speak, stagger, s#eat,

    talk, trot, t#inkle, #alk, yell

    et us consider a few sentences containing denominal verbs taken from ale W )eyser &**$-

    and Iadford &**1--7

    J$- a% e may protest b% e was lying c% e complained d% e was fishing

    $

  • 7/23/2019 Syntactic Alternations in Englishv1

    40/149

    e% 0e were golfing f% The child laughed g% 3ary sang h% The cow calved

    The interesting fact about these intransitie erbs is that they generally have erb 6 nounparaphrases:

    JG- a% e may make a protestb% e was telling lies

    c% e made a complaint d% e was catching fish e% 0e were playing golf f% 0e had a good laugh g% 3ary did her new song h% The cow had a calf

    This shared property of the two classes of verbs is captured if, following ale and )eyser

    &**+-, both structures are assigned the structure in J/-7

    J/- a- F a!- F

    F 5. F ?

    had a laugh laugh

    4n the case of the transitive construction J/a-, the complement 5. is realised by the process of

    7er%ein the usual manner%

    The intransitive construction, on the other hand, involves a phonologically empty verb and a

    bare nominal root% This structure can be viewed as forming an implicitly transitive verb i%e%, a

    hidden transitive-, constructed by merging the noun laughwith an abstract verb %

    The de(nominal verb laugh is formed by CONFLATION i%e% fusion- of the noun complement into

    the verb% The semantic structure associated unambiguously with the syntactic structure in

    J$a!-, would be7 an action or event implicates! an entity of the le'ical category ? i%e% e n-%

    Thus, the implicating event is completed or perfected by virtue of the creation!, production! or

    realisation! of the relevant entity%

    The structures in J/a, a!- corresponding to the relational structures for unergatives and simple

    transitive verbs respectively, clearly must have subectsas arguments in s(synta' as it is the

    case with all the members of the category F which at some point pro@ect a sub@ect-%

    40

  • 7/23/2019 Syntactic Alternations in Englishv1

    41/149

    The fact that the sub@ect=specifier of unergative verbs is always an 2gent, that argument will

    necessarily be an e'ternal argument%

    4f unergatives are hidden transitives it also means that, @ust as with transitives, their structure

    should be binary in spite of the fact that they are intransitive verbs-% owever, since they are

    derived by the operation of conflation their structure becomes! binary at the level of le'ical

    structure% 3oreover, as unergative verbs @ust like transitive verbs- have an agentive e'ternal

    argument, they will also contain an agentive light verb in their structure as it is conceived in

    the 3inimalist framework Chomsky &**1-7

    JJ- 4!

    4# .

    Su !

    # F.

    F#

    2n interesting :uestion raised by ale and )eyser &***- is whether or not unergatives can

    appear in a le'ical syntactic causative construction like transitive verbs i%e% #hether unergatie

    erbs can turn transitieand whether transitive verbs can further transitivi9e, for that matter-%

    To see this in the case of unergatives, we have to test whether the structure in J/a!- can

    appear as a complement of a verbal head that has its e'ternal argument within a le'ical

    pro@ection% Consider the following e'amples with unergative verbs where an e'ternal sub@ect

    has been inserted7

    J1- a% QThe clown laughed the children cf% got the children to laugh-b% Q

  • 7/23/2019 Syntactic Alternations in Englishv1

    42/149

    the children F ?

    laugh

    The meaning of unergative verbs does not imply a causal relation, rather they are used in a

    performatie sense,e%g%, he laughedcan be paraphrased as he performed the act of laughing3%

    4t follows that unergatives have no inner sub@ect! in their le'ical argument structure

    representation i%e%, they do not pro@ect an internal 5. sub@ect the children- in le'ical synta' as

    the transitiveputdoes-%

    0e may also wonder why should this be the case since a structure such as J+- has not been

    prohibited by any principle so far% 4n other words, the :uestion is the following7 what is it that

    forces the appearance or absence of a sub@ect> The answer is provided by a general syntactic

    principle according to which linguistic structures must be fully interpreted3% This principle is

    called the &r"n!"p'e o6 F'' Interpretat"on Chomsky &*+J- and states that elements that

    contribute to the semantic interpretation of a structure must be present in the syntactic

    representation of that structure% The .rinciple of Aull 4nterpretation will guarantee that verbs of

    change of locationsuch asput- have a sub@ect 5. in the inner .(pro@ectionN absence of the

    sub@ect would leave the complement 5. of this pro@ection uninterpreted% The same principle will

    guarantee that unergative verbs lack a sub@ect! in their le'ical argument structure

    representations% 2 sub@ect!, if present, in an unergative argument structure representation,

    would itself go uninterpreted for lack of a predicate in the complement position remember that

    the complement laughmerges and ultimately conflates with an empty head F-% 4t follows that the

    subectof an unergative verb is a true external subect3%

    The crucial conse:uence of unergatives lacking a specifier! of their own is that they do not give

    rise to transitive pendants of the kind Qthe clo#n laughed the children% They do not participate

    into the transitivity alternation% This follows from the fact that neither the verbal head of the

    le'ical structure of laughnor its complement the noun laugh- belongs to the type of elements

    that force the appearance of a specifier in the pro@ection of the host verb% ence, there is no

    place in the le'ical structure for the surface ob@ect the children in the hypothetical transitive

    clause Qthe clo#n laughed the children% Unergative verbs are unable to transitivi9e% ;f course, insentential synta', the sub@ect will be in the Spec position of a functional category 4.-%

    2s mentioned above, these remarks on internal sub@ects do not only apply to unergative verbs

    that involve conflationN to the same e'tent, they do not apply to their 7analytic3counterparts in

    which the main verb appears with an overt complement, i%e%, to transitie erbs% Consider the

    following e'amples that contain the verbs gie and hae7

    4"

  • 7/23/2019 Syntactic Alternations in Englishv1

    43/149

    J*- a% The cow gave birth b% The cow had a calf

    These verbs pro@ect the same structure as the empty verb of J/a!-7

    1#- a% F b% F

    F 5. F 5.

    give birth have calf

    These are analytic representationsof the simple head(complement configuration ( they are the

    result of 7er%ealone% The verb of J/a! i%e%, laugh-, on the other hand, represents the synthetic

    type, so called because it is the result of both 3erge and concomitant conflation% Thus, in the

    intended sense, unergatie erbs are the synthetic counterparts of their analytic transitie

    paraphrases%

    2s stated above, the synthetic and analytic forms share the property that the head pro@ects no

    specifier and, conse:uently, neither can undergo transitiisation% Thus, @ust as Qthe clo#n

    laughed the children- is ungrammatical, so also the sentences in 1&- are e:ually

    ungrammatical7

    1&- a% Q2n in@ection gave the cow birth early

    b% Q2n in@ection had the cow a calf

    4t follows that transitie structures cannot further transitiie% The insertion of 1#- in the

    complement position of a matri' empty verb leads to the same transitivity failure as noted in

    relation to J+- above7

    1"- F&

    F& F"

    F" 5.

    give birth

    have calf

    The failed interpretations here are appro'imately an inection brought it about that the co# gae

    birth early3and an inection brought it about that the co# had a calf3% owever, the sentence the

    4$

  • 7/23/2019 Syntactic Alternations in Englishv1

    44/149

    co# had a calfis not the source of the co# caled% The claim is simply that they share the same

    argument structure configuration the same specifier(head and head(complement relations- and

    conse:uently, they evince a certain syntactic behaviour%

    Since the sub@ect of F" i%e% the co#, is an e'ternal argument, it will not appear as a specifier in

    the le'ical argument structure of that verb, by hypothesis% 4t will therefore not be possible for it to

    function as the sentential syntactic ob@ect of the derived verb% 2 structure such as 1"-, where

    the verb gieraises to F&, will not result in the putative Qgie the co# birth, Qhae the co# calf%

    The 5. the co#simply cannot appear in the ob@ect position of gieor hae%

    This is accounted for under the assumption that the verb which heads the complement R i%e%,

    gie, haeR does not pro@ect a specifier, @ust as the empty verb of J/a!- does not%

    4.,. On Co%nate O5e!t)

    0e have already taken the theoretical stance that the process of Con6'at"on, since it is not a

    movement rule, does not leave traces% 0hat conflation does is to substantiate a defective

    phonological matri' by 3erge and copy of the resulting constituent onto a higher le'ical item%

    Conflation leaves the entire structure unaffected with respect to synta' and semantic structure%

    !ognate obects do occur as direct ob@ects of unergatie (intransitie) erbs% 4n a standard

    manner, cognate ob@ects are root(identical to the nominal component of the associated de-

    nominal erbas in 1$- ale and )eyser "###-7

    1$- a% She slept the sleep of the @ustb% e laughed his last laugh

    c% e died a peaceful death

    The structures in 1$- turn transitive, and they are headed by de(nominal verbs whose ob@ect is

    headed by a noun which is root(identical to the verb% This latter property of cognate ob@ect

    constructions is important as the ob@ect cannot be @ust some random noun but it is necessarily

    root(identical to the verb% Consider the following e'amples where this condition is not met7

    1G- a% QShe slept her last nap = along winter slumberb% QShe laughed a surreptitious giggle = chuckle

    0e follow ale and )eyser "###- and hypothesise that cognate ob@ects are derived by the

    process of conflation

    0e have seen that in the case of the derivation of the de(nominal verb laugh, the conflating

    44

  • 7/23/2019 Syntactic Alternations in Englishv1

    45/149

    element the noun laugh- has been the head of a le'ical pro@ection necessarily sister to the

    target verb the derived verb laugh-% 4n this case, the noun laugh, which provides the

    phonological matri' to the verb, is taken as a maximal lexical proection 9not an e'tended

    pro@ection 5.-%

    To e'plain the formation of cognate ob@ects we assume that the verb enters into a relation not

    @ust with the ma'imal le'ical pro@ection ?, but with its extended maximal proection Pto give

    sleep the sleepof the ust-% Consider the derivation of sleep the sleep of the ustproposed by

    ale and )eyser "###-7

    1/- F

    F 5 the , sleep-

    sleep 5 ? sleep-

    the ? .

    sleep of the @ust

    This derivation is in conformity with the general sisterhood limitations on the process of

    conflation% This follows under the assumption that certain features of the le'ical head ?- pro@ect

    to its e'tended pro@ection 5 the sleep-% 4f the features that pro@ect so high include the

    phonological matri' of the le'ical head ?, then, the defective verb is a structural sister to the

    node 5 that bears the phonological matri' of the le'ical head% The nominal phonological matri'

    Dsleep is then copied onto the defective label of F% 4n this way, the cognate ob@ect head the

    sleep- becomes sister = complement to the target verb as its relation to the latter cannot be

    more distant than this%

    ?otice that it does not do to move the inner ? to F and ad@oin it to F to give sleep the sleep- as

    the determiner 5- is a head that intervenes between the noun and the verb% E'traction across

    5. would violate ead 3ovement Constraint Travis &*+G-% Het, in the particular case of the

    cognate ob@ect, conflation does look like operating at long distance!, but only in the sense that it

    bypasses the functional category structure of an e'tended pro@ection% Even in this case,

    conflation is still restricted to the head(complement relation%That is why the cognate ob@ect construction bears the name that it does R the conflating element

    ?- is necessarily represented by a copy, its cognate!, in the target head position

    etymologically, cognate! derives from related by blood!, related on the mother!s side!-%

    The above e'ample plainly shows that conflation is a copying process% Under certain conditions,

    copies are realised overtly in both positions, i%e%, target and source% This latter possibility is

    4!

  • 7/23/2019 Syntactic Alternations in Englishv1

    46/149

    evinced in case the source position is dominated, within its e'tended pro@ection, by certain

    functional categories, e%g%, 5% ;therwise, the copy is deleted in the source position, as in the

    canonical conflation relation, illustrated below7

    1J- F

    F ?

    laugh

    :.0. The Le*"!a' Ar%(ent Str!tre o6 Una!!)at"3e Ver)

    ike unergatives, unaccusative verbs such as arrie, remain, fall- are monadic predicates but

    their uni:ue argument is attributed a direct ob@ect position at the level of 5(structure% Their sole

    argument ends up in the sub@ect position in the S(structure as in the configurations below7

    unaccusative7 5(Structure

    verbs 4.

    ?.& 4L

    4o F.

    FL

    Fo ?."

    4. S(Structure

    ?.& 4L

    4o F.

    FL

    Fo ?."

    t

    Consider the s(synta' of the italicised arguments in the following unaccusative structures7

    4

  • 7/23/2019 Syntactic Alternations in Englishv1

    47/149

    11- a% There arose an unfortunate misunderstanding b% There came a cry of anguishfrom inside the house c% There appeared a ghostly faceat the window d% 4n front of the house, there stands a statue of 1eneral 1houly

    Iadford &**1--

    4n some respects, the italicised arguments seem to behave like complements as they occupy

    the postverbal position typical for complements%

    owever, in other ways, they seem to behave like sub@ects as the italicised argument agrees

    with the verb preceding it% Aor instance, standsin 11d- is a singular form which agrees with the

    singular nominal a statue of 1eneral 1houly%

    3oreover, the post(verbal argument carries the nominatiecase associated with sub@ects, not

    the accusative case associated with complements% 0e can easily pre(pose the KcomplementK of

    these verbs and the structures obtained are also well formed7

    1+- a%An unfortunate misunderstandingarose b A cry of anguishcame from inside the house c A ghostly faceappeared at the window d A statue of 1eneral 1houlystands in front of the house

    This is not a general property of verbs% ;nly certain types of verbs seem to allow postverbal

    sub@ectsN notice the ungrammaticality of the sentences below from Iadford &**1-7

    1*- a% Q4n the dentistLs surgery, there groaneda toothless patientb% QThere has apologied3a@or 3uddle for his minor indiscretions

    Ferbs like those in 11-, which can have post(verbal sub@ects, are referred to as na!!)at"3e

    3er)% 4n contrast, verbs with 2gent sub@ects but no overt ob@ect like those in 1*- are, as said

    above, ner%at"3e 3er)%

    4n the case of unaccusatives, two factors are essential7

    i-% their uni:ue argument is normally Theme not 2gent- and, as such, they lack an e'ternal

    argument% Their uni:ue argument is pro@ected in the synta' as an underlying ob@ect, in the frame

    DP5.%

    ii-% they typically designate states be, remain, lie, etc%- or changes of state occur, arrie,

    appear, arise,etc%-%

  • 7/23/2019 Syntactic Alternations in Englishv1

    48/149

    3oreover, they allow There(4nsertion as in There came a cry of anguish from inside the house

    The name unaccusatiesuggests that these verbs are unable to assign 2ccusative Case and

    that is why the underlying ob@ect! ends up as surface sub@ect%

    Aollowing .erlmutter &*1+-, they are referred to as unaccusatie erbs%

    4n the 3inimalist framework, their syntactic structure is as in +#-N they lack an e'ternal

    argument and, as such, they do not contain a . different from transitives and unergatives-%

    They are made up of @ust a F.7

    +#- 4!

    4# F.

    F!

    F# Su

    This type of intransitive verbs is intrinsically intransitive and they do not participate in transitive

    alternations%

    Bur9io &*+&,&*+J- identified another class of verbs such as melt, sink, crack, split, open,

    close, increase, break, bend, collapse, tighten, shorten, loosen, redden, slide, moe, shake,

    float, dropetc%- that display the following properties7

    i-% ike unaccusative verbs, they are one(internal argument intransitive verbs as in The ship

    sank

    ii-% ike unaccusative verbs, they pro@ect their The(eargument in the ob@ect position at 5(

    structure that at S(structure becomes sub@ect%

    iii-% 5ifferent from unaccusatives, they hae a transitie pendantas in The enemy sank the ship

    = The ship sank% The two(argument structure of sink correlates with a semantic difference

    between the two uses of sink% 4n its intransitive use the Kob@ectK theship- is engaged in some

    activity the sinking-N in its transitive use, the e'ternal argument specifies who is responsible for

    the sinking as in The enemy sank the boat% 2n element of causation is, thus, involved in the

    transitive use of verbs of the type sink%

    iv-% Unlike unaccusative verbs of state or changes of state, the sinkclass of verbs do notappear in KthereK construction as in QThere sank t#o ships last #eek%

    This latter class of intransitive verbs is referred to as er%at"3e 3er)Bur9io &*+&-, &*+J--% et

    us compare the structure of a transitive construction with that of an ergative construction and

    see how the derivation works out for this latter class of verbs7

    4(

  • 7/23/2019 Syntactic Alternations in Englishv1

    49/149

    +&- The artillery sank two enemy ships vs%+"- Two enemy ships sank

    4. 5(structure=S(structure

    ?. 4L

    The artillery 4o F.

    (ed FL

    Fo ?.

    sink two enemy ships

    4. 5(structure

    ?. 4L

    De 4o F.

    (ed FL

    Fo ?.

    sink two enemy ships

    This e'ercise is instructive because we notice that in both the transitive and the intransitive

    configurations t#o enemy shipsbears the same theta role Theme- in the same 5(structure

    position, despite the change in its grammatical function% KThe assumption is that thematic roles

    assigned by a verb are assigned to the same syntactic elements in different constructionsK

    oekstra &*+G--% Baker &*++- formali9ed this grammatical empirical fact in a hypothesis

    called Uniform Theta 2ssignment ypothesis UT2- which states7

    Uniform Theta 2ssignment ypothesis

    Identical theta relationships bet#een items are represented by identical structural relationships

    bet#een these items at the leel of -structure

    4

  • 7/23/2019 Syntactic Alternations in Englishv1

    50/149

    Since ergative verbs do participate in transitive alternations we shall devote an e'tensive

    description of their le'ical argument structure%

    The above distinction between unaccusatieand ergatie verbs is indeed a very fine(grained

    classification of intransitive verbs% owever, since both the arrieclass of verbs and the sink

    class of verbs involve essentially the same process of base(generating their Theme in ob@ect

    position with its subse:uent movement to the sub@ect position in the synta', the general practice

    is to place them under the larger umbrella of Kunaccusatie erbsK The Unaccusative

    ypothesis! first formulated by .erlmutter and .ostal &*+G--%

    The widely acknowledged pattern in +&(+"- is assumed to hold only for a restricted set of verbs

    and this pattern is known as the inchoatie alternation Ieinhart "###-% 2n overwhelming

    ma@ority of unaccusative verbs across languages have an active transitive alternate% Het, other

    unaccusatives do not e%g% the English die, come, the 4talian crescere-% To account for this gap in

    the paradigm Chierchia &*+*- and Ieinhart "###- hypothesi9e that, cross(linguistically,

    unaccusatives with no transitive alternate are derived from some abstract transitive verb whose

    result is a fro9en form% That the le'icon contains fro9en forms! is not that surprising as such

    forms can be identified in other areas, such as that of intrinsic refle'ives e%g% ich-schamenbe

    ashamed! in 5utch or behaein both 5utch and English are fro9en refle'ives that can not be

    used transitively-%

    2 traditional e'planation of the peculiar behaviour of unaccusatives! sub@ect, due to Bur9io

    &*+J-, is that the sub@ect of unaccusatives does not originate as the sub@ect of their associated

    verbs but rather as their complements as shown above-% 2t the level of 5(structure,

    unaccusative structures with postverbal arguments involve leaving the relevant argument Kin

    situK, in F.(complement position% Het, the unaccusative verb cannot assign the 2ccusative case%

    Bur9io &*+J- argues that unaccusative verbs are also unable to theta mark their sub@ects%

    Since the sub@ect does not receive a theta role and the ob@ect ?. is not case marked by the

    unaccusative verb, the ob@ect ?. has to move to the sub@ect position to receive ?ominative

    case% This generali9ation has come to be known as Bur9ioLs

  • 7/23/2019 Syntactic Alternations in Englishv1

    51/149

    ii- 2 verb which fails to assign 2CCUS2T4FE case fails to theta(mark an e'ternalargument Bur9io, &*+J 7&+G-

    ?otice that Bur9ioLs generali9ation captures the relation between two independent modules of

    grammar, namely Theta Theory and Case Theory% 2s far as the latter theory is concerned, there

    is a universal re:uirement that all overt ?.s must be assigned Case% inguists have tried to

    relate this re:uirement to other properties of the grammar% Aor instance, it is known that

    argument ?.s must be assigned a theta role% The idea is then that a predicate can only assign

    a theta role to ?.s that are isibleCase renders a ?. visible% The isibility re;uirementon ?.s

    is related to Theta Theory7 in order to be recogni9ed as an argument of some predicate a ?.

    must be made visible-%

    :.1. 7ore on the Le*"!a' Ar%(ent Str!tre o6 Una!!)at"3e) and the La!$ o6 A!!)at"3eCa)e A))"%n(ent

    2lthough it is clear that unaccusative verbs are unable to assign 2ccusative Case and they do

    not participate into the transitivity alternation, the :uestion hale and )eyser "###- rise is what

    peculiarity of their argument structure prohibits the assignment of Case%

    2fter all, some unaccusative verbs such as arrieappear to select a .(pro@ection as in +$a and

    +$b below- @ust like the transitive verb put in +G-7

    +$- a% 3any guests arrived at the party

    b% There arrived many guests at the party+G- 8ohn put the book on the table

    Iemember that a verb which selects as its complement a structure pro@ected by a preposition, is

    e'pected to be transitive7

    +/-