Noun Phrase Accessibility and Universal Grammar

Download as pdf or txt
Download as pdf or txt
You are on page 1of 38

Noun Phrase Accessibility and Universal Grammar

Author(s): Edward L. Keenan and Bernard Comrie


Source: Linguistic Inquiry, Vol. 8, No. 1 (Winter, 1977), pp. 63-99
Published by: The MIT Press
Stable URL: http://www.jstor.org/stable/4177973
Accessed: 22/12/2009 12:04

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=mitpress.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].

The MIT Press is collaborating with JSTOR to digitize, preserve and extend access to Linguistic Inquiry.

http://www.jstor.org
Linguistic Inquiry Volume 8 Number I (Winter, 1977) 63-99.

Noun Phrase Accessibility and


Edward L. Keenan
Bernard Comrie Universal Grammar*

In section 1 we present the Accessibility Hierarchy, in terms of which we state three


universalconstraintson Relative Clause Formation.In addition,we present the data in
support of these constraintsand discuss certain partialcounterexamples. In section 2
we propose a partialexplanationfor the hierarchyconstraintsand present furtherdata
from Relative Clause Formationsupportingthese explanations. Finally, in section 3 we
refer briefly to other work suggesting that the distributionof advancementprocesses
such as Passive can be described in terms of the Accessibility Hierarchy;these facts
show that the proposed explanationfor the Hierarchyneeds to be generalized

1. The Accessibility Hierarchy


1.1. Two Methodological Preliminaries
We are attemptingto determinethe universal properties of relative clauses (RCs) by
comparing their syntactic form in a large number of languages. To do this it is
necessary to have a largely syntax-freeway of identifyingRCs in an arbitrarylanguage.
Our solution to this problemis to use an essentially semanticallybased definition
of RC. We consider any syntactic object to be an RC if it specifies a set of objects
(perhapsa one-memberset) in two steps: a largerset is specified, called the domain of
relativization, and then restricted to some subset of which a certain sentence, the

* This article is a developed version of Keenan and Comrie (1972). We attempt in this version to
account for several objectionsto the earlierformulationthat have since been broughtto our attention.
We would like to acknowledgethe followingsources for significanthelp with this article:
(i) One LinguisticInquiryreviewerfor constructivecriticismsof an earlierdraft;
(ii) Supportfrom an NSF postdoctoralresearchgrant and two Wenner-Grengrants, 2384 and 2944,
for work on the Malayo-Polynesianlanguagesreportedherein;for supportof the psycholinguisticwork on
the Accessibility Hierarchywe are indebtedto a grantfrom the Nuffield Foundation;
(iii) The willing and substantive help from the following linguists concerning languagesthey have
worked with: G. Brettschneider(Basque), P. Brown (Tzeltal), M. Butar-Butar(Toba Batak), S. Chung
(Indonesian, Tongan), 0. Dahl and J. Allwood (Swedish), K. Ebert (North Frisian, Kera), J. Hawkins
(German,Slovenian), A. Janhunen(Finnish), M. Perera(Catalan),H. van Riemsdijk(ZurichGerman),J. de
Rooij (Dutch), A. Salmond(Maori), P. Sgall (Czech), R. Tanaka(Japanese),B. Vattuone (Genoese), N.
Vincent (Italian);
(iv) The willingand substantivehelp of our informants.
64 EDWARD L. KEENAN AND BERNARD COMRIE

restricting sentence, is true.' The domain of relativization is expressed in surface


structureby the head NP, and the restrictingsentence by the restrictingclause, which
may look more or less like a surface sentence dependingon the language.
For example, in the relative clause the girl (that) John likes the domain of
relativizationis the set of girls and the head NP is girl. The restrictingsentence is John
likes her and the restrictingclause is (that) John likes. Clearly, for an object to be
correctly referredto by the girl that John likes, the object must be in the domain of
relativizationand the restrictingsentence must be true of it. We shall refer to the NP in
the restrictingsentence that is coreferentialwith the head NP as the NP relativizedon
(NPrei); in our example, this is her, i.e. the direct object of John likes her.
Note that we only consider definite restrictive RCs in this study. The role of the
determinerthe is held constant and ignored, and the term RC is used to apply to the
collocation of the head NP and the restrictingclause.
Note furtherthat our semanticallybased notion of RC justifies consideringas RCs
certain constructions that would perhaps not have been so considered in traditional
grammar. Thus, in German, alongside the traditional RC in (1) we also count the
participialconstructionin (2):
(1) der Mann, der in seinem Buro arbeitet
the man who in his study works
'the man who is workingin his study'
(2) der in seinem Buro arbeitendeMann
the in his study working man
'the man who is workingin his study'
As the German data above illustrate, not only do different languages vary with
respect to the way RCs are formed, but also within a given languagethere is often more
than one distinct type of RC. We shall refer to distinct ways of forming RCs as
different relative clause forming strategies. Different strategies differ with regard to
which NP positions they can relativize. Thus, the participialstrategy in (2) above can
only relativize subjects (that is, the head NP can only be understoodto function as the
subject of the main verb of the restrictingclause), whereas the strategy in (1) above
functions to relativize almost any major NP position in simplex sentences. Conse-
quently, generalizationsconcerningthe relativizabilityof different NPs must be made
dependenton the strategiesused. It will be criticalthereforeto provide some principled
basis for deciding when two different RCs have been formed with differentstrategies.
There are many ways RCs differ at the surface, and hence many possible criteria
for determiningwhen two strategies are different. We have chosen two criteria that
seem to us most directly related to our perception of how we understandthe meaning
of the RC-that is, of how we understandwhat propertiesan object must have to be
1 A more formal statement of this semantic notion in which the logical structure of an RC is represented
as a pair, a common noun phrase and an open formula, can be found in Keenan (1972a).
NOUN PHRASE ACCESSIBILITY AND UNIVERSAL GRAMMAR

correctly referred to by the RC. The first concerns the way the head NP and the
restrictingclause are distinguishedat the surface, and the second concerns how the
position relativized is indicated.
In the first case we consider two RCs to be formed by different strategies if the
relative position of the head NP and the restricting clause differs. There are three
possibilities: the head occurs to the left of the restricting clause, as in (1) above
(postnominal RC strategy); the head occurs to the right, as in (2) (prenominal RC
strategy);or the head occurs within the restrictingclause (internal RC strategy),as in
(3) and (4), from Bambara(Bird (1966))and Diguefio (Gorbet (1972)), respectively:
(3) a. ne ye so ye.
I Past horse see
'I saw a horse.'
b. ne ye so min ye
I Past horse which see
'the horse that I saw'
c. tye ye ne ye so min ye san.
man Past I Past horse which see buy
'The man bought the horse that I saw.'
(4) a. tOnay ?awa:+ 0 ?awu:w.
yesterday house DO I-saw
'I saw the house yesterday.'
b. ?owa: + pu + LI ?ciyawx.
house Def in I-will-sing
'I will sing in the house.'
C. [NP[S tanay ?;wa: + 0 ?3wu :w]] + pu + LU?ciyawx.
yesterday house DO I-saw Def in I-will-sing
'I will sing in the house that I saw yesterday.'
In the second case we consider two RCs to be formed from differentstrategiesif
one presents a nominal element in the restrictingclause that unequivocallyexpresses
which NP position is being relativized, and thus we know exactly what the restricting
clause is saying about the head NP (that is, we can recover the restrictingsentence
from surface) (+case RC strategy). For example, the English strategy that forms the
girl who John likes is not case-coding since who, the only relevant particle in the
restrictingclause, can be used as well if the role of the head NP in the restrictingclause
is different, e.g. the girl who likes John (-case RC strategy). On the other hand, in
comparablesentences in Russian, (5a) and (5b), the form of the relative pronoundoes
unequivocally tell us the role of the head NP, so that strategy in Russian is case-
coding:
(5) a. devuska, kotoruju Dzon Ijubit
girl who (accusative)John likes
'the girl who John likes'
66 EDWARD L. KEENAN AND BERNARD COMRIE

b. devuska, kotoraja ijubit Dzona


girl who (nominative)likes John
'the girl who likes John'
Note, however, that RCs in English like the chest in which John put the money are
considered case-coding, since the prepositionin, which indicates the role of the head
NP, is present in the restrictingclause.
In addition to the use of relative pronouns, case can be coded in another way in
the languagescovered in our study. Namely, a personalpronouncan be present in the
NP position relativized, as in Hebrew:
(6) ha- isha she- Yon natanla et ha- sefer
the woman that John gave to-her DO the book
'the woman that John gave the book to'

1.2. The Accessibility Hierarchy and the Hierarchy Constraints


1.2.1. Statement of the Hierarchy and the Constraints. On the basis of data from
about fifty languages,we arguethat languagesvary with respect to which NP positions
can be relativized, and that the variationis not random. Rather, the relativizabilityof
certain positions is dependenton that of others, and these dependencies are, we claim,
universal. The Accessibility Hierarchy(AH) below expresses the relative accessibility
to relativizationof NP positions in simplex main clauses.

Accessibility Hierarchy (AH)


SU > DO >LO > OBL > GEN > OCOMP

Here, ">" means 'is more accessible than'; SU stands for 'subject', DO for 'direct
object', 10 for 'indirectobject', OBL for 'majoroblique case NP' (we intendhere NPs
that express arguments of the main predicate, as the chest in John put the money in the
chest ratherthan ones having a more adverbialfunction like Chicago in John lives in
Chicago or that day in John left on that day), GEN stands for 'genitive' (or
'possessor') NP, e.g. the man in John took the man's hat, and OCOMP stands for
'object of comparison', e.g. the man in John is taller than the man.
The positions on the AH are to be understood as specifying a set of possible
grammaticaldistinctionsthat a languagemay make. We are not claimingthat any given
languagenecessarily distinguishesall these categories, either in terms of RC formation
or in terms of other syntactic processes. For example, some languages (e.g. Hindi)
treat objects of comparison like ordinaryobjects of prepositions or postpositions. In
such cases we treat these NPs as ordinaryOBLs, and the OCOMP position on the AH
is unrealized. Similarly, in Gary and Keenan (1976) it is argued that the DO and 10
positions are not syntactically distinguished in Kinyarwanda, a Bantu language.
Further, it is possible that in some language RC formationmight distinguishbetween
NOUN PHRASE ACCESSIBILITY AND UNIVERSAL GRAMMAR

two types of DOs. If this were so, we would have to expand the AH at that point and
say that languageslike English do not make the distinction. For the moment, however,
we take the AH as specifying the set of possible grammaticaldistinctionsto which RC
formation(from simplex main clauses) may be sensitive, since our data do not appear
to justify any furtherrefinementin the categories.
In terms of the AH we now give the HierarchyConstraints(HCs):

The Hierarchy Constraints (HCs)


1. A languagemust be able to relativize subjects.
2. Any RC-formingstrategy must apply to a continuous segment of the AH.
3. Strategiesthat apply at one point of the AH may in principlecease to apply at
any lower point.

The HCs define conditions that any grammarof a human language must meet. HC1
says that the grammar must be designed to allow relativization on subjects, the
uppermostend of the AH. Thus, for example, no languagecan relativize only DOs, or
only locatives. It is possible, however, for a language to allow relativizationonly on
subjects (and this possibility is in fact realized;see 1.3.1 for examples). HC2states that,
as far as relativizationis concerned, a languageis free to treat adjacentpositions on the
AH as the same, but it cannot "skip" positions. Thus, if a given strategycan apply to
both subjects and locatives, it can also apply to DOs and 1Os. And HC3 states that
each point of the AH is a possible cut-off point for any strategythat applies to a higher
point. This means that in designingthe grammarfor a possible human language,once
we have given it a strategy that applies at some point on the AH, we are free to
terminateits applicationat any lower point.
Note that it is HC2 that justifies the actual orderingof terms in the AH. Further,
HC2allows as a special case that a particularRC-formingstrategy may apply to only a
single position. Thus several languages(e.g. Javanese (see Table 1)) have recourse to a
case-coding strategy for positions low on the AH (e.g. genitives), whereas the strategy
for major NPs is not case-coding. On the other hand, HC3 states that each point on the
AH is relevant. Thus, if no languagecould have an RC-formingstrategythat appliedto
DOs but not to 1Os, then the data would not justify makingthis distinctionin the AH.
In section 1.3 we substantiatethe claim that each point on the AH is in fact a possible
cut-off point by showing that for each point on the AH some languagehas a strategy
that cuts off at that point.

1.2.2. The Primary Relativization Constraint. In 1.2.1 we stated that the AH


determines, universally, the degree of accessibility to RC formation. But it is perhaps
not obvious that the HCs actually make that intuitionexplicit. However, the Primary
Relativization Constraint(PRC) below does make that intuition more explicit, and it
logically follows from the HCs. To state it we first define an RC-formingstrategyin a
68 EDWARD L. KEENAN AND BERNARD COMRIE

given languageto be a primarystrategy (in that language)if it can be used to relativize


subjects. The termprimaryis justified in that, by HC1, a languagemust have a primary
strategybut need have no other. Thus, of the various RC-formingstrategiesa language
may have, only primaryones are necessary. We can now state:

The Primary Relativization Constraint (PRC)


1. A languagemust have a primaryRC-formingstrategy.
2. If a primary strategy in a given language can apply to a low position on the
AH, then it can apply to all higherpositions.
3. A primarystrategy may cut off at any point on the AH.

Clearly, PRC1is just a restatementof HC1. PRC2follows directly from HC2 and the
definitionof primary, since a primarystrategythat can relativize a low position on the
AH is one that can relativize subjects(since it is primary)and a low position, and so by
HC2 it can relativize all intermediatepositions. And PRC3 is simply a special case of
HC3.
Thus the PRC states that, restricting our attention to primary RC-forming
strategies, relativizabilityof a low position on the AH entails relativizabilityof all
higher positions. And the converse fails. We may be able to relativize only subjects,
both subjects and DOs, or subjects, DOs, and 1Os, etc. So the possibility of
relativizing (with a primary strategy) decreases as we go down the AH, and in that
sense the furtherwe descend the AH the harderit is to relativize.
Note that there is a more obvious, and stronger,way to express the intuitionthat
relativizabilitydecreases as we descend the AH. Namely, "if a languagecan relativize
any position low on the AH, then it can relativizeall higherpositions". This claim says
in effect that RC-formingstrategies always distribute themselves so as to cover an
initial segment of the AH. But this formulationis falsified by our data.
Thus Toba Batak (a Malayo-Polynesian language spoken in Sumatra) has a
primarystrategy, illustratedin (7), which is postnominaland -case:
(7) a. manussi abit boru-borui.
wash (active) clothes woman the
'The woman is washing clothes.'
b. boru-boruna manussiabit i
woman that wash clothes the
'the woman who is washing clothes'
(Note that the basic word order is VOS. See Silitonga(1973)for furthersubstantiation
of this point.)
However, direct objects cannot be relativized using this or any other strategy in
Toba, as we see from (8a). The only way to achieve the semantic effect of (8a) is first
to passivize the underlyingsentence, (8b), and then relativize on the derived subject,
NOUN PHRASE ACCESSIBILITY AND UNIVERSAL GRAMMAR

(8c):
(8) a. *abit na manussiboru-borui
clothes that wash woman the
'the clothes that the woman is washing'
b. disussi ni boru-boruabit i.
wash (passive) by woman clothes the
'The clothes were washed by the woman.'
c. abit na nisussi ni boru-borui
clothes that washed by woman the
'the clothes that were washed by the woman'
(Note that the passive prefix di- becomes ni- in subordinateposition.)
Thus Toba, like many Malayo-Polynesianlanguages (Keenan (1972b)), cannot
directly relativize NPs that can be promotedto subject. It must promotethem and then
relativize them as subjects. However, in distinction to Philippine languages and
Malagasy, for instance, Toba has a ratherlimited promotion("voicing") system. NPs
governed by prepositions,including1Os, cannot be systematicallypromotedto subject.
To relativize them, a second strategy, this time +case, can be used. The relativization
markeris different, and a personal pronounis retainedin the position relativized:
(9) dakdanaki, ima- na nipaboani si Rotua turi-turian-i tu ibana
child the namely that told by Art Rotua story the to him
'the child that Rotua told the story to'
Clearly, then, Toba has "gapped" RC-formingstrategies. One applies only to
subjects, the other to 1Os, OBLs, and GENS. The NPs in the "gap", in this case the
DO, must be promotedto subjectto be relativized. All the languagesin our samplethat
present gapped strategies (see Table 1) are similar to Toba in this respect. That is,
unrelativizableNPs can be systematically promoted to higher positions on the AH,
whence they can be relativized. The fact that NPs lying in strategygaps can always be
promoted to accessible positions does justify the following form of the strong
constraint: If a languagecan relativize any position on the AH, then it can relativize
any higher position either directly or by promotingit to a position that can itself be
relativizeddirectly.
We consider now, in section 1.3, the data that supportthe HierarchyConstraints.
In particular,we show that each point on the AH represents a cut-off point for some
primarystrategy. Then, in 1.4, we consider some of the problematiccases and possible
counterexamples.

1.3. Justification of the Hierarchy Constraints


1.3.1. Subjects Only. In many Western Malayo-Polynesianlanguages, only subjects
can be relativized;considerfor instance Malagasy,which has basic word order V 0 X S
and a developed system for promotingany major NP to subject position. RCs place
70 EDWARD L. KEENAN AND BERNARD COMRIE

the head NP to the left, followed optionally by an invariablerelativizerizay, followed


by the restrictingclause with no pronounin the NPrei position (Keenan (1972b)):

(10) a. Nahita ny vehivavy ny mpianatra.


saw the woman the student
'The student saw the woman.'
b. ny mpianatraizay nahitany vehivavy
the student that saw the woman
'the student that saw the woman'
c. *ny vehivavy izay nahitany mpianatra
the woman that saw the student
'the woman that the student saw'
d. Nohitan' ny mpianatrany vehivavy.
seen (passive) the student the woman
'The woman was seen by the student.'
e. ny vehivavy izay nohitan'nympianatra
the woman that seen the student
'the woman that was seen by the student'

Other Malayo-Polynesianlanguages in our sample that have primary strategies


that apply only to subjects are Javanese, Iban, Minang-Kabau,and Toba Batak; also
Tagalog, on the assumption that the "focus" NP is the subject (see further 1.4.1.1).
Finally, many Europeanlanguages(e.g. German, Russian, and Polish) have participial
RC-formingstrategies that apply only to subjects; cf. (2) above.

1.3.2. Subjects-Direct Objects. In Welsh, the primary strategy applies only to


subjects and direct objects; it presents postnominal RCs introducedby the particlea,
with deletion of NPrel. Other positions are relativized by a different postnominal
strategy, introducedby the particley and with a personalpronounin NPrei:

(11) a. y bachgen a oedd yn darllen


the boy who was a' reading
'the boy who was reading'
b. Dyma 'r llyfr y darllenaisy stori ynddo.
here-is the book that I-read the story in-it
'Here is the book in which I read the story.'

As a second case, one of the primarystrategiesin Finnish applies only to subjects


and direct objects. This strategy places the relative clause prenominally, uses no
relativization marker, and puts the subordinate verb in a nonfinite form (different
dependingon whether the head NP functions as its subject or object). As this strategy
is explicitly discussed in Karlsson (1972), we merely note here the illustrative
NOUN PHRASE ACCESSIBILITY AND UNIVERSAL GRAMMAR

examples:
(12) a. Poydallatanssinut poika oli sairas.
on-table having-dancedboy was sick
'The boy who had danced on the table was sick.'
b. Nakemani poika tanssi poydalla.
I-having-seenboy danced on-table
'The boy that I saw danced on the table.'
As a final case in this category, consider Malay. The primary strategy uses
postnominal restricting clauses introduced by the invariable particle yang with no
pronounretainedin the position relativized:2
(13) Ali bunoh ayam yang Aminahsedang memakan.
Ali kill chicken that AminahProg eat
'Ali killed the chicken that Aminahis eating.'
But this strategyapplies only to subjects and direct objects. Thus from (14a)we cannot
form (14b) or (14c), either strandingor moving the preposition:
(14) a. Ali beri ubi kentangitu kapadaperempuanitu.
Ali give potato the to woman the
'Ali gave the potato to the woman.'
b. *perempuanyang Ali beri ubi kentangitu kapada
woman that Ali give potato the to
c. *perempuankapadayang Ali beri ubi kentangitu
woman to who Ali give potato that
The only systematicallyelicitable possibility was (15):
(15) perempuankapadasiapa Ali beri ubi kentangitu
woman to who Ali give potato the
'the woman to whom Ali gave the potato'
Here, the interrogativepronounsiapa preceded by the prepositionis used to introduce
the relative clause. Clearly, the use of the interrogativepronoun, which can take
prepositions (i.e. +case), is a differentstrategy from that using the invariableparticle
yang. In fact, this use of the interrogativepronoun was elicited only under pressure,
and was felt to be bookish and clumsy. The preferredalternativewas (16), in which the
original sentence has been reorganizedlexically so that the position relativized is the
subject:
(16) perempuanyang menerimaubi kentangitu daripadaAli
woman that received potato the from Ali
'the woman that received the potato from Ali'
2 Our two informantswere from Malaya. MacDonald and Soenjono (1967) present a
slightly more
restrictedstrategyfor Indonesian.
72 EDWARD L. KEENAN AND BERNARD COMRIE

1.3.3. Subject-Indirect Object. The indirect object position is perhaps the most
subtle one on the AH. For purposes of relative clause formation,it appearsthat many
languages either assimilate indirect objects to the other oblique cases (e.g. English,
Malay)or to direct objects (e.g. Shona, Luganda).Nonetheless, Basque does appearto
discriminateindirectobjects from both its immediateneighborson the AH.
In Basque, subject, direct object, and indirectobject are representedin the verb by
verbal affixes (Lafitte (1962, 193-194)). Relativization on any of these positions is
effected naturallyby deleting the position relativized, putting the restrictingclause in
prenominalposition, and markingthe juncture with an invariable marker -n. Thus,
from (17a) we can generate any of the three relative clauses (17b)-(17d):

(17) a. Gizon-a -k emakume-a -ri liburu-a eman dio.


man the SU woman the 10 book the give has
'The man has given the book to the woman.'
b. emakume-a -ri liburu-a eman dio-n gizon-a
woman the 10 book the give has Rel man the
'the man who has given the book to the woman'
c. gizon-a -k emakume-a -ri eman dio-n liburu-a
man the SU woman the 10 give has Rel book the
'the book that the man has given to the woman'
d. gizon-a -k liburu-a eman dio-n emakume-a
man the SU book the give has Rel woman the
'the woman that the man has given the book to'

However, once we attemptto relativize on positions that are not explicitly coded
in the verb, a variety of difficultiesarise. For speakersof what de Rijk(1972)has called
the restricted dialect, no further relativization is possible. For other speakers a
somewhat greatervariety of positions may be relativized, but often a differentstrategy
is used. For instance, a pronoun may be retained in the position relativized and the
relative clause may occur postnominally. Consequently, for speakers of the restricted
dialect, as well as for certain other speakers, the primarystrategy works only on the
subject, direct object, and indirectobject.
Anotherlanguagein this category is Tamil, a Dravidianlanguageof southernIndia
and Sri Lanka (Ceylon). One RC-forming strategy puts the restricting clause in
prenominal position, with the participial (nonfinite) ending -a on its verb, and no
indication in the restricting clause of the syntactic function of NPrei. This strategy
applies to subjects, direct objects, and indirectobjects:

(18) a. Jan patu-kir -a penmani(y)-ai kan-t -an.


John sing Pres Partwoman DO see Past Sg-3rd-Masc
'John saw the woman who is singing.'
NOUN PHRASE ACCESSIBILITY AND UNIVERSAL GRAMMAR

b. Anta manitanati-tt -a penmani(y)-ai jan kan-t -an.


that man hit Past Partwoman DO John see Past Sg-3rd-Masc
'John saw the woman that that man hit.'
c. Jan puttakatt-ai(k)koti-tt -a penmani(y)-ai nan kan-t -en.
John book DO give Past Partwoman DO I see Past Sg-lst
'I saw the woman to whom John gave the book.'
It does not work on other positions, e.g. instrumentals,where a different RC-forming
strategy is required,retainingNPrel in the restrictingclause:
(19) Enna(k)katti(y)-dl kori(y)-ai anta manitan
which knife with chickenDO that man
kolaippi-tt -an anta katti(y)-ai jan kan-t -an.
kill Past Sg-3rd-Mascthat knife DO John see Past Sg-3rd-Masc
'John saw the knife with which the man killed the chicken' (literally: with
which knife the man killed the chicken, John saw that knife).
Roviana, a Melanesian language spoken in New Georgia, Solomon Islands, also
provides some supportfor discriminatingthe indirectobject position. There, in simplex
sentences, indirect objects are treated like other oblique case NPs in that they are
preceded by a preposition, whereas direct objects are not:
(20) Ele ponia Jone koe Mere sa buka.
Past give John to Marythe book
'John gave the book to Mary.'
(21) Vekoa Jone sa bereti pa tevelo.
put John the bread at table
'John put the bread on (the) table.'
On the other hand, in relative clauses indirect objects are relativizedjust like direct
objects and subjects-a postnominal strategy, in which the case of the relativized
position is not marked, as the relativizationmarkersapu is morphologicallyinvariable
and no pronoun is retained in the position relativized. But in the oblique cases the
function of NPrelis coded, either in the form of a stranded adverb or in the variable
form of the relativizationmarker:
(22) a. sa buka sapu ele ponia Jone koe Mere
the book that Past give John to Mary
'the book that John gave to Mary'
b. sa barikalegesapu ele ponia buka Jone
the woman that Past give book John
'the woman that John gave the book to'
(23) sa tevelo vasina vekonia Jone sa bereti
the table where put (by?) John the bread
'the table where John put the bread'
74 EDWARD L. KEENAN AND BERNARD COMRIE

1.3.4. Subject-Oblique. The primary strategy in Korean (Tagashira(1972)) places


the restrictingclause to the left of the head NP, separatedfrom it by the suffix -()n,
-nin. NPrei is simply deleted for all NPs on the AH down to and including obliques:
(24) hy;nsik -i ki la -lil ttali-n maktaki
Hyensik SU the dog DO beat Rel stick
'the stick with which Hyensik beat the dog'
However, where genitives are relativized, a pronounmust be retained:
(25) chaki-ij la -ka chongmyongha-n ki salam
he of dog SU smart Rel the man
'the man whose dog is smart'
In fact, many languages besides Korean change RC-formingstrategies at the genitive
position by presentinga pronominalelement in the position relativized;see Table 2 in
section 2.2.2.
Furthersupportfor distinguishinggenitives on the AH comes from languageslike
Catalan and North Frisian (Fering dialect), where genitives (and objects of compari-
son) are simply not relativizable at all, although all NPs higher on the AH than
genitives are relativizable,e.g. North Frisian:
(26) a. John kland det wuf's henk.
'John stole the woman's chicken.'
b. det henk wat kland John
'the chicken that John stole'
c. *det wuf wat's henk John kland
'the woman whose chicken John stole'
1.3.5. Subject-Genitive. A greatmany well-knownlanguageshave primarystrategies
that permit relativization on all the positions on the AH except that of objects of
comparison. Thus in French we cannot relativizele jeune homme in (27a):
(27) a. Marie est plus grandeque le jeune homme.
'Marie is biggerthan the young man.'
b. *le jeune homme que que Marie est plus grande
'the young man than whom Marie is bigger'
The same situation holds in Spanish, German,and Romanian(nonparticipialstrategy,
in each case).

1.3.6. Subject-Object of Comparison. Few languages that distinguish objects of


comparison from direct objects or oblique NPs permit them to be relativized. In
English we do have phrases like (28), though some find them ratheruncomfortable:
(28) the man who Mary is taller than
NOUN PHRASE ACCESSIBILITY AND UNIVERSAL GRAMMAR

And in Urhobo, whose primarystrategy is postnominal,and pronoun-retainingfor all


positions on the AH, we have:
(29) oshale na 1- i Maryrho n- o
man the that Marybig thanhim
'the man that Mary is biggerthan'
This completes the argumentthat primaryRC-formingstrategies can discriminate
all the positions on the AH. Thus, the PRC is established, and we may conclude
generally that the Accessibility Hierarchy determines the relative ease of relative
clause formationfrom unmarkedsimplex sentences across languages.
Table 1 (p. 76) summarizesour data concerningthe relativizingpower of the RC-
formingstrategieswe consideredin the languagesin our sample. A key to the entries in
the row-column intersections is given at the end of the table. Needless to say, such a
summarycannot adequatelyrepresentall the language-particular problemsinvolved in
determiningthe nature of an RC-formingstrategy.

1.4. Problems and Possible Counterexamples


We shall consider here two types of difficulties with the analysis we have proposed.
The first concern methodologicalproblemsinvolved in identifyingsubjects and relative
clauses (RCs). The second concern specific counterexamplesto the Hierarchy Con-
straints.
1.4.1. Methodological Problems
1.4.1.1. IdentifyingSubjects. We are using a largely traditionalnotion of subject. An
attempt to make explicit the large number of specific properties that comprise this
notion is given in Keenan (1976b)and will not be discussed here. It is clear from that
investigation, however, that the NPs we call subjects in some languages are more
subject-likethan those of other languages.That is, they possess a greaternumberof the
propertiesthat are characteristicof subjects in general. A languagein which the subject
propertieswere systematicallydistributedacross two or more NPs then mightarguably
be said not to have a single category of subject. In such a case the AH for that
languagewould lack the subject position, much as the AH applied to other languages
may lack an OCOMP position (see 1.2.1), and the predictionsmade by the HCs would
be reduced and furtherprincipleswould have to be found to account for the degree of
relativizabilityof NPs not on the AH.
Fortunately,many languagesdo appearto present NPs that conformfairly well to
the traditionalconcept of subject. Nonetheless, three categories of possibly subjectless
languages have been recently discussed in the literature: Tagalog and Philippine
languagesgenerally (Schachter(1976)), "topic" oriented languageslike Lisu and other
Sino-Tibetan languages (Li and Thompson (1974; 1976)), and ergative languages
(Tchekhoff (1973))(see 1.4.2 below).
76 EDWARD L. KEENAN AND BERNARD COMRIE

Table 1

Language Relativizable positions


Relative clause forming
strategy Subj DObj IObj Obl Gen OComp
Aoban (North-East)
1. postnom, -case + - - - - -
2. postnom, +case - + + + + +
Arabic (Classical)
1. postnom, -case +
2. postnom, +case - + + + + +
Basque
1. prenom, -case + + +
Batak (Toba)
1. postnom, -case +
2. postnom, +case - - + + +
Catalan
1. postnom, -case + + + - - -
2. postnom, +case - - - +
Chinese (spoken Pekingese)
1. prenom, -case + +
2. prenom, +case - + + + + +
Czech (colloquial)
1. postnom, +case + + + + + +?
Dutch
1. postnom, -case + +
2. postnom, +case - - + + +
English
1. postnom, -case + +
2. postnom, +case - - + + + +

Finnish
1. postnom, +case + + + + +
2. prenom, -case + + - - -
French
1. postnom, +case + + + + +
North Frisian (Fering dialect)
1. postnom, -case + + +
2. postnom, +case - - -(?) +
Fulani (Gombe dialect)
1. postnom, -case + + *
2. postnom, +case - - * + +
NOUN PHRASE ACCESSIBILITY AND UNIVERSAL GRAMMAR

Table 1 (Continued)

Language Relativizable positions


Relative clause forming
strategy Subj DObj JObj Obi Gen OComp
Genoese (Zeneyze)
1. postnom, -case + + - - - -
2. postnom, +case - + + + + +
German
1. postnom, +case + + + + +
2. prenom, -case + - - - - -
Gilbertese
1. postnom, -case + - - - - *
2. postnom, +case - + + + + *
Greek (Modern)
1. postnom, -case + + - - - -
2. postnom, +case - - + + + +
Hausa
1. postnom, -case + +
2. postnom, +case - - + + +

Hebrew
1. postnom, -case + + - - -
2. postnom, +case - + + + + +/*
Hindi
1. postnom, +case + + + + + *
2. internal, +case + + + + + *
Iban (Sea Dayak)
1. postnom, -case + - - - - -
2. postnom, +case - - - +/-
Italian
1. postnom, -case + + - - -
2. postnom, +case - - + + +
Japanese
1. prenom, -case + + + +1- +1- -?
2. prenom, +case - - - - +1- -?
Javanese
1. postnom, -case +
2. postnom, +case - - - - +
Kera
1. postnom, -case + - - - - *
2. postnom, +case - + + + + *
78 EDWARD L. KEENAN AND BERNARD COMRIE

Table 1 (Continued)

Language Relativizable positions


Relative clause forming
strategy Subj DObj IObj Obl Gen OComp

Korean
1. prenom,-case + + + +
2. prenom, +case - - - - +
Luganda
1. postnom, +case + + * - - *
Malagasy
1. postnom, -case +
Malay
1. postnom, -case + +/- - - - *
2. postnom, +case - - + + + *
Maori
1. postnom, -case +
Minang-Kabau
1. postnom, -case + - - - - *
2. postnom, +case - - - + + *
Persian
1. postnom, -case + + - - - *
2. postnom, +case - + + + + *
Polish
1. postnom, +case + + + + +
Romanian
1. postnom, +case + + + + + *
Roviana
1. postnom, -case + + +
2. postnom, +case - - - + +
Russian
1. postnom, +case + + + + +
Sinhala
1. prenom, -case + +/+? +? +/-
Shona
1. postnom, -case + +? + - - *
2. postnom, +case - - - + + *
Slovenian
1. postnom, +case + + + + + +
NOUN PHRASE ACCESSIBILITY AND UNIVERSAL GRAMMAR

Table 1 (Continued)

Language Relativizable positions


Relative clause forming
strategy Subj DObj IObj Obl Gen OComp
Spanish
1. postnom, -case + + + - - -
2. postnom, +case - - - + +
Swedish
1. postnom, +case - - + + +? -?
2. postnom, -case + + - - - -
Tagalog
1. postnom, -case +
2. prenom, -case +
Tamil
1. prenom, -case + + + - - *
2. internal, +case + + + + + +
Tongan
1. postnom, -case + + - - -
2. postnom, +case + - + + +
Turkish
1. prenom, -case + + + +
2. prenom, +case - - - - + +?
Urhobo
1. postnom, +case + + + + + +
Welsh
1. postnom, -case + +
2. postnom, +case - - + + + +
Yoruba
1. postnom, -case + + * * - *
2. postnom, +case - - * * + *
Zurich German
1. postnom, -case + +
2. postnom, +case - - + + +??
Key; + means that the strategygenerallyappliesto that NP position; - meansthat it does not. +? meansit
applies, but with loss of acceptabilityand perhapssome informantdisagreement.-? meansit does not apply,
althoughthe resultis notjudged too bad by informants.An entry of the formx/y meansthat for certainof the
NPs in that positionx is the correctentry, and for other such NPs y is the correctentry.x(?) means that our
data are not entirelyunequivocal,but our best judgmentis thatx is the appropriateentry. * means that that
NP position does not exist as such, but ratheris treatedas some other position: e.g. objects of comparison
are often either direct objects of verbs like exceed as in Shona or oblique case NPs governed by pre- or
postpositionsas in Hebrew and Hindi. A blankmeans that we lack the relevantdata.
80 EDWARD L. KEENAN AND BERNARD COMRIE

Of these categories, Schachter's claim that the "focus" NP in Tagalog cannot be


regarded as a subject is the most damaging to our claim, since Tagalog was one of the
languages that justified HC1 and HC3 (a language may relativize only subjects).
However, it has been shown in Keenan (1976a) that others of the Western Malayo-
Polynesian languages, notably Malagasy, do not present the evidence that supports
Schachter's claims for Tagalog, and consequently there is still sufficient data to support
HC1.
The evidence Li and Thompson present from Sino-Tibetan is less damaging, since
the NPs that are most subject-like do not present any Hierarchy violations. What they
argue, however, is that the syntactic category "subject" has a very low functional load
in these languages in that few if any syntactic processes need to be made sensitive to
the NP that is the subject. Further research in those languages might then reveal that
categories other than those on the AH, e.g. topic, are relevant in determining RC-
forming possibilities.
With the possible exception of Dyirbal (and perhaps Eskimo; see Woodbury
(1975)) (see section 1.4.2.1), the claim that ergative languages lack subjects has been
refuted by Anderson (1976), who shows that with respect to many major syntactic
processes transitive subjects (the ergative NP) and intransitive subjects behave alike, in
distinction to DOs, thus establishing the fact that there is a category of subject in those
languages.
1.4.1.2. Identifying Relative Clauses. We have been considering an RC to be any
syntactic structure that designates an object (or set of objects) in a certain way,
namely, by first specifying a larger domain of objects and then restricting it to a subset,
perhaps a one-member subset, of which a certain sentence, the restricting sentence, is
true. However, many languages present sentence types that appear to designate objects
in this way but in which there is no surface constituent with which we can associate the
designating properties. One such example concerns extraposed RCs, as in (30a) below:
(30) a. The student finally arrived who we had been waiting all morning for.
b. the student who we had been waiting all morning for
Clearly, to evaluate the truth of (30a) it is necessary to determine that the main
predicate holds of an object that is, first, a student, and second, is such that the
sentence we had been waiting all morning for him is true of him. That is, the object in
question is clearly the one designated by the RC in (30b), and in this sense the logical
structure of (30a) contains the RC in (30b). But (30a) itself does not present in surface a
constituent that has the referential properties of (30b) and so does not contain an RC.
In this case, however, it is easy to argue that at a fairly shallow level of underlying
structure (30b) does occur in the syntactic structure of (30a), and that a late rule of RC
Extraposition moves the restricting clause around the main verb. Thus the syntactic
structure of (30a), considered as the sequence of phrase markers representing its
NOUN PHRASE ACCESSIBILITY AND UNIVERSAL GRAMMAR

derivation,does contain an RC, and it is clear, then, that our definitionof RC does not
requirethat, in surface, the head NP and the restrictingclause be a constituent.
It is clear, furthermore,that the most ordinarytype of RC in English is expressed
by structures like (30b), those in (30a) being more marked and of more restricted
distribution. But in many languages the usual translations of English sentences
containing RCs do not present the head NP and the restrictingclause as a constituent.
(31) from Hindi and (32) from Walbiri(Hale (n.d.)) illustratethis type:
(31) Mai us aurat ko janta hoonjis- ko Ramne kitabdiya.
I that woman DO know who IO Ram Erg book gave
'I know the woman that Ram gave the book to.'
(32) ijatjulu-Ju 'k-payankiripantu-nu, kutja-lpa tapa ia- nu.
I Erg Aux emu spear Past CompAux water drink Past
'I speared the emu that was drinkingwater/whileit was drinkingwater.'
In these cases it is not clear that what correspondsto the restrictingclause in English
ever occurs as a constituentwith the head NP in underlyingstructure.Both Hale (n.d.)
and Andrews (1975) argue that it does not.
However, languageswith this type of constructionvery regularlyexhibit a related
type of constructionin which there is a constituentthat meets our semantic conditions
of relative clause-hood. The correspondingversions of the Hindi ((31)) and Walbiri
((32)) are given below in (33) and (34), respectively:

(33) [Ram ne jis aurat ko kitab diya] us (aurat) ko maijanta hoon.


Ram Erg which woman IO book gave that woman DO I know
'I know the woman that Ram gave the book to.'
(34) [yankiri-li kutja-lpa ijapa Da- nu] ula pantu-nu ijatjulu-lu.
emu Erg CompAux water drinkpast that-one spear Past I Erg
'I speared the emu that was drinkingwater.'

It is not clear that the bracketed constituents above should be considered NPs.
Certainlythey present the normalsyntax of full sentences, except that they contain the
markersjis and kutja. Whatever the grammatical category of these constituents,
however, it does appearthat they meet our definitionof RC-they specify a domainof
objects, those marked by jis and kutja, and they give the restrictingclause, the one
determinedby the entire clause in which they occur. Thus, like the examples (3) from
Bambara and (4) from Diguenlo mentioned earlier, they illustrate an RC-forming
strategy in which the head NP occurs within the restrictingclause. Our definition of
RC then does not requirethat an RC be an NP, nor does it requirethat the head NP
commandthe restrictingclause (a use of "head" that differs from the usual one in the
linguisticliterature).
The RC-formingstrategiesillustratedabove in Walbiriare perhapsresponsiblefor
the claim sometimes heard that, in violation of HC1 (which states that subjects are
82 EDWARD L. KEENAN AND BERNARD COMRIE

universally relativizable), Australian languages do not have RCs. Walbiri at least


clearly does not present embedded RCs in which the head NP commands the
restrictingclause and in which the two form an NP constituent. But it certainly does
present constituents that meet our semantically-basedcriteriafor what constitutes an
RC, so HC1 is not violated here.
1.4.2. Some Possible Counterexamples
1.4.2.1. Possible Counterevidenceto HC1. We consider first the best documented
counterexampleto HC1(subjectsare always relativizable).The counterevidencecomes
from Dyirbal(Dixon (1969; 1972)),an Australianlanguage. Dyirbal appearsergative in
the standardsense that full NP subjects of intransitivesentences and full NP DOs of
transitive sentences are case-marked in the same way (zero, in this case), whereas
subjects of transitive sentences carry a special marker,the ergative. In the analysis to
follow we shall refer to intransitive subjects and transitive DOs collectively as
absolutives. Examples are from Dixon (1972, 100-101).
(35) bayi yara banagajiu.
Det-Abs man-Abs return
'The man is returning.'
(36) bayi yuri baijgul yataggu bagan.
Det-Abs kangaroo-AbsDet-Erg man-Ergspear
'The man speared a kangaroo.'
Dixon demonstrates that absolutives can be relativized from both transitive and
intransitive sentences, but the ergative NP cannot be relativized. To talk about the
man who speared a kangaroo, it is necessary to apply a transformationthat promotes
the ergative NP to absolutive status (37a), whence it can be relativized(37b):
(37) a. bayi yara bagalUajiu bagul yurigu.
Det-Abs man-Abs spear-'"Pass" Det-Instr kangaroo-Instr
'The man speared the kangaroo.'
b. [Npbayi yara bagal-rja -iju bagul
Det-Abs man-Abs spear "Pass" Rel Det-Instr
yurigu] banagajiu.
kangaroo-Instrreturn
'The man who speared the kangaroois returning.'
On the basis of the Dyirbal data, Johnson (1974a) has suggested that the upper
segment of the AH is inapplicableto ergative languagesgenerally, and that an Ergative
Hierarchy (EH), ABS > ERG > 10 > OBL, etc. more adequately describes the
accessibility to RC formationin those languages. The EH differs significantlyfrom the
AH only in that it predicts the possibility of RC-formingstrategies that relativize only
absolutives, such as in Dyirbal. But general support from ergative languagesfor this
prediction is lacking. The majority of ergative languages known to us permit both
NOUN PHRASE ACCESSIBILITY AND UNIVERSAL GRAMMAR

ergative and absolutive NPs to relativize. We have already illustratedrelativizationon


ergatives in Basque (17b) and Walbiri (34). Examples (38)-(42) below illustrate
relativizationon ergatives from a few other languages.
Hindi
(38) Ram us larkiko pasandkartahe,jis- ne kapre dhoye.
Ram that girl DO likes Rel Erg clothes washed
'Ram likes the girl who washed clothes.'
Tongan (Anderson (1975))
(39) te mo fetaulakimo e tangata'oku ne fua 'a e sioki vai.
Fut 2 Du meet with Art man Pres 3Sg carry Part Artjug water
'You will meet a man who is carryinga jug of water.'
Gahuku(Deibler (1973))
(40) izegipa get -a ve
child begot 3Sg man
'the man who had begotten the child'
Jacaltec (Craig (1974))
(41) x- 0- w- 'il [naj x- 0- watx'e-n hun ti']
Asp him-Abs I-Erg see man Asp him-Abs make Rel one this
'I saw the man who made this.'
Greenlandic Eskimo (Woodbury (1975))
(42) qimmi-O -a tuquk-kiga nalu Nil- saNa.
dog Abs-Sg Poss-3Sg kill TransignorantNeg Indic-3Sg-lSg
'He knows about me, who killed his dog.'
Woodbury(1975) does argue, however, that absolutives are more relativizablein
Greenlandicthan are ergatives, on the grounds that (1) RCs formed on ergatives are
somewhat more restricted in the distributionin matrix clauses (p. 21) than are those
formed on absolutives, and (2) for certain verb classes ergatives cannot be relativized
out of the active participle(p. 27). To relativize that NP from an active participle,the
ergative must first be promotedto an absolutive via Antipassive. Nonetheless, it is the
case that ergatives can in general be relativized, even if not quite so freely as
absolutives, and so the AH does appearto apply, althoughmore work would need to
be done to distinguishaccordingto our criteriawhat the differentRC-formingstrategies
in Greenlandicare.
The general claim, then, that in ergative languagesabsolutives are more relativiza-
ble than ergatives receives little support. This in turn suggests that the Dyirbal data
might be analyzed differently. One reanalysis that would have the effect of making
Dyirbal conform to the AH would be simply to regardthe absolutive NP in transitive
sentences (as well as in intransitiveones) as the subject. Such an analysis again calls
into question the definingcriteriafor subjecthoodand would requiremuch independent
84 EDWARD L. KEENAN AND BERNARD COMRIE

support. Perhapssurprisingly,however, Dyirbaldoes provide such support. Below we


summarizethis evidence, citing page numbers from Dixon (1972) for the supporting
data.
1. Absolutives are the most essential NP in the sentence. Thus an unspecified
agent (ergative) may simply be eliminated from a sentence, much like unspecified
agents of passive sentences in English. But unspecified absolutives cannot be so
eliminated. If we want to eliminate an unspecified patient, we must first "passivize"
the sentence so that the ergative becomes a surface absolutive and the former
absolutive becomes oblique, whence it can be deleted when nonspecific (p. 70).
2. Absolutives are the only target of advancement rules. Dyirbal has several
rules, rather like Passive in English, that promote e.g. ergatives and instrumentalsto
the absolutive position of underlyingtransitive sentences, as we illustratedin (37a).
3. Absolutives normallyprecede ergatives in simplex sentences (p. 291).
4. Only absolutives can be coreferentiallydeleted by operations like Equi and
ConjunctionReduction (pp. 73, 67). To coreferentiallydelete an ergative NP, it must
first be advanced to an absolutive.
5. It appears that certain demonstratives, always definite and presupposing
reference-a characteristicpropertyof subjects-are restrictedto absolutive positions
(p. 218).
6. In the few cases where we can tell, absolutives appear to control reflexives.
Usually reflexivizationis done by verbalaffixing, so the derived verb is intransitiveand
the subject is absolutive on independentgrounds. However, in a few examples the
object is an inalienablebody part and is expressed in surface structureeven thoughthe
verb is reflexivized. And in these cases the controlleris absolutive (p. 153):
(43) bayi yara mala da1gaymarijiu.
Det-Abs man-Abs hand-Abseat-Refl-Pres/Past
'The man chewed his finger.'
If we adopt the analysis of Dyirbalin which absolutives in transitivesentences are
the subjects, then Dyirbal will be typologically anomalous in only one majorrespect:
namely, in the least markedtype of transitive sentence the NP with the referentialor
topic propertiesof subjects does not express the agent. And since the predominanceof
topic propertiesover agency propertiesin this case is quite large (very few transforma-
tions mention ergatives, but very many mention absolutives), it turns out that the most
subject-likeNP in basic transitivesentences is the absolutive and does not express the
agent.
This analysis at least has the (metalinguistic)advantage of isolating at a single
point the anomaly of Dyirbal-namely, transitive subjects in basic sentences are not
agents. The ergative analysis on the other hand makes Dyirbal anomalous in a great
many respects: the word order is OSV; DOs are the most essential constituent of
transitive sentences, subjects being freely deletable when unspecified; etc. Further-
NOUN PHRASE ACCESSIBILITY AND UNIVERSAL GRAMMAR

more, it is possible that when the lexicons of more languages have been analyzed in
detail, it will be seen that many languageshave classes of verbs that select goal subjects
but allow agents present in oblique cases. Biggs (1974) has recently argued, for
instance, that Fijian has one large class of verbs that select goal rather than agent
subjects, whereas another class selects agent subjects, as is more usual. Perhaps
Dyirbal will then be seen to be merely near the end of a continuumin that most of its
transitive verbs select nonagentsubjects.

1.4.2.2. ApparentExceptions to HC2. A second type of counterexamplewe consider


are apparent exceptions to HC2, that a given RC-formingstrategy must apply to a
continuous segment of the AH. Hausa (Abrahams (1959); Schachter (1973)) and
Yorubaappearto use RC strategiesthat apply to discontinuoussegments of the RC.
Thus Hausa seems to use a +case strategy to relativize subjects, since a personal
pronoun seems present in the position relativized:

(44) dokin da ya mutu


horse Rel it died
'the horse that died'

Relativizationon DOs, however, does not permitretentionof a pronounin the position


relativized, and as no other nominal particle is present to code the case of the
relativized NP (the relativizerda is invariable), RC formationon DOs appears to be
effected by a -case strategy:

(45) mutumin da na gani (*shi)


person Rel I saw him
'the person that I saw'

However, when we relativize on OBLs in Hausa, a pronoun is again present. (In


relativizingon 1Os, the presence of a pronounis optional.)

(46) yaron da suka gaya {wa/masa}


child Rel they said to to-him
'the child whom they told'
(47) wuqad da ya kashe ta da ita
knife Rel he killed her with it
'the knife with which he killed her'

It appears, then, that the postnominal, +case strategy in Hausa does not apply to a
continuous segment of the AH, in violation of HC2.
On examining further the pronoun that appears in the RC when a subject is
relativized, however, we note that this same pronoun is also required in simplex
86 EDWARD L. KEENAN AND BERNARD COMRIE

sentences with a subject NP:


(48) a. Yusufu ya zo.
Joseph he came
'Joseph came.'
b. *Yusufu zo.
Quite generally in Hausa, full subject NPs must be accompaniedby clitic pronouns.
This suggests a different line of analysis for such pronouns. Instead of regardingthem
as constituent parts of a +case strategy, we may regardthem as an instance of verb
agreement. In many languagesverb-agreementaffixes are known to derive from (clitic)
pronouns(Givon (1976)), and it is often difficultto draw a precise dividing-linebetween
clitic pronouns and agreement affixes. We are therefore suggesting that where the
presence of a pronoun is requiredby the presence of a full NP, then the pronounbe
regardedas an instance of verb agreement,and not as an instance of NP case marking.
In Hausa, such "internal pronouns" (for this terminology, see Keenan (1972a, 447-
450)) occur only with subjects; with DOs, etc., we have (49a) and (49c) for instance,
but not (49b) and (49d):
(49) a. Na gani mutumin.
b. *Na gani shi mutumin.
'I saw the person.'
c. Ya kashe ta da wuqad.
d. *Ya kashe ta da ita da wuqad.
'He killed her with the knife.'
Another similar possible counterexample,from Tongan (pointed out in Anderson
(1975)), is not so easily disposed of, however. Here it appearsthat pronounsmay, and
in some cases must, be retainedwhen subjects are relativized. (39) already illustrates
this. Similarly, relativizationon 1Os, OBLs, and GENs obligatorilyleaves behind a
pronominaltrace (examples from Anderson (1975), Chung(personal communication)):
(50) ko e 'eiki eni na'e langa mo'ona 'a e fale lahi.
Part Art chief this Past build for-3Sg Part Art house big
'This is the chief for whom the big house was built.'
(51) fakaha mai kiata au 'a e tamasi'i na'e
show towards to-Pers me Abs Art child Past
ngalo 'ene polosi fulunifo.
disappearhis brush toothbrush
'Show me the boy whose toothbrushdisappeared.'
However, it is not possible to present pronominaltraces when DOs are relativized:
(52) 'oku 'ikai 'ilo 'e ha taha 'a e tangatana'a ku taa'i (*ia).
Pres not know Erg any one Abs Art man Past I hit him
'Nobody knows the man who I hit.'
NOUN PHRASE ACCESSIBILITY AND UNIVERSAL GRAMMAR

Furthermore,the type of explanationused above for Hausa and Yorubadoes not


in general seem applicable. The subjectpronounspresent when subjects are relativized
do not in general function as agreementparticles, althoughthe pronounpresent in the
RC is the clitic one, not the independentone. Thus it occurs preverbally,whereas the
independentpronounsand full NPs functioningas subjects normallyoccur postverbally
(Tongan is VSO).
(53) na'e 'ave 'e Sione ho'o telefone.
Past take Erg John your telephone
'John took away your telephone.'
(54) *na'a ne 'ave 'e Sione ho'o telefone.
Past he take Erg John your telephone
Furthermore,the clitic pronouns can be present when the subject is an independent
pronoun, althoughthe meaninghere is emphatic:
(55) na'a ne taa'i 'e ia pe ia.
Past he hit Erg he Emph him
'He hit himself.'
This use of the clitic could conceivably develop so that the clitic would cooccur with
full NPs in subject position without the emphatic meaning. This at least would be a
normalway for an agreementto arise (see Givon (1976)for justificationof this claim).
But this would be to anticipate the development of Tongan and cannot be used to
justify its current status as a counterexampleto HC2.
Perhapsthe best explanationwe can offer for the patternof RC-formingstrategies
in Tonganis a historicalone. As we have mentioned, Tonganis an ergativelanguagein
which the ergative marker 'e is cognate with the passive agent marker in related
Polynesian languages. Hohepa (1969) has presented evidence that the ergative para-
digm present in Tongan and the closely related Niuean has evolved from an original
nominative-accusative paradigmin which the productive distinction between active
and passive verb forms was lost and the passive morphology retained on the major
NPs. Thus, historicallyspeaking,the ergative subjectof Tonganis a passive agent, and
the absolutive DO in Tongan is historically the subject. We suggest here that the
evolution of the agent NP to subject status in Tongan is not quite complete. While
Anderson (1976) has shown that it does possess many general propertiesof subjects, it
may retain a few traces of its originaloblique case status. For example, in simple active
sentences the ergative NP can be omitted when unspecified, like the ergative NP in
Dyirbal and the passive agent in English, as illustratedin (56) and (57) below (both
from Churchward(1953)):
(56) na'e tamate'i 'e Tevita 'a Koliate.
Past kill Erg David Abs Goliath
'David killed Goliath.'
88 EDWARD L. KEENAN AND BERNARD COMRIE

(57) na'e tamate'i 'a Koliate.


Past kill Abs Goliath
'Goliath was killed.'
Now, since OBLs in general in Tongan requireretentionof a clitic pronounwhen
relativized (in one case a nonclitic pronoun can be retained, but this is quite
exceptional), the ergative NP in Tongan is behaving like the other OBLs in this
respect. On the other hand, the DO does not allow a pronoun to be retained under
relativization. But, as we show in 2.2.2, this is characteristicof subjects in general.
That is, even in languages like Hebrew in which pronoun retention is normal under
relativization,subject pronouns are normallynot retained. So, in this way, the DO in
Tongan is behaving like a subject. In these very minor respects, then, perhaps the
ergative and absolutive in Tongan betray their earlier status as passive agent and
subject, respectively.

2. Towards an Explanation of the Hierarchy Constraints


2.1. The AH as a Psychologically Valid Entity
Given that the HCs do make correctpredictionsabout RC formationin a wide rangeof
languages, it is natural to wonder why this should be so. We propose the following
explanation: The AH directly reflects the psychological ease of comprehension. That
is, the lower a position is on the AH, the harderit is to understandRCs formedon that
position.
If the AH does reflect the psychological accessibility of NPs to RC formation,
then we can use this fact, in conjunctionwith certain other assumptions,to explain the
HCs in the following way. First, it would be naturalthat a way of relativizinga certain
position might not be applicable at the next lower position (HC3) on the general
assumption that syntactic processes are ways of encoding meanings; and, if one
meaning is inherently more difficult to encode than another, then a strategy for
encoding the first need not apply to the second. By the same token, a strategy that
applies to one position but fails to apply to the next lower position would not be
expected to apply to a still lower position (HC2). For, if a given strategy is used to
encode a fairly easy meaningand that strategy is "strong" enough to encode a rather
difficult meaning, then it is surely strong enough to encode the meaningsof intermedi-
ate difficulty. However, this ratherinformalline of reasoningcan be slightly extended
to yield an argumentthat languagesshould not have "gapped" RC-formingstrategies.
Thus, if speakers can in general encode a meaningof a certain difficulty in some way,
then they should certainly be able to encode meaningsof lesser difficulty(possibly in a
differentway). Yet we have seen that Toba Batak, for example, can relativize directly
subjects using one strategy, and 1Os, OBLs, and GENs using another. But DOs
cannot be directly relativized using any strategy. As we mentioned, however, in all
cases in our sample in which languagespresent gapped strategies,the NPs that occur in
NOUN PHRASE ACCESSIBILITY AND UNIVERSAL GRAMMAR

the gap can always be promoted to positions that can be directly relativized via
operationslike Passive. In these cases, then, we would like some independentevidence
that it is psychologically easier, in those languages, to for instance promote a DO to
subject and then relativize it rather than to relativize it as a DO. We have no direct
psychological evidence of the right sort. It is worth noting, however, that in those
languagesthat promote an NP to a higherposition on the AH in order to relativize it,
the promotionsystem (e.g. the ways of converting NPs low on the AH to ones higher
on the AH) is usually well developed, very commonly used, and has a wide syntactic
distribution,in distinctionfor instance to Passive in English, which is, by comparison,
a less usual, more markedform. In fact, in Hawkins and Keenan (1974) it is reported
that on certain types of repetition tasks, English-speakingchildren (10-12 years) do
significantlyless well on RCs like the boy who was seen by Mary than they do on ones
like the boy who Mary saw. Thus, promotingto relativize is certainly not universally
easier than relativizingdirectly.
Finally, note that this psychologicalinterpretationof the AH cannot fully account
for HC1, that subjects must, in general, be relativizable. It only justifies the claim that
subjects are easier to relativize than any other position on the AH, but it would allow
in principlethat, in some language,no position on the AH be relativizable.

2.1. Evidence in Support of the Psychological Validity of the Accessibility Hierarchy


Several recent experimentalstudies do provide partial support for the claim that the
AH does represent the psychological accessibility to RC formation. Thus, in Legum
(1975) it is shown that English-speakingchildren, aged 6-8, comprehend RCs formed
on subjects better than ones formed on DOs. Similarresults were achieved by Brown
(1971) for children aged 3-5 years. Hatch (1971) shows that young children respond
more rapidlyto RCs formed on subjects than on objects, and Cook (forthcoming)has
shown that children and adults recognize with fewer errors instances of head NPs
bearingthe subject relation to the subordinateverb in RCs than instances of the head
bearing the object relation to the verb. And Valli et al. (1972), in a productiontest,
show that French children at the level of 61 (approx. 12 years of age) produce RCs
formed on subjects much more readilythan on objects, and RCs on possessor NPs are
much rarerstill and in the majorityof instances incorrectlyformed.
It should be noted that the purposeof these studies was not specificallyto validate
the AH. Variables other than the position on the AH of the NP relativized were in
general shown to be relevant (e.g. whether the RC was embeddedor not, whether the
function of the RC in the matrixwas the same as that of the head in the RC (Sheldon
(1974));see Legum (1975)for some discussion of these other parameters).
On the other hand, Hawkins and Keenan (1974) report a study specifically
designed to test the psychologicalvalidity of the AH. Children,aged 10-12, were given
repetition tests involving RCs formed on all the positions on the AH. Correctnessof
recall correlated significantly (<.05) with the AH. In fact, the recall hierarchy
90 EDWARD L. KEENAN AND BERNARD COMRIE

established in that study was:

SU > DO > IO > OBL, OCOMP > GEN


Thus, RCs formed on subjects were recalled with fewer errors than those formed on
DOs, which were better than ones formed on 1Os, which were better than those
formed on either OBLs or OCOMPs, which were better than ones formed on
possessor NPs. Clearly the only position out of place on the recall hierarchy is the
OCOMP position, which was treatedby the childrenas the same in accessibilityas the
OBL position. All RCs on OBLs left the prepositionstranded,e.g. the boy who Johnny
took the toy from. Apparently strandingthe comparative particle, e.g. the boy who
Johnny is taller than, was interpretedas being similarto prepositionstranding.
While it would certainly be premature, on the basis of these few studies, to
conclude that the AH does establish a hierarchyof psychological accessibility, it does
appear that the available evidence points in that direction. If further research along
these lines justifies the psychological interpretationof the AH, then we will have an
explanationfor two furtherfacts from our own study. Alternatively,these facts can be
considered further support for the psychological interpretationof the AH. The first
concerns a pattern of judgments of relative accessibility of RCs from languages in
which most positions on the AH can be grammatically relativized. The second
concerns the distributionof personalpronounsretainedin the RC.

2.2.1. Intralanguage Relative Accessibility Judgments. One natural way to extend the
HCs would be to interpretthem not only as a cross-languageorderingof grammatical-
ity, as we have done, but to consider them as an acceptability ordering within each
language. Thus one might expect that in general, within a given language, RCs formed
on the high end of the AH would be judged more acceptable than RCs formed on the
low end. And in the extreme cases this appearsto be correct.
RCs formed on subjects are always among the most acceptable in any given
language, and those formed on objects of comparison, where possible at all, are often
judged to be only marginallyacceptable. For instance, many informantsin English are
uncomfortable with the man who Mary is taller than. Similar judgments of relative
acceptability obtained for our Hebrew informants.Thus, even in languages in which
objects of comparisoncan be relativized, there will be a preference to express the RC
as one formed from a semanticallyequivalent sentence in which the semantic object of
comparisonis presented as a subject. That is, the preferredway to express the idea of
the man who Mary is taller than will be, in general, the man who is shorter than Mary.
A somewhat more subtle performancepreference is illustrated by the genitive
position on the AH. Although a majorityof languages in our sample possessed some
way of relativizinggenitives, there was often some awkwardnessin doing so, and not
infrequently,in specific cases, a preferredalternativewas offered-one that relativizes
NOUN PHRASE ACCESSIBILITY AND UNIVERSAL GRAMMAR

a position higher on the AH. For example, (58) is perfectly grammaticalin French:
(58) la femme dont le manteaua ete vole
the woman whose the coat has been stolen
'the woman whose coat was stolen'
But in practice people will tend to say:
(59) la femme qui s' est fait voler le manteau
the woman who herself is caused to-steal the coat
(59) is literally 'the woman who got her coat stolen', although there is no necessary
implicationof her having engineered the theft; relativizationis on the subject, rather
than the genitive. The translationsabove indicate that a similaralternativeto genitive
relativization exists in English. A like situation holds in Swedish, where the (b)
alternativein each pair is preferred:
(60) a. kvinnan,vars kappablev stulen
woman whose coat was stolen
b. kvinnan, som fick sin kappastulen
woman who got her coat stolen
(61) a. kvinnan, vars man ar p'asjukhuset
woman whose husbandis in hospital
b. kvinnan, som har sin man pa sjukhuset
woman who has her husbandin hospital
In Yoruba, inalienable possessives/genitives can often be paraphrasedby other
constructions, and even with simple sentences these paraphrasesare preferred:

(62) a. John lu ts Qkunrinnaa.


John strike leg man the
'John struck the man's leg.'
b. John lu okunrinnaa 1' se.
John strike man the on leg
'John struck the man on his leg.'

Under relativization,the preferenceis even more strongly marked,and our informant


was very hesitant about admitting(63a) below:
(63) a. ?*Mo ri okunrinti John lu esq ri.
I see man that John strike leg his
'I see the man whose leg John struck.'
b. Mo ri Qkunrinti John lu 1' ese.
I see man that John strike on leg
'I see the man that John struck on the leg.'
92 EDWARD L. KEENAN AND BERNARD COMRIE

Bamgbose (1966, 158-159) claims that the relation between sentences like (63a) and
(63b) is transformational.
2.2.2. Pronoun Retention. A less obvious patterningdeterminedin part by the HCs
concerns the distributionof personal pronouns in relativized positions. (We exclude
pronouns that are instances of verb agreement;see section 1.4.2.2). We have already
noted that Semitic languages"characteristically"present such pronouns, e.g. Hebrew:
(64) ha- isha she- David natanla et ha- sefer
the woman that David gave to-her DO the book
'the woman that David gave the book to'
Further, it has been argued in Keenan (1972a; 1975a)that such RC-formingstrategies
present in surface structuremore of the logical structureof the RC than do languages
like English that do not present such pronouns. The reason, in brief, is that in the
pronoun-retainingstrategies the restrictingclause in surface is a sentence-one that
expresses exactly the restricting sentence of logical structure. That is, it is just the
sentence that must be true of the referent of the RC. So a possible referent of (64)
above must be a woman of whom the sentence David gave the book to her is true. And
the translationof that sentence in Hebrew is precisely David natan la et ha-sefer-the
surface restrictingclause. Note that the correspondingclause in English, David gave
the book to or to whom David gave the book, is not a surface structure sentence, and
so is not immediatelyperceived as the sort of linguisticentity that is true of objects.
It has been shown elsewhere (Keenan (1972a, and especially 1975a))that the RC-
forming strategies that retain pronouns are applicableto a greater range of otherwise
"difficult" environments(e.g. it is often, but not always, possible in these languagesto
relativize into coordinate NPs, other relative clauses, indirect questions, and even
sentence complementsof NPs). The reason is that the logically more explicit strategies
still successfully express the basic meaningof the RC in contexts where the meaningis
otherwise difficultto perceive. Consequently,we are led to predict that, as we descend
the AH, languageswill exhibit a greatertendency to use pronoun-retainingRC-forming
strategies. Table 2 (p. 93) dramaticallysupportsthis prediction. Languagesthat do not
normallyretainpronounsin any position are not included in the table; nor is the use of
pronouns as markersof verb agreement(see 1.4.2.2).
It should be clear from Table 2 that not only does the tendency to present
pronouns in positions relativizedincrease as we descend the AH, but also that once a
languagebegins to retainpronounsit must do so for as long as relativizationis possible
at all. This is a naturalconsequence of the hypothesis that pronoun retentionwill be
used in proportionto the difficultyof the position being relativized, though the critical
point of difficulty is differentfor differentlanguages.

2.2.3. Explaining the Psychological Validity of the AH. To some extent, explaining
the HCs by interpretingthe AH as a psychological hierarchymerely pushes back the
NOUN PHRASE ACCESSIBILITY AND UNIVERSAL GRAMMAR

Table 2
Pattern of Pronoun Retention in Relative Clauses
Language Subj DO 10 Obl Gen OComp
Aoban (North-East) (+) + + + + +
Arabic - + + + + +
Batak - 0 + + +
Chinese (Pekingese) - +/- + + + +
Czech (colloquial) - +/- + + + +
Fulani (Gombe) - - * + +
Genoese - (+) + + +
Gilbertese - + + + + +
Greek (Modern) - - +(?) +(?) + +
Hausa - - (+) + +
Hebrew - + + + + +
Japanese - - - - +/-
Javanese - - - - +
Kera - + + + +
Korean - - - - + 0
Malay - - - - + *
Minang-Kabau - - - -/+ + *
Persian - (+) + + + +
Roviana - - - - + 0
Shona - - - (+) +(?) *
Slovenian - + + + + +
Turkish - - - - + +
Urhobo + + + + + +
Welsh - - + + + +
Yoruba - - * * +
Zurich German - - + + + +
Key: + means that personalpronounsare normallypresent in that position when it is relativized,using that
RC-formingstrategy which admits of pronounretention. (+) means optional retention. +/- means that in
some cases the pronounis retainedand in others it is not. - means that pronounsare usuallynot retained.*
means that that NP position does not naturallyexist in that language. 0 means that that position is not
relativizable,and a blankmeansthat we lack the relevantdata. An entry of the formx(?) means that our data
are uncertainbut x is our best guess.

problemof explanationone step. We would still want to know why it is psychologically


easier to relativize subjects than objects, etc. We have two speculationsto make here.
2.2.3.1. A Recognition Strategy. Impressionistically,the initial portion of the AH
appears to coincide with the degree to which NPs are required to appear in simple
sentences. Thus, lexical predicates almost always require a subject. Many require
94 EDWARD L. KEENAN AND BERNARD COMRIE

DOs, some require 1Os, and a few (e.g. put) require OBLs. Furthermore,a few verbs
(R. Stockwell, P. Schachter, personal communication)such as gnash (one's teeth),
blink (one's eyes), and water (his eyes watered) appear to require that their arguments
be possessed body parts, similarto the more idiomaticconstructionslose one's nerve,
blow one's cool, etc. Finally, no lexical predicate in English requires that it be
construed in a comparativeconstruction.
Perhaps, then, there is a kind of universal recognitionstrategy of the sort: "If an
NP plays a role in another clause, interpretit as a subject unless there are indications
to the contrary, otherwise try the DO slot, etc." Needless to say, such a general
recognition strategy would require a more precise formation and much experimental
research before it could constitute a serious explanationof the AH.
2.2.3.2. Independent Reference. Another explanation, which pertains only to the
relative accessibility of subjects over other NPs, was offered in Keenan (1974). There it
was arguedthat heads of RCs share a logical propertywith subjects of sentences but do
not share this property with nonsubjects. Thus, more of what we need to know to
understand the meaning of a RC formed on subjects is already contained in the
meaningof simple sentences than is the case when the RC is formed on a nonsubject.
In more "transformational"terms, RC formationon subjects distorts the meaningof
the underlyingsentence less than RCs formed on objects.
The logical propertysharedby heads and subjects is that of independentreference.
Thus, in simple sentences we cannot generally make the reference of the subject NP
dependenton that of some other NP in the sentence. For instance, if subjectand object
are markedas coreferential,it must be the object which is marked(if anythingis), by
for instance a reflexive pronoun. Thus, the reference of object phrases can be made
dependent on that of subjects, but not conversely. But it is in the inherent nature of
subjects to be independentlyreferring.(For a weakening,but not abandonment,of this
principlefor more complex cases, see Keenan (1974).)
Similarly,to understandthe meaningof an RC such as the girl that John likes, we
must be able to understandwhat set is designatedby the head NP independentlyof the
reference of the NPs in the restrictingclause. For instance, the head NP can never be
coreferentiallypronominalizedby an NP in the restrictingclause, even if, as in many
languages, the restricting clause precedes the head. But of course NPs within the
restrictingclause can be stipulatedas, for instance, being identical in reference to the
head, as in (1), repeated below:
(65) der in seinemi Buro arbeitendeManni
the in his study working man
'the man who is workingin his study'
Consequently, if we relativize on a nonsubject,the resultingstructurewill contain two
necessarily independentlyreferringexpressions, the head of the RC and the subject of
the restrictingclause. This explains, for instance, why the pronounin the man who he
NOUN PHRASE ACCESSIBILITY AND UNIVERSAL GRAMMAR

hit cannot be understood as being necessarily coreferentialwith the head.3 But if we


relativize on a subject, there is only one necessarily independentlyreferringexpression.
In this sense, then, subject relatives are psychologically simpler than nonsubject
relatives.

3. FurtherProspects
In this article, we have restrictedourselves by and large to the particulardata area that
originally led us to propose the Accessibility Hierarchy, namely, restrictions on
relative clause formation.We wish here to indicatebrieflytwo other areas where recent
work has shown the possible relevance of the same Accessibility Hierarchy.
In Comrie(in press), it is shown that the AH is useful in the syntactic description
of causative constructions, in particularsynthetic causative formations, in a variety of
languages. Summarizingthe argumentpresented in the cited work, we may say that the
syntactic position used to encode the causee of a causative construction (i.e. the
individualcaused to carry out some action) will be the highest position on the AH that
is not already occupied. The following French examples illustratethe general trend of
the data:
(66) J'ai fait courir Henriette.
'I made Henriette run.'
In (66), the causee is presented as a DO (Henriette);note that the constructionalready
has a subject.
(67) J'ai fait mangerles gateaux a Henriette.
'I made Henriette eat the cakes.'
Here, the causee is presented as an IO (a'Henriette); the construction alreadyhas a
subject (je) and a DO (les gateaux).
(68) J'ai fait ecrire une lettre au directeurpar Henriette.
'I made Henriette write a letter to the director.'
In (68), the causee is an OBL (parHenriette);there is alreadya subject(e), a DO (une
lettre), and an IO (au directeur).
A second extension of the rangeof applicabilityof the AH is in determiningcross-
languagerestrictionson advancementprocesses (Keenan and Comrie (1972); Perlmut-
ter and Postal (1974); Johnson (1974b); Keenan (1975b); Trithart (1975)). By an
"advancement process" we shall mean a productive syntactic process that converts
sentences containing NPs on a low position on the AH into roughly synonymous
I Similar facts obtain in
languages in which the RC precedes the head. Thus in Japanese and Basque,
for instance, in the equivalent of "(the) he hit man" (= "the man that he hit"), the subject pronoun of the
main verb in the RC cannot be understood as being coreferential with the head. See Keenan (1974) for
examples. Thus the Crossover Principle (Postal (1971)) is not the correct explanation for these cases.
96 EDWARD L. KEENAN AND BERNARD COMRIE

sentences containing that NP on a higher position on the AH. An obvious example


would be Passive, which in many languagesserves to advance direct objects to subject
position. The work cited above leads us to posit two Advancement Conditions:
(AC 1) The Target Condition
The higher an NP is on the AH, the more accessible it is, in general, as a
target of an advancement.
(AC2) The Distance Condition
The fartherX is from a fixed position Y higherthan X, the more difficultit
is, in general, to advance X to Y.
(The qualification"in general" is requiredbecause many such advancementprocesses,
though productive, are subject to idiosyncraticgovernmentby individuallexical items.)
The Target Conditionwould claim, for instance, that if a languagehas a rule advancing
non-DOs to DO, then it must also have a rule advancingnon-Subjectsto Subject;that
is, since DO is a target for such processes, there must also be processes with Subject
as their target. Indonesian(Chung(1976))is an instance of a languagewith advancement
of 10 to DO, and, as predictedby AC1, it also has advancementof DO to Subject. The
Distance Condition would claim, for instance, that if a languagehas a rule advancing
OBL to Subject, then it must also have a rule advancing DO to SU; that is, since OBL
is farther away from Subject than DO is, OBL is a less likely candidate for
advancementthan DO. Malagasy (Keenan (1972b))has a rule advancing OBLs (and
also 1Os) to Subject, and also has a rule, as predicted, advancing DO to Subject. For
furtherexamples and detailedjustification,the works cited above should be referredto.
In conclusion, this article has aimed to demonstrate that the Accessibility
Hierarchy correctly characterizes the relative accessibility of different syntactic
positions to relative clause formation; other work indicates that the Accessibility
Hierarchy may play a more general role in determining the accessibility of noun
phrases as candidates and targets for syntactic processes, although the precise
delimitationof the area of relevance of the Accessibility Hierarchyremains a task for
future research.

References
Abrahams, R. C. (1959) The Language of the Hausa People, University of London Press,
London.
Anderson, S. R. (1975) "Tongan," unpublishedmanuscript.
Anderson,S. R. (1976)"On the Notion of Subjectin ErgativeLanguages,"in C. Li, ed. (1976).
Andrews, A. (1975) Studies in the Syntax of Relative and ComparativeClauses, unpublished
Doctoral dissertation, MIT, Cambridge,Massachusetts.
Bamgbose, A. (1966)A Grammarof Yoruba,CambridgeUniversity Press, Cambridge.
Biggs, B. (1974) "Some Problemsof Polynesian Grammar,"Journal of the Polynesian Society
83, 401-426.
NOUN PHRASE ACCESSIBILITY AND UNIVERSAL GRAMMAR

Bird, C. (1966) " Relative Clauses in Bambara," Journal of West African Languages 5, 35-47.
Brown, H. D. (1971) "Children's Comprehension of Relativized English Sentences," Child
Development 42, 1923-1926.
Chung, S. (1976) "An Object-Creating Rule in Indonesian," Linguistic Inquiry 7, 41-87.
Churchward, C. (1953) Tongan Grammar, Oxford University Press, London.
Cole, P. and J. Sadock, eds. (1976) Syntax and Semantics: Grammatical Relations, Academic
Press, New York.
Comrie, B. (in press) "Causatives and Universal Grammar," Transactions of the Philological
Society, 1974.
Cook, V. J. (forthcoming) "Strategies in the Comprehension of Relative Clauses," Language
and Speech.
Craig, C. G. (1974) "A Wrong Cyclical Rule in Jacaltec?" in M. W. LaGaly, R. Fox, and A.
Bruck, eds. (1974).
Deibler, E. (1973) Gahuku Verb Structure, unpublished Doctoral dissertation, University of
Michigan, Ann Arbor, Michigan.
de Rijk, R. P. G. (1972) "Relative Clauses in Basque-a Guided Tour," in P. M. Peranteau, J.
N. Levi, and G. C. Phares, eds. (1972).
Dixon, R. M. W. (1969) "Relative Clauses and Possessive Phrases in Two Australian
Languages," Language 45, 35-44.
Dixon, R. M. W. (1972) The Dyirbal Language of North Queensland, Cambridge University
Press, Cambridge.
Gary, J. 0. and E. L. Keenan (1976) "On Collapsing Grammatical Relations in Universal
Grammar," in P. Cole and J. Sadock, eds. (1976).
Giv6n, T. (1972) "Complex NPs in Word Order and Resumptive Pronouns in Hebrew," in You
Take the High Node and I'll Take the Low Node, Chicago Linguistic Society, Chicago.
Giv6n, T. (1976) "Topic, Pronoun, and Grammatical Agreement," in C. Li, ed. (1976).
Gorbet, L. (1972) "How to Tell a Head When You See One: Disambiguation in Diguefno
Relative Clauses," unpublished manuscript.
Grossman, R. E., L. J. San, and T. J. Vance, eds. (1975) Papers from the Eleventh Regional
Meeting of the Chicago Linguistic Society, University of Chicago, Chicago Illinois.
Hale, K. (n.d.) "The Adjoined Relative Clause in Australia," mimeo, MIT, Cambridge,
Massachusetts.
Hatch, E. (1971) "The Young Child's Comprehension of Relative Clauses," Technical Note 2-
71-16, Southwest Regional Laboratory, Los Alamitos, California.
Hawkins, S. and E. L. Keenan (1974) "The Psychological Validity of the Accessibility
Hierarchy," Summer Meeting, LSA.
Hohepa, P. (1969) "The Accusative to Ergative Drift in Polynesian Languages," Journal of the
Polynesian Society 78, 259-329.
Johnson, D. E. (1974a) "On the Role of Grammatical Relations in Linguistic Theory," in M. W.
LaGaly, R. Fox, and A. Bruck, eds. (1974), 269-283.
Johnson, D. E. (1974b) "Prepaper on Relational Constraints on Grammar," unpublished
manuscript, Math. Sciences Department, IBM Research Center, Yorktown Heights,
New York.
Karlsson, F. (1972) "Relative Clauses in Finnish," in P. M. Peranteau, J. N. Levi, and G. C.
Phares, eds. (1972).
Keenan, E. L. (1972a) "On Semantically Based Grammar," Linguistic Inquiry 3, 413-461.
Keenan, E. L. (1972b) "Relative Clause Formation in Malagasy," in P. M. Peranteau, J. N.
Levi, and G. C. Phares, eds. (1972).
98 EDWARD L. KEENAN AND BERNARD COMRIE

Keenan, E. L. (1974) "The Functional Principle: Generalizing the Notion of Subject-of," in M.


W. LaGaly, R. Fox, and A. Bruck, eds. (1974), 298-309.
Keenan, E. L. (1975a) "Logical Expressive Power and Syntactic Variation in Natural Lan-
guage," in E. L. Keenan, ed., Formal Semantics of Natural Language, Cambridge
University Press, Cambridge.
Keenan, E. L. (1975b) "Some Universals of Passive in Relational Grammar," in R. E.
Grossman, L. J. San, and T. J. Vance, eds. (1975), 340-352.
Keenan, E. L. (1976a) "Remarkable Subjects in Malagasy," in C. Li, ed. (1976).
Keenan, E. L. (1976b) "Towards a Universal Definition of Subject of," in C. Li, ed. (1976).
Keenan, E. L. and B. Comrie (1972) "Noun Phrase Accessibility and Universal Grammar,"
Winter Meeting, LSA.
Lafitte, P. (1962) Grammaire Basque, Editions des "Amis du Musee Basque" et "lkas",
Bayonne.
LaGaly, M. W., R. Fox, and A. Bruck, eds. (1974) Papers from the Tenth Regional Meeting of
the Chicago Linguistic Society, University of Chicago, Chicago, Illinois.
Legum, S. (1975) "Strategies in the Acquisition of Relative Clauses," Technical Note 2-75-10,
Southwest Regional Laboratory, Los Alamitos, California.
Li, C., ed. (1976) Subject and Topic, Academic Press, New York.
Li, C. and S. Thompson (1974) "Subject and Topic: A New Typology of Language," Winter
Meeting, LSA.
Li, C. and S. Thompson (1976) "Evidence Against Topicalization in Topic Prominent Lan-
guages," in C. Li, ed. (1976).
MacDonald, R. and D. Soenjono (1967) Indonesian Reference Grammar, Georgetown Univer-
sity Press, Washington, D.C.
Peranteau, P. M., J. N. Levi, and G. C. Phares, eds. (1972) The Chicago Which Hunt, Chicago
Linguistic Society, Chicago.
Perlmutter, D. M. and P. M. Postal (1974) Linguistic Institute Lectures, unpublished ms.
Postal, P. M. (1971) Crossover Phenomena, Holt, Rinehart and Winston, New York.
Schachter, P. (1973) "Focus and Relativization," Language 49, 19-46.
Schachter, P. (1976) "The Subject in Philippine Languages: Topic, Actor, Actor-Topic, or None
of the Above," in C. Li, ed. (1976).
Sheldon, A. (1974) "The Role of Parallel Function in the Acquisition of Relative Clauses in
English," Journal of Verbal Learning and Verbal Behavior 13, 272-281.
Silitonga, M. (1973) Some Rules Reordering Constituents and their Constraints in Batak,
unpublished Doctoral dissertation, University of Illinois, Urbana, Illinois.
Tagashira, Y. (1972) "Relative Clauses in Korean," in P. M. Peranteau, J. N. Levi, and G. C.
Phares, eds. (1972).
Tchekhoff, C. (1973) "Some Verbal Patterns in Tongan," Journal of the Polynesian Society 82,
281-292.
Trithart, L. (1975) "Relational Grammar and Chicewa Subjectivization Rules," in R. E.
Grossman, L. J. San, and T. J. Vance, eds. (1975), 615-624.
Valli, H., N. Hernandez, M. Archard-Baule, and M. F. Beretti (1972) Compte-rendu d'une
experience realisee dans une classe de 6e 2, dont le but etait d'etudier les mecanismes de
la production des relatives chez l'enfant, Institut de Didactique et Pedagogie, Universite
de Provence.
Woodbury, A. (1975) Ergativity of Grammatical Processes: A Study of Greenlandic Eskimo, M. A.
Essay, Department of Linguistics, University of Chicago, Chicago, Illinois.
NOUN PHRASE ACCESSIBILITY AND UNIVERSAL GRAMMAR

Keenan
Department of Linguistics
UCLA
Los Angeles, California 90024

Comrie
Department of Linguistics
Sidgwick Avenue
Cambridge CB3 9DA
England

You might also like