F -F : A

F -F : A
FUNCTIONAL FORMS-FORMAL FUNCTIONS: AN ACCOUNT OF COEUR D'ALENE CLAUSE
STRUCTURE
by
Shannon T. Bischoff
_________________________
A Dissertation to the Faculty of the
DEPARTMENT OF LINGUISTICS
In Partial Fulfillment of the Requirements
For the Degree of
DOCTOR OF PHILOSOPHY
In the Graduate College
THE UNIVERSITY OF ARIZONA
2007
2
THE UNIVERSITY OF ARIZONA
GRADUATE COLLEGE
As members of the Dissertation Committee, we certify that we have read the dissertation
prepared by Shannon T. Bischoff
entitled Functional Forms-Formal Functions: An Account of Coeur d’Alene Clause
Structure
and recommend that it be accepted as fulfilling the dissertation requirement for the
Degree of Doctor of Philosophy.
_______________________________________________________________________
Heidi Harley, Committee Co-Chair
_______________________________________________________________________
Jane Hill, Committee Co-Chair
_______________________________________________________________________
Andrew Carnie
_______________________________________________________________________
Richard Demers
_______________________________________________________________________
Simin Karimi
_______________________________________________________________________
Mary Ann Willie
Date: 12/1/06
Date: 12/1/06
Date: 12/1/06
Date: 12/1/06
Date: 12/1/06
Date: 12/1/06
Final approval and acceptance of this dissertation is contingent upon the candidate’s
submission of the final copies of the dissertation to the Graduate College.
I hereby certify that I have read this dissertation prepared under my direction and
recommend that it be accepted as fulfilling the dissertation requirement.
________________________________________________ Date: 12/1/06
Dissertation Director: Heidi Harley
________________________________________________ Date: 12/1/06
Dissertation Director: Jane Hill
3
STATEMENT BY AUTHOR
This dissertation has been submitted in partial fulfillment of requirements for an
advanced degree at the University of Arizona and is deposited in the University Library
to be made available to borrowers under rules of the Library.
Brief quotations from this dissertation are allowable without special permission,
provided that accurate acknowledgment of source is made. Requests for permission for
extended quotation from or reproduction of this manuscript in whole or in part may be
granted by the head of the major department or the Dean of the Graduate College when in
his or her judgment the proposed use of the material is in the interests of scholarship. In
all other instances, however, permission must be obtained from the author.
SIGNED: Shannon T. Bischoff
4
ACKNOWLEDGEMENTS
The most important acknowledgement must go to Takae, whose sacrifices and support
made this entire project possible. For that, she deserves the greatest acknowledgment.
I also owe a great deal of thanks to Anthony Mattina. A self-described
“cantankerous old bastard,” whose generosity, kindness, and encouragement sent me on
this path. It is rare that we meet our heroes in the flesh, and rarer still that after such an
encounter we find them unflawed. Anthony Mattina embodies all this is good in a
scholar: wisdom, hard work, determination, attention to detail, respect for the facts, a
sense of responsibility to the communities he finds himself working with and in,
dedication, commitment, and a willingness to share his knowledge and experience with
others freely. He is far from a “cantankerous old bastard.” I also must thank Dr. Mattina
and the University of Montana Department of Anthropology for hosting me for a year to
collect data from the Reichard manuscripts in their Linguistic Lab.
Here I would like to acknowledge my committee, without whom you would not
be reading this work. I was repeatedly warned not to have a large committee, that
“problems would arise” working with so many members. However, I cannot imagine a
more successful “collaboration” than that between my committee members and me. To
have such scholars to work with was an incredible privilege, that they work so well
together and with me was a gift. Each showed great respect for the other member’s
opinions and ideas, and provided constant, constructive, and positive feedback at each
stage of the dissertation.
I especially wish to thank the co-chairs of my committee, Heidi Harley and Jane
Hill. Heidi pushed me in directions that I had not expected to go. Also like Jane, the
breadth and depth of her knowledge was at times overwhelming. Her enthusiasm for the
dissertation was greatly appreciated. Throughout my tenure, Heidi was encouraging and
unwavering in her support for my research on Coeur d’Alene. She showed great respect
for my ideas and my ignorance (kindly pointing me in the right direction on more than
one occasion). Further, despite her many commitments she was always available and
always provided thought provoking commentary on various drafts of this dissertation.
Jane’s willingness to work with a dissertation not quite Anthropological was very
generous. Her knowledge of, and experience with, Indigenous languages of the world
was invaluable. Her attention to the data and insightful questions regarding the structure
of the language always left me with new perspectives. She inspired me to look closely at
the forms under discussion and not to take anything for granted, to push myself and my
understanding as far as possible in order that I might come to terms with the facts in a
fruitful fashion. She also allowed a great deal of “constructive” bitching and moaning, for
which I am forever indebted.
Andrew Carnie was extraordinary in his commentary. He, more than anyone else,
challenged my assumptions and conclusions. He happily served the role of “devil’s
advocate,” constantly directing me to problems in my analysis and leading me to
solutions. His kindness and generosity were remarkable, as were his insights and support.
I am especially thankful for his willingness to fill the role of advisor, and all the
responsibilities that come with job, when Heidi was away at Harvard and Cambridge.
5
Simin Karimi became my “go to” committee member. I exploited here “open door”
policy to the limit. She was always available whenever I had an inspiration for a new or
improved analysis. I spent dozens of hours in her office drawing trees, discussing the
literature, and generally taking her time. Before a draft was ever written it was discussed
in great detail with Simin, and chances are the first draft of any idea was given to her to
before anyone else. She had a remarkable knack for identifying what was the crux of any
given issue, helping me recognize this, and directing me to the relevant literature.
Dick and Mary were indispensable. Dick’s knowledge of Salish was a comfort
throughout. His insights regarding linguistics in general helped me to keep a perspective
on what the meaning of this endeavor, linguistic inquiry, is. His advice, beyond simply
that directed to the content of the dissertation itself, kept me focused, motivated, and not
afraid to hit a “brick wall” in my analysis. His love of linguistic inquiry was infectious
and inspiring. He constantly reminded me how much fun this is! It was Mary who kept
me on track. She inspired me to take myself and this work seriously, a difficult task at
times. Linguistics is a rather esoteric field, and it is sometimes difficult to find meaning
in what we do. Mary inspired me to recognize the valuable skills that I have acquired in
this process. She also gave me a sense of how to employ these skills in a fruitful and
satisfy way. This is perhaps the greatest lesson of this entire dissertation.
Among my colleagues, I must give special thanks to Yosuke Sato and Mans
Hulden for being brilliant and willing to share their insights. Naomi Ogasawara is also
owed a great deal of thanks. Our collaborative work on pro in Japanese was the starting
point for much discussion in Chapter 3. I also need to thank the “Serious Dissertators
Group,” especially Meg. For coffee Polly and piece of mind Emily and Jaime deserve
thanks.
I wish to also thank Ivy Doak, and Raymond Brinkman for helping with the data
questions and various inquiries throughout the dissertation. Terry Langendoen, who
funded me as an RA throughout my tenure also deserves a great deal of thanks, as does
Amy Fountain for being one of my linguistic heroes. Mike Hammond deserves a round
of applause for being a great Chair and an all around “good dooby,” not to mention for all
of his support. Also, Akira Watanabe deserves thanks for hosting me for a summer at the
U of Tokyo where I conducted research on pro. Further, the generous financial support of
Kazuo and Satoko Sasaki must be acknowledged. It should also be noted that work in this
dissertation was funded in part by a fellowship from NSF and the Japanese Society for
the Promotion of Science.
6
DEDICATION
For Takae, Dorothy, Tom, and Gladys
7
TABLE OF CONTENTS
L IS T
OF
T AB LES …………………………………………………………………..11
L IST
OF
S YMBOLS
AND
A BBREVIATIONS …………………………………………12
A B S TR A C T ………………………………………………………………………..13
C HAPTER 1: I NTRODUCTION ………………………………………………………14
Section 1: Introduction………………………………………………………...14
Section 2: Goals of dissertation……………………………………………….15
Section 3: Theoretical considerations…………………………………………19
Section 4: General outline of the dissertation…………………………………22
C HAPTER 2: A N
OVERVIEW OF
C OEUR
D 'A LENE MORPHOSYNTAX ………………..24
Section 1: Introduction………………………………………………………...24
Section 1.1: Basic sentence structure……………………………………….24
Section 1.2: Person marking………………………………………………..26
Section 1.2.1: Intransitive person markers………………………………26
Section 1.2.2: Transitive person markers………………………………..28
Section 1.2.3: Genitive person markers………………………………….32
Section 1.2.4: Predicate pronouns………………………………………..34
Section 2: Sentential aspect, stem aspect, and mood………………………….35
Section 2.1: Sentential aspect………………………………………………36
Section 2.2: Stem aspect……………………………………………………37
Section 2.3: Mood…………………………………………………………..39
Section 3: Tense……………………………………………………………….39
Section 4: Transitivizing morphemes…………………………………………41
8
TABLE OF CONTENTS – Continued
Section 4.1: The lone
and directive
Section 4.2: Causative transitivizer
transitivizers……..………………...42
…...………………………………...44
Section 4.3: Applicative transitivizers………………………………………...44
Section 4.3.1: Possessor applicative
……………………………………..45
Section 4.3.2: Benefactive applicative
Section 4.3.3: Dative
…………………………...…… 46
…………………………………………………..47
Section 5: Determiner phrases………………………………………………...48
Section 6: Summary….………………………………………………………..52
C HAPTER 3: C R B ASIC
CLAUSE STRUCTURE :
A
FORMAL ACCOUNT ……………….53
Section 1: Introduction………………………………………………………...53
Section 2: Conflation………………………………………………………….56
Section 3: Null arguments and bound pronouns………………………………59
Section 3.1: Bound pronoun analysis………………………………………59
Section 3.2: Radical argument drop analysis……………………………….61
Section 3.3: Bound pronouns over radical argument drop…………………64
Section 3.3.1: Null arguments in Cr: What are they?................................65
Section 3.3.2: Licensing and identification considerations……………...66
Section 3.3.3: Learnability considerations………………………………69
Section 3.3.3.1: Null argument analysis: The hybrid system………....69
Section 3.3.3.2: Bound pronoun analysis: The unified system……….71
Section 3.4: Bound pronouns vs. agreement: Some typological
considerations….…………...………………………………....73
9
TABLE OF CONTENTS – Continued
Section 4: Interim summary………………………………………………...…76
Section 5: Bound pronouns as
-pronouns……………………………………77
Section 5.1: The basic tenets of Déchaine and Wiltschko (2002)…………..77
Section 5.2: DP-Pronouns: The Halkomelem
(Salishan/Central Coast) data………………..……………………79
Section 5.3:
-Pronouns: The Shuswap (Salishan/N. Interior) data………..82
Section 5.4: Coeur d'Alene (Salishan/S. Interior) emphatic pronominals….88
Section 6: Bound pronouns: Possible diachronic antecedents………………...92
Section 6.1: Grammaticalization……………………………………………94
Section 7: Conclusions………………………………………………………...99
C HAPTER 4: L EXICAL A FFIXES …………………………………………………..101
Section 1: Introduction………...……………………………………………..101
Section 2: Lexical Affixes in Cr…………………………………………..…102
Section 2.1: Stem+stem vs. LAs: The differences……………………….105
Section 2.2: LAs historically……………………………………………..108
Section 2.3: LAs in context………………………………………………113
Section 3: LAs: Little n………………………………………………………119
Section 4: Formal analysis of LAs…………………………………………...120
Section 4.1: Conflation revisited………………………………………….121
Section 4.2: LAs: An incorporation account……………………………...122
Section 4.3: Baker, Aranovich, and Golluscio (2004)…………………….126
Section 4.4: Stem+stem incorporation in Cr………………………….…...131
10
TABLE OF CONTENTS – Continued
Section 4.5: Conflation over feature deletion………………………….…133
Section 5: Conclusion………………………………………………………..136
CHAPTER 5: ELEVEN PARTICLES IN COEUR D'ALENE: SOME THEORETICAL
C O NS ID ER A TIO N S
...…………………………………………………138
Section 1: Introduction………………………………………………….……138
Section 2: Temporal and sentential adverbial particles………...…………….140
Section 2.1: Temporal adverbial particles:
and
,
,
………………………………………………….140
Section 2.2: Sentential adverbial particles:
Section 3: Modals:
,
, and
and
………………..142
………………………………………145
Section 3.1:
future intentional, permissive, mild request…………….145
Section 3.2:
and
: ought and possibility…………………………148
Section 4: Aspectual particles:
Section 5: Interrogative
and
………………………………150
…………………………………………….……..155
Section 6: Rizzi’s split CP…………………………………………………...158
Section 7: Conclusions……………………………………………………….162
C HAPTER 6:C ONCLUSIONS ………………………………………………………165
Section 1: Introduction……………………………………………………….165
Section 2: Recapitulation…………………………………………………….165
Section 3: Future inquiry…………………………………………………….171
A P P E N D IX ……………………………………………………………………….173
REFFERENCES…………………………………………………………………………...178
11
LIST OF TABLES
T ABLE 1, Intransitive subject marking morphemes………………………………27
T ABLE 2, Transitive object morphemes…………………………………………..28
T ABLE 3, Nontopic ergative person marking morphemes…………………...…..28
T ABLE 4, Transitive subject person marking morphemes……………………….30
T ABLE 5, Transitive object-subject combinations……………………………….32
T ABLE 6, Genitive person marking morphemes…………………………………32
T ABLE 7, N. Mattina's properties of aspect……………………………………...36
T ABLE 8, Particles analyzed in Chapter 5……………………………………….41
TABLE 9, Some cross-linguistic facts regarding null arguments………………...68
T ABLE 10, Nominal proform typology…………………………………………..78
T ABLE 11, Halkomelem independent pronouns…………………………………79
T ABLE 12, Shuswap independent pronouns……………………………………..83
T ABLE 13, Coeur d'Alene Emphatic Pronouns………………………………….89
TABLE 14, Lexical affixes with no phonological similarity to free forms in
Coeur d'Alene…………………………………………………….…104
T ABLE 15, Lexical affixes with similar phonological content to free forms in
Coeurd'Alene………………………………………………………...104
TABLE 16, Lexical affixes with similar phonological content to free forms in
Coeur d'Alene…..………………………………………………...….109
TABLE 17, Coeur d'Alene particles….……………………………………………..139
T ABLE 18, Coeur d'Alene particles…..…………………………………………..158
TABLE 19, Coeur d’Alene particles…..…………………………………………..169
12
LIST OF SYMBOLS AND ABBREVIATIONS
abs
acc
aug
C
con
cont
count
ct
cust
cvc
dat
det( )
devel
deic
dim
dt
erg
absolutive
accusative
augmentative
C reduplication
connective
continuative
counter
causative
customary
CVC reduplication
dative
determiner
developmental
deictic
diminutive
directive transitivizer
ergative
fut
future
g
genitive
hab
habitual
imp
imperative
inv
involuntary
irr
irrealis
intr
interrogative
int
intentional
inchoat inchoative
indef
indefinite
loc
m
ncr
neg
nom
nte
pl
fposs
prox
rel
rdp
sg
s.t.
sub
t
term
unrlz
proximate
relational
reduplication
singular
something
subordinator
transitive
terminative
unrealized
vc
$
%
VC reduplication
root
morpheme boundary
lexical suffix
&
intransitive subject person
marker
compound indicator
indicates gloss of reduplicated
element
incomplete analysis
exaggeration (vowel lengthening
C1__
< >
**
'
"
!
locative
noncontrol resultive
negative
nominative
nontopic ergative
plural
future possibility
"#
"
1
"This suffix ... is difficult to classify. It is used
especially with kin-terms and names for
persons ... It is part of certain verbs also."
Reichard (1938:621)
2
The exact nature of this morpheme
is not clear. The functions of this morpheme
include indicating: middle constructions;
continuative; a patient subject or an agent subject
depending on root type; detransitivization. It is
also used in causative constructions (cf. Doak
1997:79).
3
Doak (1997) labels this morpheme “oblique.”
However, the oblique is homophonous with an
indefinite determiner that is also glossed
“oblique” by Doak (214-15). Throughout I label
both , and provide discussion where clarity
demands.
13
ABSTRACT
Coeur d’Alene, also known as Snchitsu’umshtsn, is a Southern Interior Salishan language
no longer learned by children. Descriptive work on the language has been carried out
since the early nineteenth-century (Tiet 1904 through 1909 in Boaz and Tiet 1930;
Reichard 1927-29, 1938, 1939; Doak 1997); however, a formal account of the basic
clause structure of this polysynthetic language has until now not been proposed. This
thesis presents such a formal analysis within the Minimalist Program (Chomsky 1995,
1998, 2000, 2001a, 2001b; Lasnik 1999a, 1999b, 2000; among others), employing the
tenets of Distributed Morphology (Halle and Marantz 1993; Harley and Noyer 1999;
among others). Demonstrating that an analysis of person marking morphemes as bound
pronouns (Jelinek 1984) is more “economical” in terms of Chomsky’s (1995:367)
Elementary Principles of Economy, the thesis goes on to account for the phenomena of
lexical affixation (Carlson 1990; Kinkade 1998; Gerdts 2003; among others), in Coeur
d’Alene as incorporation. Appealing to Hale and Keyser’s (2002) theory of conflation as
Head-movement (Harley 2004), an approach to incorporation is proposed which captures
Chomsky’s (1995) claim that head-movement is phonological while at the same time
illustrating that lexical affixes in Coeur d’Alene serve as incorporated arguments. The
thesis concludes with an articulation of the left periphery (material above vP here), based
on the strict ordering of a series of mood, adverbial, model, and aspectual particles. It is
shown that this articulation in Coeur d’Alene patterns with Cinque’s (1999) proposed
universal hierarchy of functional and adverbial heads. In this way, the basic clause
structure of Coeur d’Alene is formally presented.
14
CHAPTER 1
INTRODUCTION
1. Introduction
One of the most influential claims in linguistic theory over the past 50 years has been that
while some languages seem quite different, they are actually far more similar than one
might assume. That is, given a language like German and a language like Chinese, one
might guess that they do not have a great deal in common. For example, in Chinese there
are a number of sounds not found in German, a variety of tones for example. In German,
when a sentence is uttered it must have a subject. In Chinese on the other hand, the
subject and object can seemingly be omitted freely. However, German and Chinese do
have elements such as ‘nouns’ and ‘verbs’, and these elements do combine to form
complete utterances, or sentences, in both languages. One of the goals of linguistic
inquiry is to understand how elements such as ‘nouns’ and ‘verbs’ are combined to create
these utterances and sentences. This is where the claim that all languages are actually
quite similar comes in.
Recognizing these seemingly basic similarities across languages linguists have
hypothesized that although languages seem different at the level of an utterance or
sentence, underlyingly they are created from the same innate elements and employ the
same innate mechanism(s) to organize those elements into utterances. That is, although
German and Chinese sound quite different, they are comprised of the same basic
elements, ‘nouns’ and ‘verbs’ etc., and they both employ the same mechanism(s) to
organize those elements into comprehensible utterances or sentences.
15
The primary goal of this dissertation is to look at a language that is on the surface
quite different from the vast majority of languages previously studied, in this case a
polysynthetic language, specifically Coeur d’Alene (Salish/Idaho USA). Polysynthetic
languages are most notable for the fact that one ‘word’ of such a language is often
translated into a complete English sentence. For example the Coeur d’Alene ‘word’
(
can be translated as the English sentence, ‘He cut wood for me’. If the
claim is that utterances in any given language are built from the same innate pieces by the
same innate mechanism(s), then even though on the surface a polysynthetic language is
uttered as a single ‘word’, and an equivalent utterance in a language like English requires
several words, it should be possible to demonstrate that both are quite similar at some
basic level. This is exactly what this dissertation hopes to demonstrate.
In the remainder of this introduction, the goals of this dissertation will be
elaborated, this will be done in Section 2. Further, in Section 3 the primary theoretical
considerations employed in this dissertation will be presented. Finally, in Section 4, a
general outline of the dissertation is presented.
2. Goals of dissertation
The purpose of this dissertation is three-fold. The first goal is to present a formal account
of the basic clause structure of Coeur d'Alene. This includes identifying the functional
projections within the basic clause structure of Coeur d'Alene in terms of Cinque's (1999)
proposed universal hierarchy of functional projections and Rizzi's (1997a) Split CP
hypothesis. To my knowledge there has not yet been a formal account of basic clause
structure in Coeur d'Alene, and neither has there been a proposal put forward regarding
the hierarchy of functional projections in this language. This dissertation provides a
unique organization of data and analysis not previously available for typological and
16
cross family comparison, thus adding to our broader understanding of the intricacies of a
specific language, and how those intricacies compare with the vast number of languages
of the world.
Second, it is a goal of this dissertation to test the long-standing claim that
although languages may seem radically different on the surface, underlyingly they are the
same. It is argued in this dissertation that Coeur d'Alene (Salishan/Idaho USA), a
polysynthetic language, adheres to the same underlying mechanisms postulated for the
typologically quite different English. Employing the primary tenets of the Minimalist
Program (Chomsky 1995, 1998, 2000, 2001a, 2001b; Lasnik 1999a, 1999b, 2000; among
others) within the Principles and Parameters approach to morphology and syntax, it is
demonstrated that the same theoretical claims applied to typologically divergent
languages can account for a range of specific phenomena in Coeur d'Alene. In addition, it
is demonstrated that inquiry into specific phenomena in Coeur d'Alene can add insight
into various syntactic and morphological phenomena attested cross-linguistically. In this
way, evidence is brought to the fore, which brings us closer to an understanding of how
knowledge of language is stored in the mind.
The third goal of this dissertation is to bring Coeur d'Alene and the Salishan work
of Gladys Reichard to a wider audience. Coeur d'Alene, like all endangered languages of
the world, is invaluable to the scientific endeavors of the linguistic community at large,
and to the community of speakers and potential speakers that hope to use Coeur d'Alene
to express themselves and their culture. It is hoped that the fascinating phenomena seen
in the data of this dissertation will inspire others to begin or continue work on less
familiar and endangered languages. It is also hoped this dissertation will highlight the
value of heritage materials, such as those of Gladys Reichard, Tom Miyal, and Dorothy
Nicodemus used here, and thus inspire others to begin to work with such resources. Too
17
often these materials are left languishing in archives, garages, offices, in homes, and
numerous other places and are themselves at great risk of loss, just as the languages they
record are in great danger of being lost forever.
Along with this goal is the push to present a general research program for Coeur
d'Alene. In the summers of 1927 and 1929, at the behest of Franz Boas, Gladys Reichard
came to Coeur d'Alene country to record the language. She collected approximately
forty-eight texts. At this time two projects are underway to develop a corpus of Coeur
d'Alene, comprised primarily of the Reichard texts, which will provide thousands of
examples of data for linguistic research and revitalization efforts.4 It is hoped that the
constructions and elements analyzed in this dissertation will serve as a possible starting
point for future formal inquiry of Coeur d'Alene, and serve as further motivation to
complete the two corpora projects.
As mentioned, the heritage materials of Gladys Reichard, Dorothy Nicodemus,
and Tom Miyal, in the form of unpublished manuscripts of the narratives of Nicodemus
and Miyal recorded by Reichard (1927-29), were used in this dissertation. Of the some 48
narratives, nine were used. These nine narratives were morpho-syntactically and morphophonologically analyzed by the current author for their use here. The narratives used in
this dissertation, along with the abbreviations employed to denote them are presented
here:
4
Currently one project underway, funded by NSF and directed by Ivy Doak, is working to digitally archive
the Reichard manuscripts and other Coeur d’Alene material as well as create an online dictionary. Another
project underway by the current author and Man's Hulden is the development of a morphological parser
designed for Navajo based on a computational system (developed by Hulden) that will be able to exploit
the digitally archived Reichard manuscripts to create a variety of morphological corpora.
18
Boy takes food
(btf)5
Calling his kind (Dorothy)6
(chkd)
Calling his kind (Tom)
(chkt)
Coyote imitates Magpie
(cim)
Coyote overpowers sun securing sun disc
(cosssd)
Coyote steals son's wife
(cssw)
Lynx
(L)
Man caught in fire coral
(mcfc)
War between Blackfeet and Coeur d'Alene
(wbc)
In addition to these nine narratives, other works of Reichard used in this
dissertation include: Reichard's 1938 grammar, her 1939 partial stem list, and her 1947
English translations of Coeur d'Alene narratives. Other data comes from Nicodemus'
(1975) Coeur d'Alene dictionaries, and Doak's 1997 description of grammatical relations
in Coeur d'Alene. Further data was arrived at in consultation with Salishan scholars
Anthony Mattina and Ivy Doak, and the Coeur d'Alene Director of Language Programs
Raymond Brinkman.
Before moving forward a brief introduction to Coeur d’Alene is perhaps
necessary. Coeur d’Alene is a language no longer learned by children. It is a Southern
Interior Salishan language. There has been descriptive work carried out on Coeur d’Alene
(see references above and references therein), but to the author’s knowledge, no formal
account of Coeur d’Alene phenomena has been presented. Coeur d’Alene is spoken by a
few elderly speakers on the reservation near Plummer, Idaho. Like many of the language
communities of the Americas, speakers of Coeur d’Alene suffered greatly as a result of
the imperialism of European powers and the western expansion of the US prior to the
5
The letters in parentheses represent the abbreviations employed before line numbers in the data. A form
like btf34 would indicate line 34 of the narrative Boy Takes Food.
6
The name here and in the following narrative refer to Reichard's informants.
19
twentieth century. In the twentieth century, Coeur d’Alene continued to demise as a result
of the pressures brought on by the dominant culture. Today there are revitalization efforts
underway, efforts that include developing corpora from the unpublished Reichard (192729) manuscripts.
3. Theoretical considerations
These data were analyzed within the framework of the Minimalist Program. This
program is extensively laid out in Chomsky (1993, 1994, 1995, 2000, 2001a, 2001b). The
general tenets of the Minimalist program employed in this dissertation are those of Merge
and Move, Features, Agree, and Economy. Merge and Move are two recursive operations
within the computational system. Merge is the simplest such operation which takes a pair
of syntactic objects,
and , and forms a more complex object J from
and . In short,
Merge combines a head with its complement which then projects and merges with a
specifier as in (1).
(1)
Spec
J
Move takes a copy of an existing element in the structure and places (‘re-Merges’) this
copy in a c-commanding position in the structure.
(2)
Spec
J
20
Features are construed as either interpretable or uninterpretable. The need to allow
grammatical convergence through eliminating uninterpretable features motivates the vast
majority of syntactic operations within the Minimalist Program. Convergence occurs
when an uninterpretable feature F of a Head H, a probe, is deleted in a matching
relationship with a c-commanded element, the Head of an XP, a goal, with matching
interpretable features. This relation between the probe and goal is an Agree relation. The
derivation of a sentence is further subject to general conditions of Economy. Chomsky
1995 states this as follows:
( 3 ) Elementary Principles of Economy
a.
Add optional
to numeration only if it has an effect at the interface.
b.
At each stage of a derivation, apply the most economical operation that leads
to convergence. (367)
Many of the other notable theoretical considerations employed in this dissertation
are derived from Distributed Morphology (Halle and Marantz 1993; Harley and Noyer
1999; among others). Distributed Morphology is a late-insertion, piece-based theory of
grammar. The architecture of the grammar, still of the Y-type, separates the lexicon into
three separate components; a set of morphosyntactic features manipulated by syntactic
operations, a set of vocabulary items corresponding to phonological content, and an
encyclopedia that gives semantic interpretation for vocabulary items in contexts. The
basic idea of Distributed Morphology (DM) is that Lexical items consist of feature
bundles comprising semantic, phonological, and formal features. Further, under this
view, insertion of phonological content of these bundles occurs post-syntactically. In this
way morphemes are subject to a distinct morphosyntactic description, representing a root
plus any material attached / added to it, and a distinct morphophonological description.
This model is diagramed below.
21
( 4 ) O’Donnell’s (2004) DM Model
MORPHOSYNTACTIC
FEATURES:
[+N]
[+singular]
[3rd person]
Syntactic
Operations
Morphological
Operations
VOCABULARY
INSERTION:
/kæt/
/-s/
ENCYCLOPEDIA:
Non-linguistic
Knowledge
Phonological Form
Logical Form
Conceptual
Interface
(Meaning)
The three main principles of DM that govern the morphological component are: late
insertion, underspecification, and syntactic hierarchical structure all the way down. Late
insertion is the anti-Lexicalist position according to which syntactic categories are
abstract and contain no phonological content. It is after syntax that phonological material
is inserted into terminal nodes through vocabulary items. Syntactic hierarchical structure
all the way down proposes that there is no principled distinction between the structures
seen in syntax and morphology: The units in both, consisting of terminal nodes and their
content, are discrete. Terminal nodes that are realized as a part of a single
morphophonological word are not the result of a combination operation in pre-lexical
component, but are rather composed and linearized by the syntactic mechanism itself.
22
4. General outline of the dissertation
The dissertation is divided into the following chapters. In Chapter 2 an overview of
Coeur d'Alene morphosyntax is provided in order that those unfamiliar with Coeur
d'Alene will be able to navigate the remaining chapters with little difficulty and a fuller
understanding of the complexities of Coeur d'Alene grammar. In Chapter 3 arguments for
treating person marking morphemes as bound pronouns (Jelinek 1984; Bhat 2004) rather
than as agreement morphology, with arguments being realized as pro (Baker 1996) are
presented. It is claimed that a bound pronoun analysis is preferred in light of various
complications that arise for an agreement analysis regarding Economy, learnability,
typological considerations, and cross-linguistic evidence that analysis of arguments as
pro are in general problematic. Further, a grammaticalization account of the diachronic
origins of bound pronouns is presented and then rejected, and an alternate account of
cross-family variation in terms of argument realization is presented.
In Chapter 4, an account of Lexical Affixes and noun incorporation phenomena in
Coeur d'Alene are presented. It is argued that employing Hale and Keyser's (2002)
Conflation operator as head-movement (Harley 2004) is the most optimal analysis for the
noun incorporating facts of Coeur d'Alene. It is further argued that such an analysis is
preferable to that of Baker, Aranovich, and Golluscio’s (2004) analysis of incorporation
that maintains traditional head-movement, contra Chomsky (1995), and proposes a family
of constraints along with a special deletion mechanism within narrow syntax.
Cross-linguistic implications of a Conflation-type analysis are also presented. In addition,
the traditional view of Lexical Affix diachronic origin is re-evaluated in terms of Health's
1998 Hermit Crab hypothesis of grammaticalization.
Chapter 5 provides an account of various particles and proposes a hierarchy of
functional heads within the Coeur d'Alene clause. This hierarchy is compared with
23
Cinque's 1999 proposed universal hierarchy of functional heads. In addition a comparison
of Rizzi's 1997a Split CP hypothesis and Cinque's hierarchy, in light of the Coeur d'Alene
data, is presented. In the final chapter, Chapter 6, concluding remarks are presented along
with discussion of areas that will be fruitful for future inquiry into the nature of Coeur
d'Alene clause structure and our understanding of language in general.
24
CHAPTER 2
AN OVERVIEW OF COEUR D'ALENE MORPHOSYNTAX
1. Introduction
In this chapter I present an overview of Coeur d'Alene (henceforth Cr), discussing the
morphosyntax of the basic clause structure.7 In this discussion, I consider the basic
clause to comprise a predicate inflected for aspect and person marking morphemes
(person and number). Further, this predicate may be accompanied by mood particles,
temporal particles, and determiner phrases, or some combination of these. It should be
noted that the following discussion benefits greatly from Doak (1997). In the first section
of this chapter I introduce what Doak (1997:50) considers the "basic sentence structure"
of Cr, which includes a discussion of the person marking morphemes. This is followed in
Section 2 by a discussion of sentential aspect, stem aspect, and mood. Section 3
introduces a minimal number of particles. In the transitive examples in these sections the
-st(u) causative transitivizer (ct) and the -nt directive transitivizer (dt) are employed (cf.
section 4 below). Next, in Section 4, I describe the transitivizer paradigm. Finally, in
Section 5, I present a brief discussion of determiner phrases. Concluding remarks appear
in Section 6.
1.1. Basic sentence structure
The basic Cr predicate is composed of the simple root and person marking morphemes,
with no marking for tense, aspect, or mode. Here “root” is used in the tradition of the
7
The phonemic inventory of Cr appears in the Appendix, with brief comments. A complete description of
Cr phonological phenomena can be found in Reichard 1938, and a sketch of important phonological
phenomena can be found in Doak 1997. Other significant work on Cr phonology includes Doak 1992,
Sloat 1966, 1968, 1972 and 1980.
25
Salishan literature to indicate morphemes of the shape CVC, and the rarer CVCC and
CCVC, which take inflectional and derivational formatives, i.e. allow affixation (N.
Mattina 1996: Doak 1997: Thompson 1992: Carlson and Bates 1990 among others). The
predicate may have the form of the bare root where inflection is null as in the case of
constructions with the null third absolutive, (5b) below.8
( 5 ) Cr basic clause structure (Doak 1997:50)
a. (particle)9
predicate
(determiner phrase)
(conjoined phrase)
(second phrase)
b. Example intransitive uninflected (null subject + root)10,11
)& $
3abs& deer
'It's a deer.'
(Doak 1997:36 modified)
c. Example intransitive inflected (subject + root)
(
& $ (
2s& go
'You go.'
8
(Doak 1997:50 modified)
A note regarding the presentation of data, Doak (1997) presents her data in both 3 and 4 line analysis. I
present her data as I find it in her text. In the examples taken from the Reichard (1927-29) manuscripts, I
include a four-line analysis so that the complex phonology may be taken into account by any reader who
wishes to offer suggestions regarding the analysis presented here. Unless otherwise noted, the term
‘modified’ indicates an additional line of analysis (the second) not provided in Doak.
9
Parentheses indicate optionality here.
10
Coeur d'Alene does not have a copula verbal element.
11
Ligature "&" indicates what is traditionally viewed as a clitic in the Salishan literature. In the remaining
chapters the ligature will be replaced with "-.” In Chapter 3 it is argued that these elements are bound
pronominals.
26
d. Example transitive (root + transitivizer + object + subject)12
(!
$
!
hit
-dt -2acc -3erg
'He hit you.'
(Doak 1997:58)
Next we turn to a discussion of person marking morphemes.
1.2. Person marking
There are four person marking paradigms in Cr. These are the intransitive subject
paradigm, transitive subject paradigm, transitive object paradigm and genitive paradigm.
Cr employs both nominative/accusative and ergative/absolutive in its case marking
system.13 There is also a set of what Reichard (1938:554.174) identifies as "independent
pronouns,” and what Doak (1997:72) describes as "predicative pronouns.” Doak
(1997:72) argues that these forms are unanalyzable roots with intransitive person marking
morphemes. Person marking morphemes are mandatory in all constructions. In the
discussion that follows, I first present the intransitive person marking morphemes
followed by the transitive set. Next I introduce the genitive pronominals, and finally I
present the "person predicates.”
1.2.1 Intransitive person markers
The simple intransitive is comprised of a root or stem14 preceded by a subject marking
morpheme. It should be noted that the third person plural
12
occurs in both transitive and
In her examples, Doak (1997) places a hyphen, '-', between the -n and the -t of transitivizing morphemes.
In the examples throughout I do not follow this pattern; I follow the notation of A. Mattina (cf. for example
1982).
13
In the description of person marking morphemes I follow Doak (1997). Doak (1997:76-78) suggests that
the Cr person marking system may have been at one time a split system, but evolved to a system with a full
set of transitive subjects marked ergative rather than simply those that complement the absolutive (third
person only).
14
"Stem" here refers to a root plus any derivational morphology such as locatives, directionals, or lexical
affixes or prefixes. Later this term will be used slightly differently (cf. Chapter 4).
27
intransitive constructions. Further,
is only used where clarity requires a distinction.
That is, when discourse does not make clear which, subject or object, is plural. These
morphemes are listed in Table 1.
Table 1: Intransitive subject marking morphemes
Singular Plural
1 nom
&
2 nom
&
&
3 abs15
)&
&
)&
Examples taken from Doak (1997) using the root *
'to see' illustrate these person
marking forms.
( 6 ) Intransitive person marking morphemes
*!
'I saw.'
*
'You saw.'
)*
*
*
)*
16
'He saw.'
'We saw.'
'You folks saw'.
'They saw.'
(Doak 1997:53-54)
In the following discussion of transitive person marking morphemes it should be noted
that the transitive person marking morphemes follow the root and transitivizing
morphology.
15
There is no means that I am aware of to determine where the null 3abs morpheme appears, therefore, I
represent the morpheme along with the overt morphemes of the given paradigm as is tradition in the
Salishan literature (cf. Doak 1997 among others).
16
Here the null morpheme is indicated by ). Doak (1997) does not include the null morpheme in her
glosses after introducing the morpheme for reason of economy. Throughout they are included and added to
the examples taken from Doak.
28
1.2.2 Transitive person markers
Basic transitive constructions comprise a root or stem followed by one of a given number
of transitivizing morphemes (discussed below in section 4), then the object morphemes
and subject morphemes in that order. Table 2 presents the transitive object morphemes. It
is important to note that the plural
occurs in the transitive paradigm as noted above.
Table 2: Transitive object morphemes
Singular
1 acc
Plural
"
2 acc
"
3 abs
)
)
Before providing examples of the different object morphology, it should be noted
there is a set of transitive subject person marking morphemes which Doak (1997)
identifies as the nontopic ergative (nte). This set of subject morphemes appears in
transitive constructions in which the ergative form, in many discourse constructions, is
replaced with -m or -t (60). Table 3 lists the morphemes in this paradigm; note again that
the plural morpheme
may also occur with this paradigm when clarity demands.
Table 3: Nontopic ergative person marking morphemes
Singular
Plural
1 nte
2 nte
3 nte
Doak provides a full account of the nontopic ergative paradigm, arguing that it does not
represent passive morphology, but rather, a demotion of topic, the subject, from first
topic to second topic. I refer the reader to Doak for a full account of the nontopic ergative
paradigm.
29
The alternate forms identified in the first person singular accusative and second
person singular accusative,
"
and
"
respectively, are selected on the basis
of the given transitivizer of the predicate: those with m primarily occur with the -st(u)
causative transitivizer, while those with s occur with the other transitivizers. The
following examples from Doak (1997) illustrate:
( 7 ) a.
(
$
kill
-ct -1acc -nte
'I got killed.'
b.
(
$
kill
-ct
-2acc -nte
'You got killed.'
c.
(!
$
!
hit
-dt -1acc -nte
'I got hit.'
d.
(!
$
!
hit
-dt -2acc -3erg
'He hit you.'
The "l" in
"
(Doak 1997:57-58)
occurs before the second person plural ergative -p and the nontopic
ergative (nte) -m.
The transitive subject person marking morphemes are presented here in Table 4.
30
Table 4: Transitive subject person marking morphemes
Singular
Plural
1 erg
2 erg
3 erg
nte
"
"
Again, the third person plural marking morpheme
demands. Plural marking
appears where clarity
may appear only once in a clause, and discourse
disambiguates whether it is the subject or object that is plural, or both, in transitive
constructions. In the following example (8), any of the three glosses are possible
renderings of the construction.
(8)
+ , $ . ,
)
go.out -m -ct
-3abs -3erg -pl
'He took them out.'/ 'They took it out.'/ 'They took them out.'
(Doak 1997:59)
At this point it should be mentioned that there is a suffix
, which Doak (1997)
labels the indefinite (65). In Cr there is a person hierarchy that makes it impossible to
express a second person agent, singular or plural, acting upon a first person plural patient
in transitive constructions. Instead, the indefinite
is attached to an intransitive root or
stem inflected with the second person nominative subject. This is illustrated in (9). In (9a)
and (9c) the constraint imposed by the person hierarchy can be illustrated. In (9b) and
(9d)
( 9 ) a.
is illustrated. In (9a) and (9c) the lone -t transitivizer (section 4 below) is used.
* $*
see -t -1pacc -2serg
'You see us.'
31
b.
*
& $*
2nom see -indef
'You see us.'
c. * $*
see -t -1pacc -2perg
'You folks see us.'
d.
*
/
& $*
2nom see -indef
'You folks see us.'
(Doak 1997:65-66)
It is important to note that (9b) and (9d) do not employ transitive morphology and
that they do employ the intransitive nominal morphemes. Although
does refer to an
additional participant, Doak (1997) notes that it would be "misleading to call
an
"object" suffix (67). However, she provides no further analysis of the morpheme.
Compelling though it may be, this morpheme will not be addressed further in the
dissertation, other than to provide an example of its presence in transitive constructions.
It will be left to future inquiries to account for the indefinite constructions illustrated
here. The transitive object-subject combinations are presented in Table 5. Recall that, the
object morpheme precedes the subject morpheme in the verbal morphology.
17
Doak (1997) glosses the second person nominal with a hyphen '-',
hyphen is not included here.
, for reasons of economy the
32
Table 5: Transitive object-subject combinations (Doak 1997:68 modified)
1s
subj
2s
3
1p
2p
nte
obj
1s
2s
3
)
)
1p
)
0
)
)
)
0
2p
Next we turn to a brief discussion of the genitive person marking morphemes.
1.2.3 Genitive person markers
The genitive person markers are used to indicate predicates of possession and indicate
second participants in a given set of detransitivized constructions. In the discussion that
follows only genitive constructions of the predicate possessive type will be addressed. All
genitive morphemes are suffixes with the exception of the first and second person
singular morphemes, which are prefixes. The genitive morphemes appear in Table 6
below.
Table 6: Genitive person marking morphemes
Singular
Plural
1g
2g
3g
Examples of the genitive morphemes follow in (10).
33
( 10 ) a.
1
$ 1
1g- thing
'It's mine.'
b.
1
$ 1
2g- thing
'It's yours.'
c.
1
$ 1
thing -3g
'It's his/hers.'
d.
1
$ 1
thing -3g -3pl
'It's theirs.'
e.
1
$ 1
thing -1pg
'It's ours.'
d.
1
$ 1
thing -2pg
'It's yours (pl).'
(Doak 1997:69-71)
When the possessed item is other than third person, a nominative subject occurs
with the genitive predicate. Doak (1997) notes that this suggests that possessive
constructions always include a pronominal subject, in the cases where the possessed is
34
third absolutive, as in (10) above, the null morpheme should be assumed to be present
(72). The following example (11) illustrates the nominative and genitive combination.
( 11 ) a.
&
$
2nom& 1g- father
2You are my father.'
b.
&
$
1nom& nom- woman -3g
'I am his wife.'
(Doak 1997:71-72)
Now we turn to the predicate pronominals.
1.2.4 Predicate pronouns
The following predicate pronouns (12) can stand alone as predicates or may be employed
as emphatic adjuncts. The constructions that follow are comprised of the nominative
morphemes and unanalyzable roots (Doak 1997:72-74).
( 12 ) Predicate pronouns in Cr
a.
*
'I/me.'
b.
c. )
d.
e.
f. ) -
*
%
'You.'
'He/him.'
'We/us.'
'You folks/All of you.'
'They/them.'
(Doak 1997:72-73)
In (f) the plural morpheme
mood in Cr.
is used. Now we turn to a brief discussion of aspect and
35
2. Sentential aspect, stem aspect, and mood
In this section I describe the three aspects, identified as sentential aspect, two identified
as stem aspect and one mood morpheme discussed. Here N. Mattina's (1996) term stem
aspect is employed, which parallels with what others have referred to as aktionsart or
lexical aspect (cf. Comrie 1976; Binnick 1991 for example). While the theoretical
approach of this dissertation does not follow N. Mattina's Lexicalist approach, the facts of
the morphemes described here parallel those of their cognates in Okanagan as described
by N. Mattina. It should be noted that aspect is more complex in Cr than presented here,
and that a full analysis of aspect is beyond the scope of the present dissertation. However,
I refer the reader to N. Mattina 1996 for a thorough account of aspect in a Southern
Interior Salishan language (Okanagan).
It is worth noting that N. Mattina's (1996) "properties of base, stem, and sentential
aspect in Okanagan" appear to parallel those of Coeur d'Alene (104). The qualities are
listed here, with the caveat that base aspect will not be addressed and that future inquiry
will be necessary to determine to what degree these generalizations hold in Coeur
d'Alene. In terms of the morphemes discussed here, the generalizations appear to hold,
but future research will be necessary to determine the exact nature of the aspect system in
Cr.
36
Table 7: N. Mattina's properties of aspect (1996:104 modified)18
base aspect
primitive situation type, inherent, ontological,
prototypical
stem aspect
derivational, serves to focus a temporal phase or
modality
sentential aspect
inflectional, sets the situation in a temporal frame
in sequence with other situations in the discourse
The discussion will begin with an overview of sentential aspect.
2.1. Sentential aspect
The three sentential aspects are: completive indicated by a null morpheme, customary
indicated by the prefix
, and continuous indicated by the
prefix morpheme. In the
examples that follow the aspect morpheme is underlined.
( 13 ) completive null ) : A situation has ended19
a.
&
) $
1s.nom& comp- smoke
'I smoked.'
(Doak 1997:83)
b. *
) $*
comp- call -t -1acc -2erg
'You called me.'
18
(Doak 1997:119)
N. Mattina (1996:103 f1) notes that her base, stem, and sentential aspect are the same categories as
Binnick's (1991) Aristotelian aspect, aktionsart, and aspect proper, respectively.
19
There is no evidence suggesting that the null morpheme appears to the left or right of the root, however, I
assume it patterns with the overt morphemes that align to the left of the root.
37
( 14 ) customary
: A situation that is viewed as characteristic of a whole period rather
than of a moment (Comrie 1976).
a.
(
&
$
1p.nom&
cust- work
'We work.'
b.
(Doak 1997:85)
3
$
cust- accompany -dt
-2acc -1p.erg
'We go with you.'
(Doak 1997:115)
: A situation in progress.
( 15 ) continuative
a.
* &
$*
1snom& cont- see
b.
'I am seeing.'
(Doak 1997:106)
Does not occur on transitive stems
(Doak 1997:44)21
Having presented the sentential aspect morphology we turn next to stem aspect.
2.2. Stem aspect
There are two significant morphemes that are glossed as stem aspect. Here what is
referred to as stem aspect parallels the notion of aktionsart or lexical aspect (c.f. Binnick
1991 for a comparison of aktionsart and lexical aspect). Comrie (1976:6 f 4) makes the
distinction between aspect and aktionsart as one of grammaticalization vs. lexicalization
(sentential aspect vs. aktionsart respectively). Within the DM framework the derivational
20
Doak (1997:115.116) glosses the object, , in this construction as ‘1erg’. The morpheme that
corresponds to ‘1erg’ however is . I take this to be a typographical error and list the gloss as ‘1p.erg’.
21
Doak notes that she has recently acquired “examples of the continuative occurring with transitive stems;
however, they are usually in subordinate clauses (one speaker seemed able to use this construction in main
clauses)” (pc).
38
morphemes associated with lexicalization in N. Mattina (1996) are considered part of the
syntactic mechanism that builds clauses and stems/words.
Unlike the sentential aspect morphemes, the stem aspect morphemes are affixes
that attach to the right edge of the root, with the exception of one morpheme, the
inchoative infix - -, discussed below. These morphemes parallel their Okanagan
cognates in function and form.
( 16 ) stative : indicating that "x has the quality y.”
)&
$4
3abs& be.good -stat
'He is good.'
(Doak 1997:44 modified)
( 17 ) inchoative and - - : indicates a change of state without reference to the act
leading up to the change.22
a.
&$
1sn& push –inch/inv
'I started to push (it).'
b.
( $
discover -inv/inch -ncr
'... it was discovered ...'
22
(Doak 1997:44 modified)
24
)
555
-(d)t -3abs -nte
(Reichard 19927-29: cosssd022)
N. Mattina (1996:87) refers to this morpheme and morphemes of this type as anticausatives, I use Doak's
(1997) term, though the morphemes are cognate. Doak (1997:44) refers to the -p morpheme as the
involuntary/inchoative.
23
Doak (1997:44) records & $
as
& $
with an extra " " before the root. I consider this
a typographical error.
24
It is not clear if this is the -nt or -t transitive as nasal reduction occurs when two nasals are adjacent in
Cr. The '( )' parenthesis indicate the unknown status of the 'n' element. It should be remembered that the
two morphemes appear to be alternate forms of one another.
39
c.
$
3abs& be. heated -inch'It became warm.'
)&
(Doak 1997:26 modified)
N. Mattina (1996:88 f 54) notes that the distribution of the infix
and affix
is in part
phonological. It must be left to future research to determine if this is true in Cr.
2.3. Mood
Here the irrealis
is introduced. This contrasts with the unmarked realis. Irrealis
indicates doubt, possibility, prediction and imperatives, depending on context. The
irrealis may appear in both transitive and intransitive constructions. There may be other
moods in Cr, but they will not be addressed here. The following example illustrates
irrealis.
( 18 )
$
irr nom- thing
'Whatever it was.'
The irrealis
diec
(Doak 1997:188)
will be discussed in greater detail, along with a number of other particles
in Cr in chapter 5.
3. Tense
The only tense that is marked in Cr is the future. If no future marking morphemes appear
in a construction it is deemed present or past based on context. There are four future
marking morphemes.
40
( 19 ) immediate future
: 'soon'
*
-
& $*
imfut
+,
6$
.,
2p.nom be.group_one.sack.full.of.meat
'You will soon each have one sack full of meat.'
(Reichard 1927-29:cssw243)
Doak (1997) identifies two other "immediate future" particles related to
and
intentional
( 20 )
:
(187). These forms may appear in conjunction with the future
.
'
1
&
$
imfut
fut 2nom int- first- eat
'Soon you are going to eat first.'
1
The future intentional
(Doak 1997:187)
indicates future intentional, permissive, mild request
(Reichard 1938:666-67). It occurs before the predicate in intransitive (21a) and transitive
(21b) constructions. In transitive constructions it is restricted to subordinate structures.
( 21 ) a.
4.
& $4.
fut 2nom love
'You are going to be loved.'
b.
7(
(Doak 1997:194)
,
$7
)
neg sub fut int- lose -dt -3abs -3erg
'Don't lose your man.'
$,
det1
2g- nom- man
(Doak 1997:198)
In chapter 5 these future particles will be discussed in greater detail. As Comrie
(1976:2 f2) notes, the future tense in many languages has modal as well as tense value. It
41
may be the case that the immediate future morphemes
,
and
are temporal adverbials (Cinque 1999) which pattern with the as yet discussed
discourse/narrative adverbial morphemes
and
2 both glossed 'then' and
'and/then' when together. These particles will be discussed in greater detail in chapter 5,
where the following particles will be analyzed.
Table 8: Particles analyzed in Chapter 5
TYPE
PARTICLE
Temporal Adverbial
'soon' immediate future (Doak 1997:186)
'soon' immediate future (Doak 1997:187)
'soon' immediate future (Doak 1997:187)
Sentential Adverbial
'and' / 'then' discourse/narrative adverbial
'and' / 'then' discourse/narrative adverbial
Mood
irrealis (Reichard 1938:669.777; Doak 1997:188)
Modal
future intentional, permissive, mild request
(Reichard 1938:666-67)
ought, obligation (Reichard 1938:669.780)
was to be but isn't, possibility (Reichard
1939:104)
Aspectual
'used to' terminative (Doak 1997:49)
'always' habitual (Doak 1997:49)
4. Transitivizing morphemes
There are six transitivizing suffixes that occur following the stem and preceding the
object. The structure of the transitivizing morphemes is presented in (22).
42
( 22 )Transitive structure
stem
object
subject
(
It will be noticed that the transitive paradigm includes the segment -t in each of the
morphemes. Cognates in other Salishan languages also include this segment. While some
scholars segment this as a morpheme (e.g. Carlson 1972; Thompson and Thompson
1992) others do not (e.g. A. Mattina 1973, 2001; Kuipers 1974). Here I follow A. Mattina
and Kuipers for reasons of notational simplicity. The specific function of each
transitivizer is addressed below with examples.
4.1. The lone -t and directive -nt transitivizers
The lone -t (t) and directive transitivizer -nt (dt) are the most commonly used in Cr.
Doak (1997) suggests that these two are alternate forms of one another with the lone -t
appearing on a limited number of roots (115). The person markers, which follow these
forms, are the same in function and form: object patients and subject agents. These
transitivizers indicate most often that the subject is an agent in control of his or her
actions. Examples follow, (23) with -nt (dt) followed by (24) -t (t). In (23e) an example
of the indefinite morpheme
is presented.
( 23 ) a.
$
)
cut
-dt -3abs -1perg
'We cut it.'
(Doak 1997:114)
43
b.
,
$,
)
fill
-dt -3abs -2perg
'You folks filled it up.'
c.
(Doak 1997:114)
+
$ .
shoot -dt -1acc -3erg
'He shot me.'
d.
(Doak 1997:114)
4
$4
irr gnaw -dt -2acc -1erg
'I will gnaw you.'
e.
25
7
$
(Doak 1997:114)
7
)
break -indef -rel -dt 3abs -3erg
'He broke something for someone.'
( 24 ) a.
(Doak 1997:66-67)
$
)
give
-t -3abs -nte
'He was given it.'
b.
(Reichard 1927-29:cssw259)
8
$
)
then back- loc- carry.on.back
-t -3abs -3erg
'Then he carried it back on his back.'
(Reichard 1927-29:cssw106)
Now the discussion to turns the causative transitivizer.
25
Reichard (1938:626, 565) provides the same example with the gloss "He broke it (someone else's
property).” She provides three other examples; these and Doak's one example provided above are the only
examples available regarding this construction that I am aware of. As mentioned above, this morpheme will
not be analyzed further.
44
4.2. Causative transitivizer
(ct)
(ct) has three primary functions. It can indicate causative
The causative transitivizer
constructions, customary aspect, or topical object constructions. Here I follow Doak
(1997:124) and use the label "causative" for all three functions throughout this
dissertation. As mentioned in 1.2.2, the unique set of M-initial objects for first and
second person singular replace the S-initial morphemes when the construction is with an
causative transitivizer. The following examples illustrate the
( 25 ) a.
transitivizer.
(
$
say
-ct
'I told you.'
b.
-2acc -1erg
(Doak 1997:126)
(
$
dir- go
-ct
-1acc -3erg
'He took me there.'
(Doak 1997:125-26)
It will not be attempted in this dissertation to address the many functions of the
causative transitivizer
. Future work on aspect should lead to a better understanding
of the exact nature and function of this complex morpheme.
4.3. Applicative transitivizers
There are three applicative transitivizers in Cr. These morphemes introduce a third
participant to the argument structure of the sentence. Further, they alter the role of the
morphosyntactic object. That is to say, the participant represented by the object person
marking morpheme (absolutive/accusative) serves as a possessor or dative with the
possessor applicative transitivizer
benefactive transitivizer
(pra) and as a beneficiary or dative with the
(bt). There is a third, much less frequent, applicative ( . I
45
have found no examples of this morpheme in the analyzed Reichard (1927-29)
manuscripts. Each applicative is introduced with examples below. It should be noted that
Cr marks only two arguments on the predicate via agreement person marking
morphemes, any third argument must be gleaned from context or discourse, or be added
in the form of an overt DP adjunct.
4.3.1 Possessor applicative
In transitive constructions with the possessor applicative
(pra), the object marking on
the predicate indicates the possessor, rather than the possessed. The following example
(26) illustrates. In (26a) only two arguments are indicated on the predicate. In (26b) the
case is the same, only two arguments indicated on the predicate, but the construction
carries reference to a third item, the thing being filled. In (26b) the ergative marks the
subject as expected, however, the object accusative refers not to the patient, but to the
possessor of the patient.
( 26 ) a.
b.
,
$,
)
fill
-ct
'I filled it.'
-3abs -1erg
,
$,
fill
-pra -2acc -1erg
'I filled it for you.'
(Doak 1997:144)
To further exemplify that it is the possessor that is marked on the predicate and the
possessed that is not, a construction with an adjunct DP is given (27), where the DP is
underlined.
46
( 27 )
-
--
$
$
take -pra -1p.acc -nte
det3 chief
-1pg
'Our chief was taken from us.'
['We were taken from (it was) our chief. ']
In some cases
(Doak 1997:146)
(pra) serves to indicate a dative construction. In these cases the
role of the object shifts to dative. In the following example the accusative morpheme in
the directive transitive (
) represents the object patient (28a), and the accusative
morpheme in the applicative transitive ( ) represents the object goal (28b).
( 28 ) a.
(
$
9 :
soon teach +rdp<aug>
'He will teach us.'
b.
-dt -1pacc -3erg
(
$
9 :
soon teach +rdp<aug> -pra -1pacc -3erg
'He will show us [how to do it].'
('He will teach x to us.')
(Doak 1997:149)
4.3.2 Benefactive applicative
In constructions with the benefactive applicative
, the argument indicated by the
ergative is the agent and that by the accusative/absolutive the beneficiary. This fact can
be illustrated with the following comparison of a simple transitive construction (29a) and
the benefactive (29b).
47
( 29 ) a.
*
$*
ask.for -t -1acc -3erg
'He invited me. '
b.
*
$*
ask.for -bt -1acc -3erg
'He begged something for me.'
(Doak 1997:153)
The benefactive applicative may also function to render objects recipients. The
following example illustrates.
( 30 ) a.
$
9 :
report +rdp<aug> -bt -1acc -2perg
'You folks told me a story.'
b.
$
sing -bt -2acc -1erg
'I sang to you.'
(Doak 1997:156)
4.3.3 Dative
As noted, this morpheme is extremely rare. It patterns with the applicative in that it
serves to introduce another participant into the clause structure. Doak (1997) notes that
the exact role of the third person is difficult to assess, as she has only a handful of
examples each with the third person or nontopic ergative person marking morphemes
(157). Examples follow in (31). In (31a) a simple directive transitive construction with
(dt) is contrasted with the dative
(dat) in (31b).
48
( 31 ) a.
. 4
$ . 4
)
look.at -dt -3abs -3erg
'He looked at it.'
b.
(Doak 1997:158)
. 4
$ . 4
)
look.at -dat -3abs -3erg
'He looked at it for him.'
In the next set of examples the dative
(32a).
( 32 ) a.
(Doak 1997:159-60)
(32b) is contrasted with the benefactive
+
$.
)
scorch -bt -3abs -3erg
'He burned it for him/somebody. '
b.
+
$.
)
scorch -dat -3abs -3erg
'He burned it for somebody. '
(Doak 1997:159)
Once all the Reichard (1927-1929) manuscripts have been completely analyzed, more
forms of
may be attested, allowing for greater understanding of this applicative
morpheme. Next, determiner phrases will be addressed.
5. Determiner phrases
What I refer to here as determiner phrases Doak (1997:214) refers to as adjuncts,
following Jelinek (1984 et series; Baker 1996). I use the term determiner phrase because
these structures are headed by a determiner, however, it should be noted that they are
considered adjuncts throughout this dissertation as Doak claims. Doak maintains that
49
these determiner phrases are optional structures used to more fully specify the referents
of pronominal arguments of the predicate, or used to introduce additional participants not
indicated by the pronominal arguments, or serving the function of prepositional phrases
(214). Unlike other Salishan languages (cf. Jelinek and Demers 1994:721; N. Mattina
1996:31 f4), it is not uncommon to find sentences that include two or more adjuncts or
adjoined clauses with unique reference in Cr.
These determiner phrases can be headed by one of three determiners in Cr;
(det1),
(det2),
(det3). Doak (1997) captures the difference between the three
morphemes by suggesting that a distinction between the three is based on distance:
near,
is far; and
is
is somewhere in between (215). Determiner phrases may also
appear without one of the three determiners when referring to a proper name only, though
proper names may also appear with a determiner as the examples below illustrate.
Determiner phrases are composed of a determiner followed by a fully inflected
predicate as in the intransitive construction in (33a) (where the predicate is preceded by
the discourse adverbial
, and transitive construction (33b). In all examples the
determiner phrase is underlined.
( 33 ) a.
- !
)&
$
!
then 3abs& went det3 Coyote
'Then Coyote went.' (lit. 'Then he went, the Coyote.')
(Reichard 1927-29:cssw11)
b.
$
irr
take -(n)t
)
$
-Ø -2erg
det1 wood
'Then you take a stick.' (lit. 'You take it, the stick.')
(Reichard 1927-29:cssw71a)
50
In general, but not in all cases, adjuncts refer to the absolutive argument in
transitive constructions. Determiner phrases in Cr are to a certain degree
nonconfigurational. In some instances only discourse can disambiguate adjunct reference.
This can be illustrated in the following example (34), where in an intransitive future
inversion construction (34a) two glosses are possible, and (34b) where either Cr example
can be glossed as the single English gloss. In the examples the adjuncts are underlined.
( 34 ) a.
:Ι
:Ι
; -,-,- (
$ ,
9#
$=
.*.<
;
$> .<*.<;
fut int- deceive +redup<ncr> -ncr -m -3g det1 Vinnie det1 Margaret
'Margaret's going to fool Vinnie.' (lit. 'Margaret's fooling will be of Vinnie.')
or
'Vinnie's going to fool Margaret.' (lit. 'Vinnie's fooling will be of Margaret.')
(Doak 1997:254)
b.
*
?
-
)
$*
$
3abs- cont- see
-cont
$?
det1 deer det1 Ben
or
*
-
)
?
$*
$?
3abs- cont- see -cont det1 Ben
'Ben is seeing the deer'.
$
det1 deer
(Reichard 1938:679)
These determiner phrases may be topicalized (Doak 1997:255) and appear
preceding the predicate as in the transitive construction in (35) where the sentential
adverbial hoi precedes the topicalized DP, and the DP refers to the absolutive object.
( 35 )
.
- -,
$
.,
4. + ,$4. 6$ . ,
then det3 Snipe
club_stick
'Then Snipe, he bludgeoned her.'
)
-dt -3abs -3erg
(Reichard 1927-29:cssw216)
51
In (36) an example of a topicalized DP in an intransitive construction is given.
( 36 )
,
$ ,
det3 meat
- 8
#
9$
rdp<dim>+object.lies -
'The body, it was small.'
(Reichard 1927-29:cssw141)
As mentioned, these determiner phrases will be assumed to be adjuncts throughout the
dissertation.
Finally, Doak (1997:214-215) notes that there is a determiner “homophonous with
the oblique”
. Doak notes elsewhere that predicates following this morpheme
indicate indefinite ergatives and absolutives (223), indefinite inanimate objects and
animate non-topic ergative subjects (232), and patients (241). Doak labels this
morpheme as “oblique” despite the apparent differences between its function and its
homophonous counterpart the true oblique. Doak claims there is no evidence that the two
morphemes are not the same (283:fn83). However, she does suggests that the oblique
is preceded by one of the previously mentioned determiners and that the determiner
is
not preceded by other determiners in transitive constructions (219,315). In intransitive
constructions Doak treats all instances
of as being oblique. In short, these morphemes
deserve further scrutiny to determine their true nature. Such an undertaking will not be
part of this dissertation. However, these morphemes will simply be glossed ‘ ’
throughout rather than “obl” or “oblique” as Doak glosses them. Where clarity demands
discussion of the morphemes to determine if the phonological form
oblique or an indefinite determiner, discussion will be presented.
represents a true
52
6. Summary
In this chapter, I have briefly described Cr clause structure. It has been shown that
syntactically intransitive constructions are minimally comprised of a subject marking
morpheme, and a root. Syntactically transitive constructions are minimally comprised of
a root, transitivizing morpheme, object marking morpheme, and subject marking
morpheme. Optionally, both intransitive and transitive clauses may include, aspect
marking morphology, a limited set of tense marking particles, mood particles, determiner
phrase(s), or any combination of these. In the next chapter a formal account of the basic
clause structure is proposed.
53
CHAPTER 3
CR BASIC CLAUSE STRUCTURE: A FORMAL ACCOUNT
1. Introduction
The goal of this chapter is to present an analysis of the basic clause structure of transitive
and intransitive constructions in Coeur d'Alene (henceforth Cr), and provide a formal
account of the surface structure morpheme ordering. The basic clause structure is
illustrated for the intransitive and transitive clause, (14) above.
( 37 ) a.
intransitive
*
$*
1pnom
cust- work
subject- aspect- root
'We work.'
b.
(Doak 1997:85)
transitive
$
cust- accompany -dt
-2acc -1p.erg
aspect- root
transitivizer object subject
'We go with you.'
(Doak 1997:115)
The independent phrasal elements (DP) that Doak (1997) treats as adjuncts, and which
may serve as what Kiss (1995:7) has described as notional subjects and/or notional
objects, will not be discussed. Throughout this chapter it will be assumed that these DPs
are not generated in argument positions as Doak suggests. In this way Cr is nonconfigurational in the sense of Baker (2001) and DP subjects and objects are adjuncts.
54
In the first half of this chapter it will be argued that the person marking
morphemes in Cr are not agreement morphemes (Kroeber 1996; among others), but
rather bound pronouns in the sense of Jelinek 1984. It will be demonstrated that an
account of bound pronouns is preferable to a radical pro drop account (Baker 1996) as it
provides a more elegant and simplified account of the Cr grammar. Further, Hale and
Keyser's (2002) Conflation as head-movement (Harley 2004) will be employed in the
analysis to maintain Chomsky's (1995) claim that head-movement is phonological in
nature.
In the second part of this chapter it will be argued that the bound pronouns are
independent -pronouns (Déchaine and Wiltschko 2002) generated in argument
positions. Further, a possible diachronic account of the bound pronouns will be presented
exploiting Speas’ (2004) notion of lexicalized agreement. A diachronic argument that
agreement morphemes were grammaticalized, in the terms of Newmeyer (1998), and thus
“upgraded” to bound pronouns historically will be considered. However, this diachronic
origin of bound pronouns will be rejected. It will instead be claimed that cross-family
variation can be accounted for if it is assumed that bound pronouns have been the norm
diachronically.
Before moving forward, a few notes regarding terminology are necessary. What
has been thus far identified as ‘agreement morphology’, or ‘person marking morphemes’,
in Cr will be referred to as ‘bound pronouns’ in anticipation of the conclusion of the
argument. Despite the leading nature of the terminology, no commonly used phrase for
such markers is neutral with respect to an analysis. Before my argument is established,
however, my use of the term should not be taken as presupposing the analysis. Also,
while many authors acknowledge Jelinek's origination of an analysis of agreement
morphology as base generated in argument positions (cf. Doak 1997; Butt 2001; Hale
55
2000; Haugen 2004, 2006; among others), and others do not (cf. Vainikka and Levy
1999; among others), it will be assumed that these analyses capture Jelinek's (1984, 2004,
2006; Jelinek and Demers 1994) claim that previously analyzed agreement morphemes
(e.g. previously analyzed as AGR-heads, etc,), are bound pronouns in a number of
languages. For clarity, when referring to such works the term ‘bound pronouns’ will be
employed to refer to any such agreement element, whether referred to as incorporated
pronouns (Haugen; Butt), pronouns (Doak; Jelinek; Jelinek and Demers), bound
pronouns (Bhat 2004), or as functioning as pronouns (Vainikka and Levy 1999).26 It
should also be noted that no claims regarding Jelinek's Pronominal Argument Parameter
will be made in the following discussion, as Jelinek's parameter is a macro-parameter,
which involves other considerations beyond argument structure (cf. Jelinek 2006 for
discussion). However, key elements of the Pronominal Argument Parameter, specifically
a ban on DPs in argument position, will be addressed in terms of micro-parametric
variation.
Further, in the discussion involving what has been traditionally termed pro drop,
the terms null argument(s), null subject(s), and null object(s) will be used. In recent
years there has been much debate as to what constitutes a null argument (cf. @3.3.1 and
references therein), and these terms will be employed for clarity. In the case of radical
pro drop, the term used for languages such as Chinese and Japanese, which allow nearly
all arguments to be dropped, the term radical argument drop will be used instead.
The chapter is organized as follows. In Section 2 Hale and Keyser's (2002)
Conflation is introduced as the mechanism by which head-movement is instantiated in a
26
It should be noted that the term bound pronoun is employed to capture the generalization that pronouns in
these discussions are somehow bound. It should not be construed as a term that attempts to capture the
syntactic mechanism involved in the various discussions noted in the references here. That is, it should be
recognized that while an incorporated pronoun is in a sense "bound", the syntactic operations and outcomes
that constitute the "binding" are different and lead to different results especially in terms of morpheme
ordering.
56
Minimalist syntactic framework (Harley 2004). In Section 3 two potential analysis of Cr
clause structure are presented, a bound pronoun analysis and a radical argument drop
analysis. Section 3 will further present various arguments against the radical argument
account. Section 4 presents a brief interim summary, and Section 5 presents the claim
that bound pronouns in Cr are -pronouns (Déchaine and Wiltschko 2002). In Section 6
a potential diachronic genesis of bound pronouns is presented and then rejected, and
finally, concluding remarks are found in Section 7.
2. Conflation
Hale and Keyser (2002) define Conflation as:
( 38 ) Conflation
Conflation consists in the process of copying the p[honological]-signature
of the complement into the p-signature of the head, where the latter is
"defective.” (63)
Hale and Keyser define a p-signature, or p-sig, as potentially being defective in one of
two ways. A defective p-sig can be either a head with no phonological content, that is the
p-sig is empty containing no phonological features, or it may be specified that the head is
an affix (63). It is important to note that Hale and Keyser make a distinction between
"null" morphemes, such as in the case of pro or PRO, which have non-defective
phonological features, and "phonologically empty" morphemes that have defective
phonological features. The difference is that a "null" morpheme, like pro, is comprised
of a p-sig that is not defective, while the "phonological empty" morpheme will comprise
a p-sig in need of phonological content.
Hale and Keyser argue that Conflation operates in tandem with Merge. They say:
57
We would like to take seriously the idea that Conflation is a
concomitant of Merge, the operation which is fundamental in
defining the projection of syntax from the lexicon (Chomsky 1995)
... To say that Conflation is a concomitant of Merge is to say that
it is in some intimate manner bound up with Merge, that is a part
of Merge in some sense. (60-61)
For Hale and Keyser Conflation occurs at Merge, and is the process of transferring
phonological material from a phonologically complete complement to a phonologically
defective head. That is, Conflation is the process of a head X merging with a
complement Y (where Y is a maximal projection) whose label contains all the features of
Y0 including phonological features, or its p-sig. If X's p-sig is defective, at Merge Y's psig is conflated into X's. This allows X to now be pronounced with Y's phonological
features. It is important to note that this feature does not motivate movement of the head,
rather it motivates transfer of only the p-sig in the Hale and Keyser formalism at Merge.
As Harley (2004) notes, this is crucially different from discussions of the Headmovement Constraint and triggers for head-movement that appeal to [±affix] feature as a
trigger for movement (5 fn5). The key assumptions of Hale and Keyser's Conflation can
be summarized as follows:
( 39 ) Conflation: Key Assumptions
a.
A label of any constituent has ALL the features of the head, including some
representation of a phonological matrix, (the 'p-sig' of the head).
b.
c.
Conflation occurs when a constituent is merged with a sister head whose
p-sig is 'defective'. The p-sig of is merged into the p-sig of .
For economy reasons, the conflated p-sig only will be pronounced once, in its
uppermost position.
(Harley 2004:3)
Harley (2004) proposes that Conflation is the mechanism with which ‘head-
58
movement’ phenomena should be analyzed within the Minimalist program. Following
Harley, Conflation is here construed as Head-movement resulting from Merge. That is,
when a head and a complement merge, a label is projected and simultaneously, the
complement’s p-sig transfers to the defective p-sig of a given head. The following
Mohawk example taken from Harley (40) illustrates this operation.
( 40 )
Owira'a
waha'-wahr-ake'
Baby
AGR-meat-ate
‘The baby ate meat.’
(Baker 1988 in Harley 2004)
As Harley notes, in the numeration for VP we start with the roots, [N wahr-], 'meat'
(N wahr-) and [V -ake], 'eat' (Vake-).27 The derivation is outlined in (41). To initiate
Conflation, we assume Vake- has a 'defective' p-sig.
( 41 ) a.
b.
c.
d.
N wahr- merges with Vake'Because Vake'- has a 'defective' p-sig, the p-sig of Nwahr- conflates into the psig of Vake'The head, now with the p-sig Vwahrake', projects (i.e. is used for a label) giving
the set {Vwahrake', {Vwahrake', N wahr- }}, which can be illustrated in the following
tree:
Vwahrake'
Vwahrake'
N wahrwahrake'
For economy reasons (because Vwahrake' is pronounced), N wahr- is not
pronounced.
In this way head-movement remains in narrow syntax and Chomsky's (1995)
generalization that head-movement is phonological in nature is maintained (cf. Chomsky
27
Following Harley (2004:3) I will "represent heads as a syntactic category label, subscripted with their
phonological realization intended to represent the p-sig of the head." As I follow the 'interpretive'
morphology of the Distributed Morphology theory, with Late Insertion, p-sigs actually represent
positions-of-exponence, waiting for Vocabulary Insertion to fill them in.
59
1995:321, 2001a 37; and Harley 2004 for discussion). Throughout the remainder of this
dissertation head-movement will be construed as Conflation as outlined here.
Next we turn to two competing analysis of Cr basic clause structure.
3. Null arguments and bound pronouns
In this section two competing analysis of Cr clause structure will be presented. In
Section 3.1 the bound pronoun analysis, adopted in this dissertation, is presented. In
Section 3.2 a radical argument drop (Baker 1996; among others: cf. Neeleman and
Szendröi 2005, 2006 for discussion of radical pro drop) analysis is presented. In Section
3.3 it is argued that licensing and identification, learnability, and typological
considerations suggest the bound pronoun analysis is preferable to a radical argument
drop analysis.
3.1. Bound pronoun analysis
Following Jelinek (1984; Bhat 2004), a bound pronoun analysis is presented here.
Assuming that bound pronouns are base generated in argument position, a rather
straightforward account of the facts emerges. Given the transitive construction in (42a),
(37b) above, the following tree (42b) illustrates the derivation outlined in (43).
( 42 ) a.
Cr basic transitive construction
$
cust- accompany -dt
'We go with you.'
-2acc -1p.erg
60
b.
TP
T
ASPP
ASP
vP
P
P
v
$P
$
( 43 ) a.
b.
The root
P
'accompany' merges with its complement the P object
'2acc'.
Neither p-sig is defective, no conflation occurs. The P is labeled with the psig of its head,
.
The P labeled
merges with an element from the numeration, the v head
'directive transitive' which has a defective p-sig and conflation occurs.
The p-sig of P,
, is copied into the defective p-sig of v, giving
.
Here we note that the v head is realized to the right of the phonological
material in P. The assumption here is that there are two types of affixes:
those with defective p-sigs and those without (the bound pronouns for
example). Affixes with defective p-sigs phonologically align with conflated
phonological material in accord with their morphological affix feature
c.
d.
e.
([±prefix]). The vP is labeled with the p-sig of its head,
.
The v labeled
mergers with the subject, a '1p.erg' -head. Neither
p-sig is defective, so no conflation occurs.
The object raises to a second specifier position of v, to check case
(Chomsky 1994, 2001a; Hornstein, Nunes, and Grohmann 2005). Neither psig is defective. The whole constituent is labeled with the p-sig of the head v,
.
The vP
merges with an element from numeration, the ASP head
,
'customary'. This element's p-sig is defective, and conflation occurs. The
p-sig of the vP,
, is copied into the defective p-sig of ASP, giving
61
. Then, the whole constituent, a projection of ASP, is labeled with
the p-sig of its head,
.
The ASPP merges with an element from numeration, a null T head. Neither
p-sig is defective, and no conflation occurs. The T projects and is labeled
f.
g.
with the p-sig of its head.28
The subject checks case via agree with the T head, as the probe/goal
relationship cannot be satisfied by any element above the subject, and attaches
to the verb-word sitting in Asp° (a process of cliticization).
In this way we capture the facts of the surface morpheme order in the basic clause
structure of Cr rather straightforwardly, without deviation from the tenets of the
Minimalist Program of Chomsky (1995, 2000, 2001a, 2001b). For reason of space,
intransitive and ditransitive29 constructions will not be discussed here, but to see that the
generalizations hold see Chapter 4 for examples of intransitive and ditransitive
constructions under the above analysis. In the next section we see that a radical argument
drop analysis raises numerous thorny theoretical issues that the present analysis is able to
sidestep entirely.
3.2. Radical argument drop analysis
Under an analysis which characterizes the bound pronouns as agreement morphology
(generated perhaps as AGR-heads and agreeing with null pronouns base-generated in
argument position), Cr fits the descriptive generalization for radical argument drop in that
with the exception of the first and second person intransitive constructions, all arguments
are dropped. That is, under such an analysis pro or the first or second person clitics
would be the only elements base generated in argument position. In this way it fits with
28
In Chapter 5 it is demonstrated that a series of particles occur between AspP and TP, providing evidence
that the predicate does not raise above ASPP.
29
Here I use the term "ditransitive" for reasons of economy to refer to the applicative and benefactive
constructions in Cr.
62
Chinese, Japanese, and other languages described as radical argument drop languages
(Neeleman and Szendröi 2005, 2006 and references therein). This was illustrated in
Chapter 2 in the discussion of bound pronouns (person marking morphemes). Given the
transitive construction in (44a) and the corresponding tree in (44b), (42) above, the
derivations outlined in (45) would represent a plausible instantiation of the radical
argument drop analysis for Cr.
( 44 ) a.
transitive
$
cust- accompany -dt
-2acc
-1p.erg
'We go with you.'
b.
TP
AGRSP
AGRS'
pro
AGRS
ASPP
ASP'
ASP
AGROP
pro
AGRO'
AGRO
vP
pro
pro v
$P
pro
( 45 ) a.
The root
'accompany' merges with its complement the pro object.
63
Neither p-sig is defective, no conflation occurs. The P is labeled with the
b.
c.
p-sig of its head,
.
The P labeled
merges with an element from the numeration, the v head
'directive transitive' which has a defective p-sig and conflation occurs.
The p-sig of P,
, is copied into the defective p-sig of v, giving
. The
vP is labeled with the p-sig of its head,
.
The v labeled
mergers with the subject, pro. Neither p-sig is defective,
so no conflation occurs.
d.
The object pro raises to a second specifier position of v to check accusative
case features. Neither p-sig is deficient. The whole constituent is labeled
with the p-sig of the head v,
.
e.
The vP
merges with an element from numeration, the AGRO head,
'2acc', which has a defective p-sig and conflation occurs. The p-sig of vP,
, is copied into the defective p-sig of AGRO, giving
. The
AGROP is labeled with the p-sig of its head,
.
f.
The AGROP
merges with an element from numeration, the ASP head
, 'customary'. This element's p-sig is defective, and conflation occurs.
The p-sig of the AGROP,
, is copied into the defective p-sig of ASP,
giving
. Then, the whole constituent, a projection of ASP, is
g.
h.
,
labeled with the p-sig of its head,
.
The ASPP merges with an element from numeration, the AGRS head , '1erg'.
This element has a defective p-sig, and conflation occurs. The p-sig of
AspP,
, is copied into the defective p-sig of AGRS, giving
AGRS projects and is labeled with the p-sig of its head,
.
The AGRSP merges with an element from numeration, a null T head. Neither
p-sig is defective, and no conflation occurs. The T projects and is labeled
with the p-sig of its head. 30
Problems with the tree in (43b) can be divided into two parts: theoretical and empirical.
Empirically, there is no evidence that movement of the predicate occurs above ASPP (cf.
30
For brevity I do not include mention of the subject and object pro raising to SpecAgrP, see Holmberg
(2005) for an analysis of pro movement within the Minimalist Program. Holmberg argues that within a
Minimalist analysis pro must transfer agreement features to the agreement morphemes, not vice-versa as
previously argued (see Jaigli and Safir (1989) among others).
64
Chapter 5 for discussion of particle heads which appear higher in the structure than the
predicate), thus a reformulation of the AGRPs in the functional structure, taking the facts
outlined in Chapter 5 into consideration, would be necessary for any radical argument
drop analysis. Again, since this is not the analysis adopted here no such reformulation
will be proposed.
Theoretically, comparing the trees in (42b) and (44b), and the derivations that
derive them, it can be seen that the bound pronoun theory is a much better fit with
Chomsky’s 1994 Elementary Principles of Economy.
( 46 ) Elementary Principles of Economy
a.
Add optional
to numeration only if it has an effect at the interface.
b.
At each stage of a derivation, apply the most economical operation that leads
to convergence. (367)
Under the radical argument drop theory, pro and AGRP combined serve to perform the
same functions as the bound pronouns in the bound pronoun analysis. Further, in post
1995-Minimalism, AGRPs are considered to be unmotivated.
In the next section, it will be argued that the radical argument drop analysis poses a
number of complications that do not arise under the bound pronoun analysis.
3.3. Bound pronouns over radical argument drop
In this section it will be argued that the radical argument drop account becomes
problematic when considerations regarding the following are taken into account: (1) the
exact nature of the null arguments (are they pro?, elided NPs?, empty nouns/, etc.); (2)
licensing conditions and identification; (3) learnability; and (4) cross-linguistic
typological facts. Each of these will be addressed separately in this section.
65
3.3.1 Null argument in Cr: What are they?
For any argument that treats bound pronouns in Cr as agreement morphology, null
arguments in the following constructions must be fully identified in terms of features,
function, and the lack of phonological content.
( 47 ) a.
b.
c.
e.
intransitive 3rd person constructions (assuming 1st and 2nd person intransitive
person marking morphemes are clitics base generated in argument positions)
transitive 1st, 2nd, and 3rd person constructions
ditransitive 1st, 2nd, and 3rd person constructions
1st, 2nd, and 3rd person genitive constructions
It is simply not enough to claim that pro is generated in argument positions (Baker 1996;
Davis and Matthewson 2003 for Salish), as cross-linguistically this has been a
problematic claim. When the precise nature of the pro proposal is investigated, wildly
varying proposals result. For example, Zushi 2003 (among others) claims that Japanese
null arguments are pro, while this claim is countered by Oku (1998), Saito (2003),Sato
(2006), and Sato and Ginsburg (forthcoming) (and reference therein) with the claim that
the null arguments are recycled arguments. Further, Huang 1984 argues that null objects
are variables in Chinese, while Neeleman and Szendröi 2005 make the claim that they are
null pronouns. In addition, as Satterfield 2003 notes, contra Rizzi 1997b, "pro Catalan
equal pro Italian, equals pro cross-linguistically" is not the case, pro is invariant crosslinguistically (225). In addition, it has been argued that pro does not exist (Panagiotidis
2003). In (48) various labels that have been employed to describe null arguments are
listed, though the list is not exhaustive by any means.
66
( 48 ) accounts of null arguments cross-linguistically
a.
b.
c.
d.
e.
f.
pros with various content (Alexiadou and Anagnostipoulou 1997; Zushi 2003
(proposes three types of pro with different syntactic features); among others)
variables (Huang 1984)
empty nouns (Panagiotidis 2003)
null pronouns in terms of Chomsky 1982; Rizzi 1986; Holmberg 200531
recycled arguments (Oku 1998, Saito 2003,Sato 2006 and references therein;
Sato and Ginsburg forthcoming)
bare NPs (Hoji 1998; Tomioka 2003)
Consequently, it seems clear that any proposal which can be implemented without
invoking such a problematic concept is to be preferred over a proposal which requires it.
In short, the radical argument drop analysis is challenged by a need to identify the null
arguments in terms of syntactic features. This problem does not arise for the bound
pronoun analysis. Further, it would be necessary for any radical argument drop analysis
to account for null 3rd person agreement morphology as well as null arguments.32 This
additional problem regarding null agreement does not arise in the bound pronoun
analysis.
Another problem for the radical argument analysis is the issue of how null
arguments are licensed and how null arguments are identified in terms of reference. This
issue is addressed in the next section.
3.3.2 Licensing and identification considerations
Over the past twenty-plus years there have been a number of claims regarding the
licensing of null arguments and the recovery of a null argument’s identity, or reference.
Some of the claims put forth in the Generative tradition are listed in (49). It should be
31
Cf. Holmberg 2005 for a general discussion of how the three differ.
Baker 2002 provides an account of null 3rd person agreement, however, it crucially relies on his macro
parameter, the Polysynthesis Parameter.
32
67
noted that many of the claims put forth regarding null arguments have focused on null
subjects only, this is indicated by simply stating "null subjects" rather than "null
arguments.”
( 49 )a.
b.
c.
d.
c.
d.
null subjects occur in languages with uniform agreement paradigms (Jaegli
and Safir 1989; Speas 1993, 2004; among others - but cf. Neeleman and
Szendröi 2005, 2006; Vainikka and Levy 1999; Ackeman and Neeleman
2006; Butt 2001; Haegeman and Ihsane 2001 among others for arguments
against this claim)
agglutinating languages allow radical argument drop (Neeleman and Szendröi
2005, 2006)
EPP features on T license null arguments (Zushi 2003; cf. Silva-Villar 1998
for a similar analysis involving the licensing of multiple specifiers of C)
Radical argument drop occurs in languages with bare NP arguments (Tomioka
2003: cf. Neeleman and Szendröi 2005, 2006 for arguments against and Zushi
2003 for further discussion of)
null arguments occur in languages where AGR can be base generated in
argument positions (Jelinek 1984; Doak 1997; Vainikka and Levy 1999; Butt
2001; Hale 2000; among others)
null arguments occur in languages without overt expletives (cf. Silva-Villar
1998 for discussion and arguments against, also Vainikka and Levy 1999;
Haegeman and Ihsane 2001; Holmberg and Nikanne 2002; Holmberg 2005;
among others for arguments against)
The following typological considerations illustrate that many, in fact all of the
generalizations that are reflected in (49) do not hold cross-linguistically.
68
Table 9: Some cross-linguistic facts regarding null arguments33
Afrk (-)
Chn (s,o)
BEngl (rs)
Cr (s,o)
EMD (rs)
Finn (s)
Frn (ro)
Galac (s)
Gale (s)
Heb (rs,o)
Ice (s)
Ital (s,ro)
Jpn (s,o)
Krn (s,o)
BPort(s,o?)
EPort (s)
Occit (s)
Rus (-)
Sist (s)
Swd (-)
UAGR
= not identified here,
−
−
EXPLTV
−
−
AGGLUT
−
−
FUSN
−
−
−
−
−
−
−
−
−
−
= language has property, − = language doesn’t have property
(-) = doesn’t allow null arguments; (s,o) = allows null subject and object; (s) = allows
null subject; (o) = allows null object; (r) = restricted; UAGR = uniform agreement
33
LANGUAGES AND SOURCES: Afrk = Afrikaans (Neeleman & Szendröi 2005); Chn = Chinese (Huang
1984; Zushi 2003; Neeleman & Szendröi 2005 – NOTE Zushi and Neeleman and Szendröi analyze null
objects differently than Huang 1984); BEngl = British English (Haegeman and Ihsane 2001); Cr = Coeur
d’Alene (here I consider the 1st and 2nd person clitics fusional); EMD = Early Modern Dutch (Ackeman
and Neeleman 2006); Finn = Finnish (Holmberg 2002, 2005; Neeleman & Szendröi 2005); Frn = French
(Rizzi 1997b); Galac = Galacian (Raposo & Uriagereka 1990; Rizzi 1997b); Gale = Galegan (Silva-Villar
1998); Heb = Hebrew (Rizzi 1997b; Vainikka and Levy 1999 – note Rizzi analyzes Hebrew as having pro
and Vainikka and Levy analyze Hebrew as having bound pronouns); Ice = Icelandic (Rizzi 1997b:277;
Silva-Villar 1997:250 – NOTE Roberts 1997:153 presents Icelandic as not a pro drop language); Ital =
Italian (Rizzi 1986, 1997b; Zushi 2003; Neeleman & Szendröi 2005- NOTE Italian object pro limited to
specific construction where an overt pronoun is not possible – cf. Rizzi 1986 for details); Krn = Korean
(Neeleman & Szendröi 2005); Occit = Occitan (Silva-Villar 1998); BPort = Brazilian Portuguese
(Rohrbacker 1994; Vainikka & Levy 1999 – NOTE Huang 1984: 541 analyzes BPort pro drop similar as
similar to Chinese – William Alexander (pc) notes that BPort is no longer a pro drop language); EPort =
European Portuguese (Rohrbacker 1994 Speas 1994; Vainikka & Levy 1999; Silva-Villar 1998); Rus =
Russian (Speas 1994; Vainikka & Levy 1999); Sist = Sisternian (Silva-Villar 1998); Swd = Swedish (Speas
1994; Neeleman & Szendröi 2005; Hulden pc)
69
paradigm; EXPLTV = expletive (overt); AGGLUT = agglutinating case or number etc; Fusn
= fusional for case
In short, if an appeal to the radical argument drop analysis were made in Cr, the
licensing and identification of the null arguments, and null third person agreement
morphology, would have to be accounted for-Not an easy task as Table 9 and (47)
illustrate. This issue is not as problematic for the bound pronoun analysis proposed, as
the discussion of learnability and typological considerations illustrate next.
3.3.3. Learnability considerations
Under the radical null argument analysis Cr children would have to learn a hybrid system
of argument marking, while under the bound pronoun analysis children learning Cr as a
first language would learn a unified system of argument marking. In what follows, it is
argued that the unified system of the bound pronoun analysis is preferable in terms of
learnability and Occam's Razor, given the known facts.
3.3.3.1 Null argument analysis: The hybrid system
Under the radical null argument analysis learners must come to terms with two divergent
systems of argument structure, clitics and agreement, and acquire null arguments as well
as null agreement.34 Recall that DPs and NPs are understood to be optional adjuncts and
never occur in argument position. Thus, in intransitive constructions the learner must
learn mandatory overt first and second person clitics and a mandatory null third person
clitic, in addition to mandatory null agreement morphology in the first second and third
person as in (50). 35,36 The learner must learn which affixes are arguments and which are
34
For reasons of economy the non-topic ergative and imperative paradigm and the third person plural
agreement affix
will not be discussed; however, this will not affect the analysis presented.
35
It is assumed following Speas (1994, 2006) and Vainikka and Levy (1999) that AGR must be licensed in
intransitive constructions in a radical argument drop analysis, even though there is no phonological
evidence for its existence.
70
agreement.
( 50 ) a. subject intransitive clitics
b. intransitive agreement
singular plural
st
1
nd
2
rd
3
)
)
1
st
2
nd
3
rd
singular plural
)
)
)
)
)
)
In the case of transitive and ditransitive constructions the learner must come to terms with
mandatory null arguments, represented here as null pronouns, and mandatory overt first
and second person agreement morphology and mandatory null third person agreement
morphology as in (51).
( 51 ) a. object trans/ditrans pronouns
b.
singular plural
st
1
nd
2
rd
3
trans/ditrans agreement
st
)
)
1
)
)
2
)
)
singular
"
nd
3rd
plural
"
)
)
In the case of genitive constructions the learner must acquire mandatory null arguments,
again represented by null pronouns, and mandatory overt first, second, and third person
agreement morphology as in (52).
( 52 ) a. genitive pronouns
1st
nd
2
3rd
36
singular plural
)
)
)
)
)
)
b. genitive agreement
1st
singular plural
2nd
3rd
Another possibility would be to analyze the first and second clitics as agreement morphology. However,
see Davis (2000) for remarks on Proto-Salish subject inflection and potential problems for such an analysis.
71
In conclusion, under the hybrid system of the radical argument drop analysis the
learner has a number of tasks, not the least of which is distinguishing between agreement
morphology and clitics in intransitive constructions. What the child must learn is outlined
in (53).
( 53 ) The hybrid system: What must be learned
(i) distinction between agreement and first and second person clitics
(ii) first and second person clitics overt in intransitive constructions
(iii) third person clitics null in intransitive constructions
(iv) agreement morphology null in intransitive constructions
(v) all arguments null in transitive, ditransitive, and genitive constructions
(vi) third person agreement null in transitive and ditransitive constructions
(vii) third person agreement overt in genitive constructions
Next the unified system is presented and it is argued that it supports the claim that the
bound pronoun analysis is preferable in terms of minimality.
3.3.3.2 Bound pronoun analysis: The unified system
Under the unified system the learner is presented with what appears to be a much simpler
task: learning three distinct pronoun paradigms. Under this analysis the learner need not
distinguish between agreement and clitics, nor learn a variety of null arguments. The
unified system requires the learner to simply learn that third person arguments are null in
all cases but the genitive, as illustrated in (54).
72
( 54 ) (i) intransitive
st
1
singular plural
1st
singular
2
2
)
)
3rd
plural
"
nd
nd
3rd
(ii) transitive / ditransitive
1st
singular plural
2nd
"
)
(iii) genitive
)
3rd
Assuming the Cue based model of parameter setting (Lightfoot 1997a, 1997b;
Dresher 1999, 2003a, 2003b), the unified system would be quite preferable over the
hybrid system. Under the unified system i, ii, and iii in (55) could be the cues for the
setting of a potential parameter for null 3rd person or the accessing of an element
available in UG, a null pronoun, via a principle. That is, it may not be a ‘parameter’, but
rather a null morpheme in the given cell provided by UG that is accessed via the
necessary “cue(s).”
( 55 ) Cues for null third person
i. mandatory pronouns would be the cue that pronouns must be in argument
positions
ii. DP and NP subjects and objects never occur in argument positions, a further
cue that pronouns are arguments
iii. no overt phonological arguments necessary in third person intransitive,
transitive and ditransitive constructions, the cue that the morpheme is null
Under the hybrid system the learner would need a number of cues to arrive at the proper
argument paradigms. The conclusions arrived at are stated in (56).
( 56 ) Conclusion:
• Learnability considerations suggests bound pronoun analysis (unified system)
preferable to radical argument drop analysis (hybrid system)
• Occam's Razor, given the known facts, would lead one to choose a single
system (bound pronouns) over a system that requires multiple systems
(radical argument drop)
73
Finally, we turn to typological considerations and how the two analyses position Cr in
terms of other languages of the world typologically.
3.4. Bound pronouns vs. agreement: Some typological considerations
In the discussion above regarding generative claims about licensing and identification of
null arguments, it was demonstrated that none of the typological generalizations held
entirely. However, Neeleman and Szendröi's 2005, 2006 claim that agglutinating
languages allow radical pro drop has the greatest coverage cross-linguistically (as
discussed below). Neeleman and Szendröi (2006) employed the World Atlas of
Language Structure (Haspelmath, Dryer, Gill, and Comrie 2005) and were able to search
hundreds of languages for counter examples to their claim.37 In that search they found
only ten languages that raised potential challenges to their generalizations (46).
However, they arrived at analyses of these ten languages that supported their claims (cf.
46-51 for discussion). In short, the typological generalization arrived at by Neeleman
and Szendröi are the strongest to date regarding radical argument drop. If the radical
argument drop analysis is correct, it would be expected that Cr would fit the typological
generalizations arrived at by Neeleman and Szendröi. As will be demonstrated, this is
not the case. On the other hand, under the bound pronoun analysis Cr fits rather
straightforwardly into the typology of Neeleman and Szendröi.
Neeleman and Szendröi (2005, 2006) argue that languages like Japanese which
have agglutinating case (individual case marking morphemes such as -ga 'nominative'
and -o 'accusative'), that attaches to pronouns and other arguments, allow radical
argument drop. Languages such as English which have fusional case (he vs. him etc.),
allow either no argument drop or allow context sensitive argument drop (such as in
37
In fact, of the dozens of languages they analyzed prior to the search of the World Atlas of Language
Structure, only one language was identified as not fitting within their typological generalizations, Finnish.
74
Italian). For languages like Chinese which allow radical argument drop but are indifferent
to case marking (case is not marked), other agglutinating elements such as number place
it typologically with Japanese. Neeleman and Szendröi’s generalizations are illustrated in
(57).
( 57 ) Neeleman & Szendröi 2005:98 Typology38
PROPERTIES OF THE PRONOMINAL PARADIGM
INVARIANT FOR CASE
AGGLUTINATING
CASE
RADICAL
PRO DROP
NO
RADICAL
PRO DROP
Japanese
Korean
Burmese
Assamese
Hindi/Urdu
Turkish
agglutinating
for number,
etc
invariant or
fusional for
number, etc
FUSIONAL
CASE
Chinese
Kokota
Cheke Holo
Papiamentu
Tok Pisin
Jamaican
Creole
Italian
Pashto
Greek
BEnglish
Afrikaans
Swedish
Dutch
CONTEXT
SENSITIVE
PRO DROP
NO
PRO DROP
Not only does the typology account for a great number of languages, it also
clearly distinguishes Chinese- and Japanese-type argument drop from Italian- and
Greek-type argument drop. Further, allowing for different motivations for argument drop
cross-linguistically allows for the conflicting accounts of argument drop seen above in
@3.3.1 and @3.3.2. In short, analyses such as that of Neeleman and Szendröi (2005, 2006)
allow for a clear distinction between languages capable of dropping nearly every
argument (Chinese, Japanese etc.) and languages that are far more restrictive in allowing
38
The typology has been modified to reflect Haegeman and Ihanse's 2001 claims regarding British
English.
75
arguments to be dropped, “context sensitive” (Spanish, Greek, Hebrew, British English,
etc.). Yet, Neeleman and Szendröi do not place restrictions on accounts of context
sensitive argument drop, allowing for argument drop to be analyzed as deriving from
similar or dissimilar grammatical mechanisms between any given context sensitive
language. For example, a claim that Hebrew and Standard Finnish argument drop are
derived from the same grammatical mechanisms (Vainikka and Levy 1999), would not
conflict with Neeleman and Szendröi's claims. Having presented the generalizations
arrived at by Neeleman and Szendröi we turn to Cr.
Under the radical argument drop analysis the typology does not hold. Under such
an analysis the first and second person clitics, fusional for case, as they do not include
any agglunative case marking yet phonologically indicate case, would predict that Cr
should not be a radical argument drop language (languages with fusional case do not have
radical argument drop). However, as the data in Chapter 2 illustrates Cr drops nearly
every single argument, Cr would thus not fit within Neeleman and Szendröi's (2005,
2006) typology of argument drop languages. That is, languages that can drop the vast
majority, if not all, arguments must have agglutinating properties that Cr does not have.
Under the bound pronoun analysis Cr would fit naturally into the typology. The
bound pronouns, being fusional for case, would lead one to assume that argument drop
would be restricted (languages with fusional case have restricted argument drop). As the
third person is the only null argument under this analysis, argument drop would indeed be
restricted (similar to object drop in Italian see Rizzi (1986) for discussion). Thus, under
the bound pronoun analysis Cr fits within the typological coverage of Neeleman and
Szendröi (2005, 2006). In short, the bound pronoun analysis is once again preferable to
the radical argument drop analysis, this time in terms of typological characteristics.
76
4. Interim summary
It has been argued that a bound pronoun analysis of arguments in Cr is preferable to a
radical argument drop analysis. It has been claimed that any account involving pro would
be problematic as there appears to be no clear consensus cross-linguistically what pro is
exactly. Further, it has been argued that licensing conditions and recoverability of the null
arguments is also complicated by a lack of cross-linguistic consensus regarding the exact
nature of these two phenomena. In terms of learnability it has also been claimed that the
hybrid system of the radical null argument analysis poses a number of challenges to the
learner that the unified system of the bound pronoun analysis avoids.
In addition, in terms of typological characterizations it was shown that the bound
pronoun analysis fits with the typological generalizations arrived at by Neeleman and
Szendröi (2005, 2006). Under the radical argument drop analysis, Cr would not fit within
these typological generalizations. It was further noted that the radical argument drop
analysis requires more syntactic elements in numeration and extra derivations not
necessary under the bound pronoun analysis, in line with Chomsky’s (1995) economy
conditions. In short, it has been argued that the bound pronoun analysis is the best
analysis given the two. Next, we turn to a claim that bound pronouns in Cr are
-pronouns in terms of Déchaine and Wiltschko (2002).
77
5. Bound pronouns as -pronouns39
In this section the notion of -pronouns is introduced. Section 5.1 presents the basic
tenets of Déchaine and Wiltschko's (2002) proposal. Section 5.2 presents Déchaine and
Wiltschko's Halkomelem (Salishan/Central Coast) data and analysis of DP-pronouns.
Section 5.3 presents their Shuswap (Salishan/Northern Interior) data and analysis with a
re-analysis of the data that strengthens their claims regarding -pronouns. Section 5.4
presents data from Cr and further analysis arguing that the bound pronouns in Cr are
-pronouns, and that the null third person is a null -pronoun.40 While the data in the
following discussion focuses on Salishan languages, Déchaine and Wiltschko's original
claims include analysis of English and French, supporting their claim that the notion of
"pronoun is not a primitive" cross-linguistically.
5.1. The basic tenets of Déchaine and Wiltschko (2002)
Déchaine and Wiltschko (2002:409) argue that the notion "pronoun" is not a primitive
and that there are at least three types of pronouns: pronoun-DP, pronoun- P and
pronoun-NP. 41 For Déchaine and Wiltschko each pronoun type has a distinct categorical
status which can be illustrated in (58) below.
39
Throughout this section I present the data as it appears in its original source in cases where DéChaine and
Wiltschko (2002) present a slight variation of the original form. In general this is simply the addition of a
null morpheme indicated in the original not present in Déchaine and Wiltschko's (2002) representation of
the data. I do however indicate where the data appears in Déchaine and Wiltschko (2002) along with where
it appears in the original source.
40
Cf. Holmberg 2005 for discussion of Déchaine and Wiltschko's (2002) -pronouns as null arguments,
specifically pro.
41
This is similar in many ways to Zushi’s (2003) claim that there are three types of pro: DP-pro, Num-pro,
and NP-pro.
78
( 58 ) a.
DP
D
b.
P
P
c.
NP
NP
NP
N
N
(Déchaine and Wiltschko 2002:410)
Déchaine and Wiltschko provide the following summary of the categorical status of these
pronominals (410).
Table 10: Nominal proform typology
Pro-DP
Pro- P
Pro-NP
Internal Syntax
D syntax; morphologically
neither D syntax nor
N syntax
Distribution
Semantics
complex
argument
definite
N syntax
argument or predicate predicate
--constant
R-expression
variable
Binding-theoretic
status
---
Déchaine and Wiltschko (2002:419) argue that not all argument expressions are
DPs, contra Longobardi (1994). However, they do agree with Longobardi’s claim that
DPs must be arguments (59a) and that NPs must be predicates (59b). Their claim departs
from Longobardi in that their P can be an argument or a predicate. The claim is then
that not all arguments are DPs (59c) and not all nominal predicates are NPs (59d).
79
( 59 ) a. DP: Argument
b. NP: Predicate
c. Argument: DP, P
d. Nominal predicate: NP, P
What concerns us here is Déchaine and Wiltschko's analysis of DP-pronouns and
P-pronouns, I refer the reader to Déchaine and Wiltschko for greater explication of their
NP-pronoun proposal.
5.2. DP-Pronouns: The Halkomelem (Salishan/Central Coast) data
Déchaine and Wiltschko (2002) argue that DP-pronouns have the syntax of determiner
phrases, are restricted to argument positions, and function as R-expressions (410). To
illustrate the properties of DP-pronouns Déchaine and Wiltschko provide the following
analysis of Halkomelem (411).
Halkomelem is a head-marking language. In addition to pronominal clitics and
affixes, Halkomelem has a set of independent emphatic pronouns. These pronouns are
listed in Table 11.
Table 11: Halkomelem independent pronouns (Déchaine and Wiltschko 2002:412)
Singular
Plural
1
te - élthe (DET-1SG)
te-lhl1melh (DET-1PL)
2
3
te-á lthe (DET-1SG.EMPH)
te-léwe (DET-2SG)
tú-tl ò (DET-3SG)
te-lhwélep (DET-2PL)
tu-tl ó:lem (DET-3PL)
thu-tl ó:lem (DET.FEM-3PL)
yu- tl ó:lem (DET.PL-3PL)
Déchaine and Wiltschko (2002) argue that the Halkomelem data shows all the properties
of full DPs. According to Déchaine and Wiltschko, independent pronouns in
80
Halkomelem are morphosyntactically complex and comprise a determiner tú and a
-pronoun, in the example (60) the -pronoun is tl'ò (3SG) (412).
( 60 ) Halkomelem Pro-DP Structure (Déchaine and Wiltschko 2002:412)
DP
D
P
tú
NP
tl'ò
Ø
Galloway (1993) identifies the non-determiner element tl'ò and the other
-elements in Table 11 above as independent pronouns (verbal), that are "used by
themselves to answer such questions as: 'Who's there?', 'Who made this?' or 'Who wants
to go?'" (171-72). Further, these -pronouns are used in sentences where the -pronoun is
in its, to use Déchaine and Wiltschko’s (2002:413) term, predicative form. The following
illustrates.42
( 61 )
[tl'ò]PRED-cha te
Bill kw'e may-th-óme
3sg
-FUT DET Bill comp help-TRANS-2SG.OBJ
'It will be Bill that helps you."
(Galloway 1993:172 in Déchaine and Wiltschko 2002:413)
Déchaine and Wiltschko note that the DP-pronouns cannot appear in what they describe
as predicative position (62).
42
Here, and elsewhere where data is Halkomelem or Shuswap, I use DéChaine and Wiltschko's (2002)
symbols and abbreviations: 1 = 1st person, 2 = 2nd person, 3 = 3rd person, AGR = agreement, ARG =
argument, CONJ = conjunction, COMP = complementizer, DEIC = deictic, DET = determiner, EXCL =
exclusive, EMPH = emphatic, FUT = future, OBJ = object, OBL = oblique, REDUP = reduplicative, SG =
singular, SUBJ = subject, TRANS = transitivizer. The abbreviation SO was not identified in DéChaine and
Wiltschko (2002) nor was I able to locate it in Galloway (1993), where the same abbreviation is employed.
81
( 62 )
*[tútl'ò]PRED-cha te
Bill kw'e may-th-óme
(Déchaine and Wiltschko 2002:413)
Further, they demonstrate that the DP-pronoun can fill what they term an argument
position (63a) while the -pronoun cannot (63b).
( 63 ) a.
[Lám-Ø]PRED [tú-tl'ò]ARG
go-3OBJ
DET-3SG
'He goes.'
(Galloway 1993:173 in Déchaine and Wiltschko 2002:413)
b.
*[Lám-Ø]PRED [tl'ò]ARG
go-3OBJ
'He goes.'
3SG
(Déchaine and Wiltschko 2002:413- modified) 43
To account for the differences between the realization of DP-pronouns and
-pronouns Déchaine and Wiltschko (2002:413) argue that Halkomelem -pronouns are
restricted to predicate position, unlike the -pronouns discussed below, as the result of a
general markedness principle that governs blocking (Wunderlich 1996; Williams 1997).
For "concreteness,” they adopt Kosters' (1997) Principle of Maximal Specialization:
( 64 ) Principle of Maximal Specialization
In a grammatical dependency relation R, select the most
specialized form. A form A is more specialized than B
if A can fulfill fewer functions than B.
(Koster 1997:224)
Déchaine and Wiltschko (2002) argue that DP-pronouns specialize for argument
functions, and thus block -pronouns from this position when a language has both types
43
In Déchaine and Wiltschko (2002) the null absolutive morpheme is not indicated though it is assumed to
be present (Wiltschko p.c.; Galloway 1993).
82
of pronoun like Halkomelem, as (62) above demonstrates (413).44 Déchaine and
Wiltschko provide the following example (65) to demonstrate that DP-pronouns can
function as 'articles' and that -pronouns are not in complementary distribution with
nouns and here cannot be equated with nouns.
( 65 )
Tl'ó-cha-1-su
qwemc1we-t-Ø
[tú-tl'ò
q'ami]ARG
then-FUT-1SG-SO
hug-TRANS-3OBJ DET-FEM-3sg girl
'Then I'm going to hug that girl.'
(Galloway 1993:174 in Déchaine and Wiltschko 2002:412)
Having introduced the DP-pronoun and the Halkomelem data, we turn now to
-pronouns and the Shuswap data of Déchaine and Wiltschko.
5.3.
-Pronouns: The Shuswap (Salishan/N. Interior) data
Like Halkomelem, Shuswap (Secwepepemctsín) is a head marking language. Arguments
are marked on the verb as clitics or agreement affixes and full DP arguments are optional
(Déchaine and Wiltschko 2002:414). Shuswap has a set of independent emphatic
pronouns similar to those presented above for Halkomelem with some variation, notably
the lack of a determiner element.
44
While a full analysis of Halkomelem is beyond the scope of this paper, I would like to suggest the
possibility that the -pronouns are blocked from argument position because either (a) the agreement
morpheme, a null 3ABS in the examples above, is filling the position or (b) pro is functioning as the
argument. This would suggest that the DP-pronoun is serving as an emphatic element possibly in an
internal or external Focus position. Such an analysis would pattern with the other Salishan data provided
below.
83
Table 12: Shuswap independent pronouns
(adapted from Kuipers 1974, Lai 1998 in Déchaine and Wiltschko 2002:414)
Singular
Plural
1
n-tset-we7 (1SG-EMPH-DIEC)
2
3
7-enwi7 (2SG-EMPH)
newi7-s (EMPH-3)
wll-enwi7-kt (PL-EMPH-1PL)
wll-enwi7-s-kucw (PL-EMPH-3-2EXCL)
wll-enwi-mp (PL-EMPH-2PL)
wll-enwi7-s (PL-EMPH-3)
Déchaine and Wiltschko identify these pronouns as -pronouns and provide the
following structure to illustrate their composition.
( 66 )
P
NP
ntsetwe7
Ø
(Déchaine and Wiltschko 2002:415)
I propose a re-analysis of (66) based on the fact that Kuipers (1974:43) identifies
the person marking morphemes in Table 4 as comprising the possessive morpheme
paradigm is presented in (67b). In a re-analysis of (66), the person marking morphemes
are the -head, as would be expected by Déchaine & Wiltschko's (2002:410) claim that
the -head has -features, person, number, possibly gender and no other content, and
their analysis of the Halkomelem data above. Further, it is suggested that the emphatic
element is a predicate perhaps in NP.45 In the case of the first singular a deictic affix
attaches to the NP. This can be illustrated in (67) where (67a) illustrates the reanalysis of
(66) and (67b) illustrates the reanalysis in general terms for the entire paradigm in Table
12.
45
Here I suggest the predicate is nominal. I do not dismiss the possibility that the predicate may be verbal,
in which case the structure would comprise a P with a verbal XP sister.
84
( 67 ) a.
P
n
b.
tset-we7
P
NP
[pers] tset
[num]
Such a re-analyses would not change Déchaine and Wiltschko's claims. On the contrary,
it would seem that it strengthens their claims and demonstrates that Shuswap emphatic
pronouns, like Halkomelem emphatic pronouns, are morphosyntactically complex.
Déchaine and Wiltschko (2002) demonstrate that P-pronouns do not have NP
syntax (425). This can be seen if we compare (68a) and (68b), where (68a) illustrates a
grammatical category N construction.
( 68 ) a.
Yirí7 te [sqélemcw]N
l
wí.w.k-t-sem-s
man
COMP see(REDUP)-TRANS-1SG.OBJ-3SG.SUBJ
'That's the man that saw me.'
(Lai 1998:41.39a in Déchaine and Wiltschko 2002:415)
DEIC OBL
b.
*Yirí7 te
DEIC
[newi7-s]N l
OBL EMPH-3
COMP
wí.w.k-t-sem-s
see(REDUP)-TRANS-1SG.OBJ-3SG.SUBJ
'That's the man that saw me.'
(Lai 1998:41.39b in Déchaine and Wiltschko 2002:415)
Déchaine and Wiltschko (2002) argue that P-pronouns do not have DP syntax in
Shuswap. Their claim is that if P-pronouns in Shuswap had DP-pronoun status it would
be expected that they could not be modified by a determiner as in (69a). Specifically, they
85
argue that the position filled by the P-pronoun in (69a) is comparable to that of (69b)
where Déchaine and Wiltschko argue that John is an NP (415).
( 69 ) a.
[Wí.w.k-t-Ø-en]PRED
[re n-tséts-we7]ARG
see(REDUP)-TRANS-3SG.OBJ-1SG.SUBJ
DET 1SG-EMPH-DIEC
'I saw him.'
(Lai 1998:28.10 in Déchaine and Wiltschko 2002:415)
b.
[re John]ARG46
[Wík-t-Ø-s]PRED
see-TRANS-3SG.OBJ-1SG.SUBJ
DET John
'I saw John.'
(Lai 1998:28.11 in Déchaine and Wiltschko 2002:415)
Déchaine and Wiltschko conclude that Shuswap -pronouns are neither NPs nor DPs.
However, Déchaine and Wiltschko demonstrate that they can function as what they
describe as arguments (70) and predicates (71).
( 70 ) a.
[Wí.w.k-t-Ø-en]PRED
[ newí7-s]ARG
see(REDUP)-TRANS-3SG.OBJ-1SG.SUBJ
EMPH-3
'I saw HIM.'
(Lai 1998:28.13 in Déchaine and Wiltschko 2002:416)
b.
[ Newí7-s]ARG
EMPH-3
[Wík-t-Ø-s]PRED
see-TRANS-3SG.OBJ-3SG.SUBJ
[Mary]ARG
Mary
'HE saw Mary.'
(Lai 1998:28.11 in Déchaine and Wiltschko 2002:416)
46
It is not clear if R-expressions such as John are considered NPs or DPs in Shuswap. It is the case that
some languages, such as Italian and German, may allow a Det0 to be sister to a DP (Longobardi 1994). This
appears to be the case in Cr where an R-expression such as John may appear with or without a Det0 in what
have been analyzed as DPs in topic positions (Doak 1997:257). The lack of clarity here will not affect the
following analysis of Cr.
86
( 71 )
[ Newí7-s]PRED
[wík-t-Ø-m-es]ARG
EMPH-3
see-TRANS-3SG.OBJ-PAST-3SG.CONJ
'It is HIM that saw him/her.'
(Lai 1998:28.13a in Déchaine and Wiltschko 2002:416)
As illustrated, the reanalysis of Shuswap -pronouns does not affect Déchaine
and Wiltschko's (2002) analysis; rather, it strengthens their claim that -heads contain
-features only. With Déchaine and Wiltschko's account of pronominals presented, we
next turn to Cr. First however, a brief account of English pronominals will be discussed
to further illustrate the distinction between DP-pronouns and P-pronouns.
In their discussion of English pronominals, Déchaine and Wiltschko first present
evidence demonstrating that one is an NP-pronoun. This is done by showing that one has
the syntax of nouns as it can occur following a determiner, a quantifier, or a modifier, as
illustrated in (72).
( 72 ) a. the one
b. someone
c. the real one
(Déchaine and Wiltschko 2002:420)
They go on to discuss one cross-linguistically to further support this claim. Next
they turn to an account of personal pronouns in English. Déchaine and Wiltschko (2002)
argue that the fact that English personal pronouns are not syntactically homogenous
(Ritter 1995) can be accounted for by an account of English personal pronouns which
includes a distinction between DP-pronominals and P-pronominals. Déchaine and
Wiltschko note that in Standard American English plural, 1st and 2nd person pronouns
can function as determiners and 3rd person pronouns cannot (421). They argue that this
87
can be accounted for if 1st and 2nd person plural pronouns are analyzed as DPs47 and 3rd
person plural are analyzed as Ps. This can be illustrated in (73) and (74).
( 73 ) a.
b.
c.
we linguists
us linguists
you linguists
*they linguists
you linguists
*them linguists
(Déchaine and Wiltschko (2002:421)
( 74 ) a.
DP Structure: 1st and 2nd Person
DP
D
we
P
NP
N
linguists
47
In their discussion, Déchaine and Wiltschko (2002) refer to these pronouns as DPs, however, in their
analysis these are D°, this does not have implications for the analysis here.
88
b.
P Structure: 3rd Person48
P
NP
they
N
)
(Déchaine and Wiltschko (2002:421)
Déchaine and Wiltschko go on to demonstrate how this analysis holds for the
pronominals discussed despite dialect variation. Now we turn to Cr.
5.4. Coeur d'Alene (Salishan/S. Interior) emphatic pronominals
Coeur d'Alene is similar to Shuswap in that arguments marked on the verb have been
described as clitics or agreement affixes (bound pronouns). As described below, full DP
arguments are optional and the independent emphatic pronouns pattern with -pronouns.
Doak (1997) identifies the following pronouns in Table 13 as being emphatic or stand
alone predicates. Further, she argues that they are constructed as "intransitive predicates
with nominative pronouns and unanalyzable roots that are nowhere else" (72). The
"nominative pronouns" in Table 13 are the intransitive bound pronouns in Cr.
48
Déchaine and Wiltschko (2002:422) give no clear indication as to why “they” does not allow for a
complement NP. They argue that in at least some dialects pronouns can be broken down into a D element
and a -element:
a. [D th-e]
b. [D th-[ is]] [D th-[ ese]] [D th-[ at]] [D th-[ ose]]
c. [D th-[ em]]
This, they suggest, is the reason for the acceptability in some dialects of "them linguists.” However, they
provide no explanation as to why "they linguists" is ungrammatical. I assume that English must have some
parametric variation that disallows this construction, a lone -element with a NP complement. Another
possibility would be that “they,” like many third person pronouns cross linguistically (Bhat 2004), is not a
“true pronoun” in the sense that first and second person are, and that it is a deictic element which may have
restrictions on its complement. It will be left to future research to determine the exact nature of this
phenomenon.
89
Table 13: Coeur d'Alene Emphatic Pronouns
Singular
Plural
1
2
3
* (1sg-emph)
(1pl-emph)
(2PL-EMPH)
* (2sg-emph)
(3-emph)
Ø-
Ø- -
(3-emph-pl)
It is important to note that the -elements in the Cr emphatic pronouns (the bound
pronouns) pattern not with the possessive bound pronouns, as in Shuswap, but rather
with the intransitive bound pronouns as mentioned above. Cr has a separate set of
possessive bound pronouns (the genitive bound pronoun set). While there are few
examples in the data, these emphatic pronouns follow the Shuswap pronouns in matching
Déchaine and Wiltschko’s (2002) requirements of -pronouns, as illustrated below where
the relevant forms are in bold italics and underlined and the -pronouns are all marked
1p-emph or 1s-emph:49
i.
They can follow determiners:
( 75 )
- (
then
.
det1
dir
.
at.a.distance det3
,
,
$
there.is fire
there det1 1p-emph
'At a distance there was a fire, which was ours.'
(Reichard 1927-29:cssw204)
49
In (76) and (77) phonological reduction of the emphatic element occurs.
90
ii.
They can be bound outside their local domain (here by Coyote):
( 76 )
- !
-
!
then say
det3
- 8
Coyotei
8
$
4
*
then 1s-emphi
4
also
-4
*
1s-emph
.4
will.do
-3g
'Coyotei said, "Ii too, I, this will be my doing."'
(Reichard 1927-29:cim53a)
iii. They can be what Déchaine and Wiltschko (2002) refer to as arguments:
( 77 )
- 8
A
* BC D E
A
1s-emph
holding.back -ct
'I will be the one to hold him back.'
)
BF D GH
-3.obj -nte.subj
(Reichard 1927-29:cossd149)
Thus, the structure of Coeur d'Alene predicative / emphatic pronouns can be
illustrated as in (78).50
( 78 )
P
NP
[pers]
[num]
*
At this point the evidence demonstrates that the -elements in Table 13 in the
predicative / emphatic pronominal paradigm, are in fact -heads in the sense of Déchaine
and Wiltschko's (2002) analysis. As these -heads are the same morphemes as the
50
Here Déchaine and Wiltschko's (2002) formalism is followed; however an alternate structure may be
possible, where the emphatic element
* is analyzed as being verbal. Such a distinction is not
necessary, though future inquiry should resolve this issue shedding further light on the nature of Cr
emphatic pronominals.
91
intransitive bound pronouns, it is also concluded that the intransitive bound pronouns are
-heads. It will further be assumed that this analysis can be extended to the transitive
subject and object bound pronouns as well.
It is not clear from Déchaine and Wiltschko's (2002) analysis if the -heads they
discuss are in fact bound pronominals or perhaps -agreement morphemes. In their
analysis they refer to the whole complex pronoun as the -pronoun. However, Déchaine
and Wiltschko (2002:432) present the possibility of an analysis of these elements as
agreement with their discussion of Mojave and Plains Cree where they label elements as
0
-agreement, however they do not use this notation with the data discussed. This should
not affect the claim that the Cr bound pronouns are -heads. In the case of the null 3rd
person, it will be assumed that it is a null -pronoun (cf. Holmberg 2005 for an analysis
of pro as a null -pronoun). In this way the bound pronouns are accounted for in terms of
syntactic content, including the null 3rd person.
Next we turn to an account of bound pronouns having their diachronic origins in
agreement elements. However, this analysis will be rejected on the grounds that it returns
us to the set of problems discussed in Section 3 above regarding an argument system
comprised of first and second person intransitive clitics, null subjects and objects in all
other constructions, and various null agreement markers. An alternate analysis in which
bound pronouns have been the historical norm will be presented. It will be demonstrated
that this analysis provides a way to account for the cross-family differences regarding
argument structure noted especially in Davis and Matthewson (2003).
92
6. Bound pronouns: Possible diachronic antecedents
In this section, it will be argued that bound pronouns may have their diachronic
antecedents in agreement morphology. The starting point will be the notion that there are
two types of agreement morphology: strong and weak (Chomsky 1989; Jaeggli and Safir
1989; Alexiadou and Anagnostopoulou 1998; Zushi 2003; Speas 2004; among others).
It has been claimed that this distinction is crucial in any analysis of
argument-drop in languages that include overt agreement morphology (cf. preceding
references). In essence, the claim is that strong agreement licenses argument-drop in
many languages, while weak agreement does not. Importantly, Alexiadou and
Anagnostopoulou (1998:521) and Speas (2004:5) adopt Rohrbacher's (1993) Full
Paradigm criterion to distinguish strong agreement from weak.
( 79 ) Rohrbacher's Full Paradigm
INFL is a referential category with lexically listed affixes in
exactly those languages where regular Subject-verb agreement
minimally distinctively marks all referential INFL-features such
that:
a. In at least one number and one tense, the person features [1st] and [2nd] are
distinctively marked
b. In at least one person one tense, the number feature [singular] is
distinctively marked. (1994:118)
A paradigm meeting the description in (79) is what Rohrbacher calls a “Full
Paradigm.” Agreement morphemes comprising such a paradigm each have individual
lexical entries according to Rohrbacher (1994).51 Speas (2004) states this in terms of
Rohrbacher's Generalization.
51
Here the discussion is couched within a lexicalist framework. The distinction would have to be syntactic
within the DM framework. One possibility is that what is identified as “strong” agreement here has
syntactic features which require the morpheme to be present at narrow syntax if present in numeration,
while “weak” agreement would be akin to Embick and Noyer’s (2005) “ornamental” morphology, that is a
93
( 80 ) Rohrbacher's Generalization
"Strong" morphemes have individual lexical entries
"Weak" morphemes do not have individual lexical entries. (6)
For Speas (2004), strong agreement is generated as a functional head, an AGR head which
projects an AGR-phrase, while weak AGR is generated within the VP-shell (6). This can
be seen in (81).
( 81 ) a.
strong AGR
AGRP
DP
b. weak AGR
AGRP
AGR'
AGR0
-af
AGR'
VP
...V...
VP
...V+af ...
(Speas 2004:6)
Later, Speas (2004:39) amends Rohrbacher's Generalization by stating that "the
agreement affixes may be listed as independent lexical entries, but they are not
necessarily so listed" (emphasis Speas). This allows Speas to postulate two types of
strong agreement: lexicalized strong agreement (Type A) and non-lexicalized strong
agreement (Type B). Speas can then account for the differences between languages such
as Spanish which have strong agreement (Type A) and allow null arguments, and
languages like Yiddish which also have strong agreement (Type B) but do not allow null
arguments. In Type A languages a specifier is not required in AGRP at spellout because
the lexicalized -features satisfy all derivations up to that point. However, Speas does not
eliminate pro in such constructions, because it is still needed to absorb -roles and check
case (at LF in the case of Italian for example). In cases where Type B strong agreement
disassociated morpheme (Halle and Marantz (1993)) inserted after narrow syntax in the morphological
component of the grammar.
94
occurs a DP must raise to the specifier position of AGRP. Weak agreement does not
project as a lexical head and is thus confined to generation in the VP-shell.
Assuming that bound pronouns are diachronically related to agreement, and that
DPs and NPs were adjuncts at the time as they are today, the historical "agreement
morphology" could be considered lexicalized if we accept Speas’ (2004) claims. That is,
historically agreement in Cr would have satisfied Rohrbacher's Full Paradigm. In the
next section it will be shown how assuming Speas’ analysis would lead to the
development of bound pronouns in Cr.
6.1. Grammaticalization
In this section, it will be claimed that in the case of Cr, the lexicalization of the agreement
morphemes would have led to the bound pronouns, and that such a process would be
interpreted as a type of grammaticalization. ‘Grammaticalization’ here is not meant as a
predictive theory of grammaticalization in the sense of Heine (1990, 1993; Heine, Claudi,
and Hünnemeyer 1991) here; rather, it is meant in terms of Newmeyer (1998) who
assumes that grammaticalization is epiphenomenal, or "nothing more than a label for the
conjunction of certain types of independently occurring linguistic changes" (237), here
reanalysis and lexicalization. Further, it is assumed that grammaticalization does not
unfold in component parts along a cline culminating in some form of "complete
grammaticalization" as Heine might suggest. That is, it is not assumed that once steps
toward grammaticalization begin, other steps will inevitably follow suit (251). While it is
the case that grammaticalization is generally assumed to be unidirectional (Heine and
Kuteva 2002; Bybee, Perkins, and Pagliuca 1994), lexical units can be grammaticized but
grammatical units cannot be lexicalized, there are a number of exceptions to the
unidirectionality principle (Heine and Kuteva 2002:5; Newmeyer 1998:263; cf. especially
95
Lightfoot 2005 for discussion of a distinction between grammaticalization and
lexicalization and the unidirectional principle). It is the exceptions discussed in
Newmeyer and elsewhere that lead to a preference for the analysis put forth by
Newmeyer.
Newmeyer (1998:263-78) provides several counter-examples to the
unidirectionality principle. Of interest to us here are the cases in which inflectional
affixes (268-69) or clitics become pronouns (270-72). Citing Janda (1980, 1981),
Newmeyer provides an example of an affix upgrading to pronoun status from Old
English which had a rich case system (268). Old English included a genitive suffix -(e)s
which was "upgraded" to a full free morpheme, the pronoun his, thus indicating that the
reanalysis from clitic to free independent pronominal word had been completed. Citing
Bybee, Perkins and Pagliuca (1994), Newmeyer presents another example of agreement
morphology upgrading to full lexical status this time, in Irish. Newmeyer shows that in
Irish, the first person plural suffix -mid/-muid has become an independent pronoun (268).
In regards to clitic upgrading to full pronouns Newmeyer (1998) states, "the
upgrading of clitic pronouns to full lexical pronouns seems to have happened repeatedly"
(270). Another example of agreement morphology upgrading to the status of pronoun is
illustrated again by historical developments of English. Newmeyer, citing Kroch, Myhill,
and Pintzuk (1982), provides examples of how the enclitics of English underwent
decliticization after 1550 to become the free pronouns “thou”, “he”, and “ye” (270).
Newmeyer notes that in the sixteenth century inverted subject pronouns such as those in
(82) were enclitics attached to the verb (270):
96
( 82 ) a.
Where dwellyth she?
b.
Why bewaylest thou thus soor, O Pelargus?
(Newmeyer 1998:270)
The spelling of the earlier period revealed their clitic status, in which the pronoun was
represented orthographically as one unit with the verb:
( 83 ) a.
b.
c.
hastow
wiltow
for 'hast thou'
for 'wilt thou'
wille
for 'will he'
(Newmeyer 1998:270)
Later, not appears for the first time between the inverted main verb and the pronoun
rather than occurring after the subject (Newmeyer 1998:271).
Newmeyer (1998) goes on to present examples of upgrading to pronominals from
Ilokano, Estonian, and a number other languages. In short, the analysis of Cr agreement
morphemes upgrading to pronouns patterns with linguistic variation cross-linguistically.
It is a case of reanalysis without semantic change or phonetic reduction (252).
Such a reanalysis of agreement morphemes to bound pronouns would have
produced a simpler Cr grammar (see contrast between number of primitives needed in the
agreement analysis and bound pronoun analysis above). This analysis would also account
for the apparent differences within the Salishan family regarding differences in argument
structure between languages like Cr and Lummi and St’át’imcets (Davis and Matthewson
2003), full reanalysis, in terms of bound pronouns, having occurred in Cr and Lummi but
not St’át’imcets. Other differences would be that Lummi is a Pronominal Argument
language as outlined by Jelinek (2004, 2006), while Cr would have the core feature of a
Pronominal Argument language, bound pronouns, but would differ in other respects,
97
notably incorporation.52 St’át’imcets on the other hand, as analyzed by Davis and
Matthewson, would not have the core characteristic, bound pronouns, of the Pronominal
Argument Parameter.
The problem with this analysis is that it brings back the thorny issues of Section 3
above that the bound pronoun analysis avoids. A simpler solution, and one that can also
account for the differences between Cr, Lummi, and St’át’imcets, would be that the
bound pronouns do not have their diachronic origin in distinct agreement morphology. A
number of languages have been identified as having bound pronouns as arguments
generated in argument position (see references above in Section 1 and references therein).
Therefore, it is not unreasonable to assume that UG provides bound pronouns as an
option for any given language and its argument deriving mechanisms. In the case of
Salishan languages, the situation has not been fully investigated to the author’s
knowledge beyond Lummi, Cr, and St’át’imcets. However, the evidence suggests rather
clearly that bound pronouns are arguments in these three Salishan languages.
The case for Lummi has been made by Jelinek as has the case for Cr above.
The case for St’át’imcets rests on the assumption by Davis and Matthewson that “the
existence of morphologically unregistered objects pose intractable problems” for the
Pronominal Argument Hypothesis (94). Davis and Matthewson identify what they refer
to as “quasi-objects,” which do not appear to be traditional adjuncts, but rather DP
arguments. Crucially, these “quasi-objects” appear in a very limited number of very
specific environments. Elsewhere, the pattern of bound pronouns appears as described for
Cr above in general. While a complete analysis of St’át’imcets is beyond the scope of the
current discussion, it can be claimed, based on the discussion provided by Davis and
52
See Haugen (2004) for illuminating discussion on the setting of macro-parameters such as the
Pronomoninal Argument Parameter, in various stages across a language family.
98
Matthewson, that the difference lies at the micro-parametric level on the various nodes
that license these DP arguments. Davis and Matthewson’s claims are that limited
constructions in St’át’imcets that allow for DP arguments, contra the tenets of the
Pronominal Argument claims, renders St’át’imcets a non-Pronominal Argument
language. This view is based on the notion that a parameter must apply to every relevant
node. However, if one views parametric variation as applying at the individual node
level, rather than as applying to every node in exactly the same way, St’át’imcets can be
seen as satisfying the tenets of a Pronominal Argument language in the majority of
syntactic constructions in terms of bound pronouns.
The claim here is that the nodes that license quasi-objects in very restricted
environments in St’át’imcets have a different parametric setting than the remaining nodes
which license bound pronouns and ban DP arguments elsewhere. In this way St’át’imcets
is similar to Hebrew as described by Vainikka and Levy (1999). In Hebrew DPs are
banned in first and second person subject positions, but allowed in third person
constructions. Under the analysis presented here, these facts represent parametric
difference between the nodes that license third person DP arguments and the nodes that
disallow DP arguments in first and second person constructions in Hebrew. The same can
be said of the facts regarding null objects in Italian as described by Rizzi (1986). In
Italian null object pronouns must occur in very specific constructions. In such
constructions overt pronouns are not acceptable. Again, this suggests a micro-parameter
at the node level. Recognizing that parametric variation is at the node level can account
for the differences seen between Cr and Lummi, and St’át’imcets. Also, it allows the
avoidance of the issues discussed in 3 above. In this way the claims made in 3 above are
maintained and the differences between the three Salishan languages are accounted for.
99
Future inquiries into the nature of this phenomena may support Pensalfini's (2004)
claim that languages of the type identified as non-configurational53 (such as Cr) may have
restrictions on the type of referential features that are allowed in argument position. In the
case of Cr, it is the case that there is a ban on DPs in argument positions. Thus, Cr clearly
fits with one element of Jelinek’s (2004, 2006) macro-parameter: DPs are not allowed in
argument positions, rather bound pronouns, -pronouns here, fill the argument
positions.54
7. Conclusions
In this chapter it has been argued that an analysis of bound pronouns is preferable to an
analysis of radical argument drop in Cr, based on considerations regarding the
identification of the syntactic features of null arguments, recoverability and licensing of
null arguments, learnability, and typological facts. Further, it has been shown that the
bound pronoun analysis arrives at the proper surface morpheme order with fewer
syntactic elements and derivations. In addition, it has been argued that bound pronouns
are -pronouns in terms of Déchaine and Wiltschko (2002). As for null arguments, it was
claimed that the only null argument was the third person pronoun and that this pronoun
was a null -pronoun that was accounted for under the Cue based model of parameter
setting (Lightfoot 1997a, 1997b; Dresher 1999, 2003a, 2003b).
Further, it has been demonstrated that employing Hale and Keyser's (2002)
Conflation as head-movement (Harley 2004) allows Chomsky's notion of
53
Here “non-configurational” refers to the fact that full DPs are not base generated in argument positions.
Under Vainikka and Levy’s 1999 analysis of Hebrew and Standard Finnish, the same claim might be
made. Vainikka and Levy claim that agreement morphology in first and second person constructions
function as arguments. It may be the case that a micro-parameter setting bans DPs in just those
constructions, requiring bound pronouns, -pronouns.
54
100
head-movement as phonological in nature to be maintained. The analysis of basic clause
structure in Cr arrived at, (37) above, can be restated here as (84).
( 84 ) a.
Cr basic transitive construction
$
cust- accompany -dt
'We go with you.'
b.
-2acc -1p.erg
TP
T
ASPP
ASP
vP
P
P
v
$P
$
P
Thus the basic clause structure for Cr is arrived at. In the next chapter we turn to a
discussion of Lexical Affixes and incorporation in Cr.
101
CHAPTER 4
LEXICAL AFFIXES
1. Introduction
In this chapter an account of lexical affixes in Coeur d'Alene is presented that builds on
the earlier work of Salishan scholars (Sapir 1911; Saunders and Davis 1975; Hagège
1978, 1980; Carlson 1990; Kinkade 1998; Gerdts 1998, 2003; Wiltschko 2006 among
others). Lexical affixes comprise a closed class of bound morphemes which productively
add semantic content to the verb in the Salishan family of languages. An example from
Coeur d'Alene (henceforth Cr) follows, where the lexical suffix in (85) is preceded by the
equal symbol "=" and is in bold and underlined:
( 85 ) a.
*
b.
) *
I
3abs- finish=mouth
'He finished eating.'
(Doak 1997:289)
*
*
)
finish -ct
-3abs -3erg
'He finished it.'
(Nicodemus 1975)
In this chapter it will be argued that lexical affixes in Cr are n heads which
incorporate into roots, where incorporation is a result of Hale and Keyser's (2002)
Conflation. It will be further argued that the analysis presented is preferable to Baker,
Aranovich, and Golluscio’s (2004) account of incorporation on the grounds that it
requires fewer parameters and no special deletion mechanism. Further, a Conflation
102
analysis maintains Chomsky's (1994) claim that head-movement is phonological (though
not postsyntactic in nature), as Harley (2004) demonstrates.
The chapter is structured as follows. After an introduction to lexical affixes and a
review of historical claims regarding their diachronic origins in Section 2, Section 3
presents the claim that lexical affixes are n heads. This is followed by the formal analysis
of lexical affixes and stem+stem incorporation in Cr in Section 4. In Section 5 are
concluding remarks.
2. Lexical Affixes in Cr
In this section lexical affixes (henceforth LAs) will be briefly introduced in terms of how
they are identified and their general semantics. LAs will then be compared to stem
compounding and incorporation processes in Cr, illustrating that the two are distinct and
involve different morphemes. Next, the possibility that LAs derived historically from
compounding will be briefly discussed. Finally, the functions of Cr LAs will be
introduced. First, an introduction to the concept of LAs is presented.
Kinkade (1998) notes that lexical affixes have at various times been referred to as
"verbal affixes that refer to nouns" (Sapir 1911:251), "field suffixes" (Vogt 1940:58), and
"somatic suffixes" (Kuipers 1974:59). It was Kinkade (1963:352) who first referred to
such forms as "lexical suffixes," which is the term used by Salishan scholars today.
Kinkade (1998) defines lexical suffixes as:
[A] group of suffixes found in Salish, Chemakuan, and Wakashan
which have semantic content analogous to specific nouns, but lack
phonological similarity. In this sense, they are unlike what are usually
perceived as derivational affixes, which more often have little concrete
meaning, rather serving to identify lexical classes. (266)
103
Mithun (1997:357) further identifies languages in Eskimoan and Tsimishian as having
similar morphemes. Fleck 2006 identifies what he considers LAs in Matses, a Panoan
language spoken in the Amazonian Peru and Brazil. Holmes and Hinchliffe (2003:534)
present a recent productive phenomenon in Swedish where nominalized roots work in
much the same way as prefixes and suffixes. In short, the phenomenon does not appear to
be an areal phenomenon as Kinkade's description might suggest.
Salishan languages have roughly 100 lexical suffixes on average (Gerdts 1998:95,
2003:346). Reichard (1938) identifies some 80 suffixes in Cr which she refers to as
“nominal” (601-603). These nominal suffixes denote body parts (hand, foot, stomach,
nose), basic physical/environmental concepts (ground, plant, vegetation, water, fire), and
human/relational terms (offspring, people, person). A number of these suffixes,
especially the body-part suffixes, extend to take on locative meanings.55
Gerdts (2003) notes that lexical affixes bear little or no phonological similarity to
free standing nouns. Wiltschko (2006) also notes that "lexical suffixes are not
transparently related to free standing nouns ..."(3). In Cr, however, a small subset of
lexical affixes may share much of, and in one case all, the phonological features of their
stem counterparts.56 In Table 14 we see examples of lexical suffixes that do not match the
phonological content of their apparent counterparts, while Table 15 presents the limited
set of lexical affixes that share some of the phonological material of their counterpart.
55
It has also been noted in the references above that in many cases lexical affixes function in much the
same was as classifiers.
56
Wiltschko (2006), among others, makes a distinction between 'lexical affixes' on the one hand and
'nominal' counterparts on the other. Here, following Mithun (1997) I make a distinction between 'lexical
affixes' and stems, where I consider stem to mean a categorized root in the sense of Distributed
Morphology where an uncategorized root becomes categorized in a complement relationship with a
categorizing head x (n, v, a, etc.).
104
Table 14: Lexical affixes with no phonological similarity to free forms in Coeur d'Alene
lexical suffix (bound)
free
a. =. ,
b. I *
c. =*!
'day, sky, atmosphere'
'stomach, heart'
'hallow object, abdomen,
wagon, canoe'
.
d. = ,
e. =
f. =
'fish'
'tooth'
'horn, hairline'
,
,J
-
K
=
'day'
'heart'
'canoe'
'fish'
'tooth'
'antler, horn'
Table 15: Lexical affixes with similar phonological content to free forms in Coeur
d'Alene
lexical suffix (bound)
free
g. = K
h. =
i. =
'camas'
'people'
'horse'
j. =
k. =
5I
m. I
'tongue, tongue-shaped'
'food, pertaining to food'
'water, liquid'
'eye, face, orifice through
I
K
'camas'
'Indian'
'horse'
'tongue'
'eat'
'water'
'face, water'
which light shines, fire'
LAs differ in a number of ways to their stem counterparts, they: (1) often have a
number of different semantic extensions compared to their stem counterparts; (2) never
themselves serve as phonological words or as the bases for phonological words; and (3)
constitute a relatively closed class of items. LAs may appear with roots in nominals or in
predicates (the LA follows the “=” symbol):
( 86 ) LAs in nominals
a.
,8 I
nom- purple =pendant
'grape'
(Reichard 1938:616.503)
105
b.
7 8 I*!
nom- dip =hollow.object
'fishnet'
(Reichard 1938:614.492)
( 87 ) LAs in predicates
a.
*
)
*
I
3abs- finish=mouth
'He finished eating.'
b.
(Doak 1997:289)
*+ ,
)
*
I,
3abs- loc- finish =head
'He finished growing.'
(Doak 1997:289)
Many scholars (Carlson 1990; Kinkade 1998; Gerdts 1998; Mithun 1997; among
others) have noticed that lexical affixation resembles nominal compounding ((86) above)
and noun incorporation ((87) above) in Salishan languages. In the next section nominal
compounding and incorporation in Cr will be compared to lexical affixation to illustrate
that the two are distinct, but have similarities.
2.1. Stem+stem vs. LAs: The differences
Like all Salishan languages (Kinkade 1998), Cr has both productive lexical affixation and
nominal compounding and incorporation. Doak (1997) notes that the process of
combining LAs in Cr with stems resembles the process of compounding and
incorporation of stems in Cr (288). The primary difference between the two operations is
that an LA shows no signs of derivation (such as the presence of the nominalizing
morpheme), or connective elements (such as the
connective or its reduced form
or
. ). The following examples illustrate these differences. In (88a) and (88b) we see what
is referred to as “compounding” in the Salish literature (Kinkade 1998; Carlson 1990;
106
Doak 1997; Mithun 1997; among others)57. It is crucially characterized by the
nominalizing prefix
and the connective
. In (89a) and (89b) we see LAs adjoining
directly to roots with no intervening morphology necessary or possible, in contrast to the
stem+stem forms in (88).
( 88 ) word compounding in Cr
a.
*
. ,
)
*
,
3abs- finish-conn-nom-write -m
'He finished writing.'
b.
*
(
*
1nom- finish -conn-nom-make
'I finished working.'
(Doak 1997:289)
( 89 ) lexical affixation in Cr
a.
*
)
*
I
3abs finish =mouth
'He finished eating.'
b.
*+ ,
)
*
I,
3abs- loc- finish =head
'He finished growing.'
(Doak 1997:289)
Kinkade (1998:11) provides the following examples, taken from Reichard (1938),
which further illustrates the differences between the two types of constructions. In (90a)
we see the connective - , the nominalizing morpheme -, and the root
57
which Doak
As Mithun 1997 notes, in the Salishan literature "compound" refers to what is recognized in other
languages as both nominal compounds (noun + noun) and noun incorporation (verb+noun) (365).
107
(1998:299.560 ) glosses as 'Indian'. In (90b) we see the lexical affix I
'people'.
Below (@2.2) it will be proposed that the - in (90b) is a phonological remnant of the
nominalizing process that remained when the complex stem
was reanalyzed as a bound morpheme I
( 9
) 'Indian'
'people'. That s- in such examples is a
phonological remnant of the nominalizing process is a claim used in Kinkade (1998) and
Carlson (1990) to further establish a link between compounding and lexical affixation
diachronically.
( 90 ) a.
(one word)
*
summon-conn-nom-Indian -sg.imp
'Call the people!'
b
)
. I
3abs- shoot =people
'He shot somebody/something.' (lit. 'He people-shot.')
(Reichard 1938:617.505)
The conclusion drawn from such data is that LA attachment is distinct from
nominal stem incorporation, as Sapir (1911), Hagège (1978), Carlson (1990), Kinkade
(1998) (among others) have noted. The merging of LAs and stems does not include
categorization and compounding morphology and therefore is assumed to be a different
process than stem compounding, or perhaps more precisely different from incorporation.
The limited set of forms represented by (90b) that appear to include the nominalizing
morpheme -, will be accounted for in the following discussion.
108
2.2. LAs historically
It has been argued by Kinkade (1998), Carlson (1990), Mithun (1997), and Gerdts
(1998) (among others), that lexical affixes have their diachronic origin in nominal
compounding, a view commonly held among Salishan scholars. Mithun states:
The precursors of the Salishan lexical prefixes and suffixes first
bonded phonologically to their hosts in compounds, at a time when
they still retained their status as roots. At this point, they lost
specific referentiality and case roles. Abstraction and extension of
meaning occurred afterward over a considerable period of time. (369)
LAs as a group show effects of grammaticalization, with apparent meanings that are more
general and abstract than their stem counterparts, phonological reduction from identified
stem cognates, and reduction in form (affixes rather than stems). These are features
widely identified as characteristic of grammaticalization (cf. Bybee, Perkins & Pagliuca
1994:9-26; Heine & Kuteva 2002:4; Newmeyer 1998:248; among others). It thus has
been the claim that LA affixes derive from stem incorporation through a process of
grammaticalization for most who have worked on LAs, such as Carlson, Kinkade,
Mithun, and Gerdts (among others).
This would explain the similarities seen in Table 15 above repeated here as Table
16. More specifically, this provides a rather straightforward account for the of =
phonological remnant from nominal incorporation.
,a
109
Table 16: Lexical affixes with similar phonological content to free forms in Coeur
d'Alene
lexical suffix (bound)
free
g. = K
h. =
'camas'
'people'
i. =
j. =
k. =
5I
'horse'
'tongue, tongue-shaped'
'food, pertaining to food'
'water, liquid'
'horse'
'tongue'
'eat'
'water'
'eye, face, orifice through
which light shines, fire'
'face, water'
I
m. I
K
'camas'
'Indian'
As noted, the view presented is widely held among Salishanists and those who
have worked with this phenomenon. It represents what I consider the "traditional" view
of the diachronic origin of LAs. This can be illustrated in the following example. In (90a)
above, repeated here as (91), we saw a compound construction with the relevant
compounding morphology: categorizing (nominal)
( 91 )
and connective
.
*
summon-conn-nom-Indian -sg.imp
'Call the people!'
For one generation of speakers this is the construction employed. For the next generation
of speakers the combination:
( 92 )
-conn-nom-Indian
is reanalyzed as a single bound morpheme. In (93b) we see remnants of the nominalizing
morphology, namely . Mithun (1997), Carlson, (1990), and Kinkade (1998) (among
others), assume the connective remnant , as is reduced before in Cr, would have
been included in the reanalysis process and that (93a) would have been the original LA
110
from which today’s I
can be traced back to. This assumption is based on the fact
that throughout the family a number of LAs still contain remnants of the connecting
morpheme - , though they still exhibit elements of phonological reduction, usually loss
of the vowel. Therefore, it would be natural, based on the phonological rules just
mentioned, to assume that the original form (unattested) of (90b), repeated here as (93b),
would have been (93a) for the first generation of speakers to exploit this form as an LA,
and only reduced to (93b) in later generations.
( 93 ) a.
)
. I
3abs- shoot =people
'He shot somebody/something.' (lit. 'He people-shot.') unattested
b
)
. I
3abs- shoot =people
'He shot somebody/something.' (lit. 'He people-shot.')
If the language had other LAs at the time of =
particular generation of speakers employing I
'’s entry into the grammar, the
for the first time would be exploiting
a category already present in the Cr grammar. The assumption would be that LAs would
have been serving the same functions as they do currently, and that the speakers that
added I
to the category of LAs would have done so to add an element that has the
particular semantics encoded in I
. The new morpheme would then allow for
constructions of the type in (93), which differ from that in (91) in the type of information
conveyed.
If on the other hand I
represented, hypothetically, the first LA to enter the
language, then the generation to first employ it would be exploiting a category
underlyingly available in UG, but not yet exploited by the Cr grammar. The fact that
items that resemble LAs have been attested in a variety of the world's language families,
111
and in geographically diverse regions, suggests rather strongly that UG provides such a
category. The case of Swedish, in particular, offers great potential to research the just
described phenomenon as it appears that a subset of the current speaking population is
beginning to exploit this process in a way that might be nearly exactly as it may have
occurred in Cr. 58 This summarizes the traditional view of LAs being derived from a
grammaticalization process involving compounds. There is however another possibility,
in contrast to the traditional view, worth considering.
Heath (1998) proposes what he refers to as "hermit crab" grammaticalization.
Heath defines this process as follows:59
A grammatical affix undergoing phonetic erosion is sometimes abruptly
replaced by a conveniently available lexical stem with which it shares one or
more phonological segments. The new affix has the phonological shape of the
old independent stem, but acquires the basic grammatical function of the old
affix, though it may also bring in a portion of the stem's own morphological
and semantic idiosyncrasies. (728)
Under the hermit crab approach, LAs would have been a category of morphemes
not derived from nominal incorporation, but that exploited nominal incorporation to
acquire new phonological material. That is, in the traditional view incorporated nominals
became LAs through a process of grammaticalization and lexicalization that resulted in
the creation of an entirely new morpheme.60 Under the hermit crab view, after an existing
58
The similarities are striking in that Swedish employs a series of fogemorphemes, one in particular s, that
is mandatory (with the exception of in specific phonological environments), in compound constructions
with two, three or more nominals. Preliminary investigation suggests that in cases where an LA appears as
the second element of a tri-nominal compound, the mandatory fogemorpheme, s in this case, does not
appear (M. Hulden pc 28/6/06). This suggests that the morpheme, generally assumed to be free, is actually
functioning as an affix. Future inquiry should reveal whether or not these may serve as arguments or
classifiers as argued here for Cr.
59
For an analysis of the Hermit Crab account of grammaticalization with in the Minimalist Program see
Hill 2003.
60
Cf. Lightfoot 2005 for discussion of grammaticalization and lexicalization along the lines presented here.
112
LA lost a certain amount of phonological material, for whatever reason, the phonological
material, and possibly some semantic and morphological idiosyncrasies, of a
phonologically similar incorporated noun were reanalyzed as the LA.61 This means that
the LA that has the semantic content "people" and is presently realized phonologically as
at a previous time may have been realized by very different phonological material
=
that became so reduced that material from the incorporated =
case of I
replaced it.62 So in the
, (93a) would not have been the original LA, but rather a pre-existing LA
with some phonological element(s) similar to I
, perhaps I
. If this were the case,
then we would expect (94a) to, hypothetically, be the underlying morpheme which "took"
phonological material from the construction in (94b) (
or
) to produce
(94c). Thus, (94c) reduced phonologically to give us present day (94d).
( 94 ) a.
)
. I
3abs- shoot =people
'He shot somebody/something.' (lit. 'He people-shot.') unattested
b.
*
summon-conn-nom-Indian -sgimp
Call the people!
c.
)
.
I
3abs- shoot =people
'He shot somebody/something.'
61
It is not clear from Heath’s discussion if the Hermit Crab hypothesis would allow the co-existence of a
morpheme X that has taken phonological material from morpheme Y after the process had taken place. If
this were the case, then the Hermit Crab approach would be potentially valid here, if not, it would be
problematic for such an analysis of the Cr data.
62
The comparative method could shed some light on this issue, however, there are number of challenges to
such an approach. One such challenge is that in some languages within the family it is not always the case
that the same compound construction was reanalyzed as the same lexical affix in terms of family wide
cognates. This should not deter such future inquiry however.
113
d
)
. I
3abs- shoot =people
'He shot somebody/something.'
Under the hermit crab view, one generation of speakers simply added new
phonological material to a grammatical element that existed in the speech of a previous
generation of Salishan speakers. This contrasts with the traditional view that some
generation of speakers first initiated the use of a universal category available to all
learners of languages but not previously employed by Salishan speakers. If the Hermit
Crab analysis were correct, it would represent a purely grammatical process in the terms
of Lightfoot (2005). If the traditional view were correct it would illustrate the apparent
convergence of grammaticalization and lexicalization as discussed in Lightfoot. It may
also be the case that at different times in the history of Salish both may have been correct.
Kuipers (2002) identifies 56 proto-LAs for Salish. Of these 56 he links 18 to protoroots. Of course the remaining 38 may have links to roots in proto-proto-Salishan or they
may be original LAs not derived from incorporation. Attempting to determine which
account is accurate is beyond the scope of this chapter. However, this will not affect the
analysis that follows. The two potential analyses have been presented to acknowledge the
two possible interpretations of the data.
2.3. LAs in context
Up to this point it has been suggested that what Salishanists refer to as
compounding and lexical affixation resembles what is commonly identified as noun
incorporation in other languages (cf. esp. Mithun 1997:365).63 In the following discussion
63
Mithun (1997:365) notes that LAs, like incorporated nouns, carry no inflectional morphology for
definiteness, number, or case, and do not serve to specify "core" arguments of the clause. Further, like
incorporated nouns they may occur with independent nominals referring to the same kind of entity they
evoke, they are used to derive new lexical items, to manipulate the selection of core arguments, and show a
114
this comparison will be strengthened by considering the nature of the data. We saw above
in (86) that LAs can occur in nominal constructions and in (87) that LAs can occur in
verbal constructions. There seems to be no restriction on individual LAs in terms of being
confined to combine to form only nominal constructions (nominalizer+V+LA) or only
verbal constructions. That is, LAs appear freely in nominal or verbal constructions as the
following example using I
'face' illustrates. In (95a) we see I
construction and in (95b) we see I
( 95 ) a.
…
I
in a nominal
in a verbal construction.
MN
L
grease =face -3g
'… my face.grease …'
b.
L
J1
L
I
1nom- again- turn =face -m
'…I will turn back…'
(Doak 1997:291)65
LAs may also appear in constructions with multiple LAs. Reichard (1938) identifies 17
"compound suffixes" (624-625). In the following examples, we can see how LAs
contribute to meaning when multiple LAs appear together.
In (96) the first suffix I
second suffix I
( 96 )
'curved motion', indicates the manner of action while the
'water', indicates the goal of motion or destination:
(
.,
I
I
. ,
loc- jump =curved.motion =water det1 lake
'He jumped in the lake.'
(Doak 1997:292)
preponderance of items pertaining to body parts, among other similarities, each of these characteristics
being derived in Mithun (1984).
64
It should be noted that Reichard identifies
‘grease’ as a verb stem (1939:94).
65
For reasons of space the first line of each example is omitted here.
115
Thus we see that LAs can function to indicate manner and goal of motion. In (97) the first
suffix I
I
( 97 )
'back' restricts the meaning of the second I
'back' andI
'foot' (the combination of
'foot' indicates 'sole' (Reichard 1938:609.459):
. K. K.
K 9 :
I
I
cust- loc- lie.flat
=back =foot
'He had boards for sandals.'
In (98) the first suffix I
'hand' modifies the free root
insertion to hand insertion, while the second suffix I
(Doak 1997:292)
.7 'insert', from general
'water' indicates the goal of
motion of the action:
( 98 )
.7 1
)
.7 I
I
3abs- loc insert =hand =water
'He hand-plunged into the water.'
(Doak 1998:292)
Thus we see in(96) and (98) that LAs may serve as goals of motion.
LAs may also serve as themes in applicative constructions as (99) illustrates.
( 99 )
(
I
cut =wood -bt -1acc -3erg
'He cut me some wood (for me).'
(Doak 1997:307)
Doak (1997:308) notes that LAs can also serve as classifying elements as the
following examples illustrate. We first see a transitive construction where the determiner
adjuncts share reference with the -pronoun arguments (100), note the transitivizing
morphology (
the transitivizer, etc.).
116
( 100 )
+
O
.
H
)
H
shoot -dt -3abs -3erg det1
‘Don shot his enemy.’
enemy -3g det1 don
(Doak 1997:295-96)
Next we see an intransitive construction, indicated by the lack of any transitivizing
morphology, with an LA filling the theme role in the verbal argument structure and an
adjunct which shares reference with the null third-absolutive subject. Note it is the
determiner phrase
( 101 )
H
+
)
which shares reference with the 3abs- subject.
H
. I
H
3abs- shoot=body det1 Don
'Don shot somebody/something.'
(Doak 1997:296)
Finally we see the same construction as (101) with an additional adjunct,
enemy (indefinite)’66 in (102a) and
‘his
‘his enemy (definite)’ in (102b),
indicating the classificatory nature of LAs. In the example that follows, as the subject is
third person absolutive, the determiner adjunct
subject, and the adjunct
reference with the LA I
( 102 ) a.
.
H
. I
3abs- shoot=body
'Don shot his enemy.'
66
shares reference with the
, which indicates the indefinite patient, shares
+
)
H
H
enemy
det1 Don
(Doak 1997:296)
Recall Doak (1997:214-15) notes, “[a] determiner homophonous with the oblique may be used without
another determiner as an indication of indefiniteness.” I assume the
used here, while labeled obl
‘oblique’ by Doak is the indefinite determiner.
117
b.
+
)
H
. I
H
3abs- shoot=body det1 enemy
'His enemy shot Don.'
'Don shot his enemy.'
det1 Don
(Doak 1997:297fn107)67
In (102) we also see that adjuncts in constructions with LAs demonstrate the same
relationships in terms of reference with LAs:
indefinite patients, and det1,2,3 DP,
(102a) we see clearly that the
DP adjuncts sharing reference with
here, adjuncts causing ambiguity. Further, in
DP defines the subset of bodies that Don shot, with the
LA expressing the superset in similar fashion to classifiers.
Finally we see that LAs may appear in constructions with their stem counterparts.
In the first example we see the LA is linked to the adjunct.
( 103 )
- 1
9# I
1nom finish+ncr =house det1 1g- house
'I finished building my house.'
(Doak 1998:299)
In the second example the adjunct is linked to the subject, and not the LA, so it is the
Indian that is doing the shooting.
( 104 )
.
Ø-
. I
3abs- shoot=people det1 nom- Indian
‘An Indian shot somebody/something.’ (lit. ‘An Indian people-shot.’)
(Doak 1998:299.560)
67
Doak notes here that the informant identified this construction as “ambiguous” without providing a gloss.
I assume “ambiguous” here is in terms of who is shooting whom. As discussed in Chapter 1, adjuncts are
non-configurational and lead to such ambiguities.
118
Doak (1997) notes that LAs "carry semantic roles equivalent to a variety of
obliques such as locative and instrument" and that they serve as classifiers (308). We
have seen LAs "carry[ing] semantic roles" of instrument, goal of motion, theme, LAs as
classifiers, and expressing manner. We also saw that LAs parallel incorporated stems in
Cr, as in all Salishan languages, in their distribution, but lack the morphological elements
associated with stem incorporation. It was thus concluded that they were separate
phenomena. Another important parallel that was noted was that between adjuncts that
refer to indefinite arguments and adjuncts that refer to definite arguments. Specifically,
we saw that adjuncts that share reference with LAs may take the same form as those that
share reference with arguments of indefinite transitive constructions, the indefinite
determiner
, and the definite determiners. Finally, it should be noted that LAs seem to
function in a manner similar to incorporated nominals cross-linguistically, especially if
filling theme or goal of motion roles.
This has led Gerdts (2003) to argue that lexical suffixes "function exactly like
incorporated nouns,” they can "internalize a core argument such as theme or causee" and
thus affect core argument structure in Halkomelem (355).68 In the analysis that follows it
will be assumed that Cr LAs function in the same way as in Halkomelem, "exactly like
incorporated nouns.” In the analysis that follows it will be seen that LAs not only
function similar to incorporated nouns, but make use of the same mechanisms that
incorporation constructions employ.
Before moving to the formal analysis of LAs, it should be noted that while it has
been seen that LAs appear adjacent to the stem, there are a limited number of
68
For a slightly different view of LAs see Wiltschko’s (2006) discussion of Halkomelem LAs. Wiltschko
identifies LAs as defective nominal roots. Such an analysis seems problematic, since LAs never function in
any of the ways that roots do. Wiltschko (2006) argues that LAs in Halkomelem are deficient “nominal
roots.” This raises a number of questions beyond the scope of the present discussion; however, it seems
rather problematic to assume that a root can be inherently categorized, i.e. nominal here.
119
constructions in most Salishan languages where an intervening morpheme, generally a
non-control resultative or involuntary morpheme, appears between the LA and the stem.
These forms are rare in the data and little discussed in the literature. They will not be
addressed here, and are left to future research.
3. LAs: Little n
Here it will be argued, based on the facts described above, that LAs are n heads (“light
nouns”) in argument position that incorporate into the selecting predicate. It is thus
concluded that that LAs are not roots because:
1.
LAs often have a number of different semantic extensions compared to roots,
which have a richer, more fixed semantic core.
2.
LAs never serve as phonological words, unlike roots
3.
LAs never serve as the base for phonological words, as roots do
4.
LAs never appear alone with nominalizing or verbalizing morphology, as
roots do
5.
LAs never appear as the sole non-D-element within a DP, as roots do
6.
LAs constitute a relatively closed class of items, unlike roots
7.
They do not appear alone with possessive marking, unlike roots
Also, LAs are not -heads:
1.
The plural affix
does not refer to them in intransitive or transitive
constructions, as it may do with -heads69
Finally, they are not D heads as they cannot take a stem as their complement.
69
Doak, in (Doak and Mattina 1997), argues the plural affix
and "subject and object" pronominals, Ps (342).
co-references the "genitive pronominals"
120
The conclusion is that LAs are n heads that lack any higher functional
categorization such as .
4. Formal analysis of LAs
In this section a formal account of LAs will be presented. It will be assumed, based on
the facts of the data seen above, that lexical affixation is a process of incorporation as
described in Mithun (1984), Baker (1988 et series), Rosen (1989) among others.
However unlike Mithun and Rosen, but similar to Baker, it will be assumed that lexical
affixation is a purely syntactic operation, not a lexical operation in line with the tenets of
Distributed Morphology. That is, LAs adjoin to stems in narrow syntax via an
incorporation operation. Unlike Baker, Aranovich, and Golluscio (2004) however, it will
be argued, following Harley (2004), that head-movement, here incorporation, in narrow
syntax is accomplished by Hale and Keyser's (2002 esp. Ch. 3) Conflation mechanism.
As Harley notes, Conflation accounts for the constraints on head-movement-like
phenomena while maintaining Chomsky's (1995) proposal that head-movement is
'phonological' in nature.
In the next section a brief review of Hale and Keyser's (2002) Conflation is
presented which includes discussion of Harley’s (2004) discussion of Conflation and
head-movement. This is followed by an analysis of Cr LAs. Next we will return to the
issue of similarity between LAs and stem+stem incorporation in Cr as described above in
1. Finally, it is shown that this account of the data and of incorporation is to be preferred
to a Baker et al-style (2004) approach to incorporation.
121
4.1. Conflation revisited
Here Conflation will be briefly revisited to remind the reader of the relevant concepts
(readers are referred to section 2 of Chapter 3 for a fuller discussion), One further
consideration will be mentioned regarding Harley (2004). As noted in the previous
chapter, Hale and Keyser (2002) define Conflation as:
( 105 ) Conflation
Conflation consists in the process of copying the p[honological]-signature
of the complement into the p-signature of the head, where the latter is
"defective." (63)
Hale and Keyser's Conflation was summarized as follows:
( 106 ) Conflation: Key Assumptions
a.
b.
c.
A label of any constituent has all the features of the head, including some
representation of a phonological matrix, (the 'p-sig' of the head).
Conflation occurs when a constituent is merged with a sister head which
has a defective p-sig.
Economy requires that the conflated p-sig only be pronounced once, in its
upper most position.
(Harley 2004:3)
The following Mohawk example taken from Harley (2004:3) illustrates
Conflation as head movement once again.
( 107 )
Owira'a
waha'-wahr-ake'
Baby
AGR-meat-ate
‘The baby ate meat.’
(Baker 1988 in Harley 2004)
As Harley notes, in the numeration for VP we start with the roots, [N wahr-], 'meat'
(N wahr-) and [V -ake], 'eat' (Vake-). The derivation is outlined in (104). To initiate
Conflation, we assume Vake- has a 'defective' p-sig.
122
( 108 ) a. N wahr- merges with Vake'b. Because Vake'- has a 'defective' p-sig, the p-sig of Nwahr- conflates into the psig of Vake'c. The head, now with the p-sig Vwahrake', projects (i.e. is used for a label) giving
the set {Vwahrake', {Vwahrake', N wahr- }}, which can be illustrated in the
following tree:
Vwahrake'
Vwahrake' N wahrwahrake'
d. For economy reasons (because Vwahrake' is pronounced), N wahr- is not
pronounced.
One more notion regarding Conflation must be presented before moving onto the
analysis of LAs in Cr. Harley (2004:8) proposes the following constraint on Conflation.
( 109 ) Conflation Economy:
Conflation must occur as early as possible. That is, a [+affix] p-sig must copy
the p-sig of its sister during Merge: it cannot 'wait' to copy some later
available p-sig in a later Merge.
Finally, Harley notes that in order to trigger incorporation of an object, it has to be
assumed that V is assigned a [+affix]70 feature as it enters the numeration. She assumes
that this is generally possible for roots in English (7). The same will be assumed here for
roots in Cr, that they may be assigned a [+affix] feature indicating a defective p-sig. Now
we turn the analysis of LAs.
4.2. LAs: An incorporation account
Taking as our starting point Hale and Keyser's (2002) Conflation, we can account for
LAs rather straight forwardly. Given a construction like (104) above repeated here as
70
Again, for notational simplicity heads with defective p-sigs will be identified [+affix], however, it
should be noted that it is assumed that this feature only triggers the transfer of a p-sig under merge. No
movement (Copy and Remerge) is triggered.
123
(110), we see how Conflation works in Cr.
( 110 )
.
Ø-
. I
3abs- shoot=people det1 nom- Indian
‘An Indian shot somebody/something.’ (lit. ‘An Indian people-shot.’)
(Doak 1998:299.560)
Suppressing the adjunct and material above vP for simplicity, we have the basic structure
in (111) for (110). The steps relevant to deriving the tree in (111) are detailed in (112).
( 111 )
vP
P
.
v
v
Ø-
.
P
.
.
.
.
nI
.
( 112 ) a. The root . 'shoot' merges with its complement the LA I
b.
c.
'people'.
The root's p-sig is defective, and conflation occurs. The p-sig of the LA,
I
is copied into the defective p-sig of the root. The P is labeled with the
p-sig of its head, .
.
The P labeled .
merges with an element from the numeration, the v
which has a defective p-sig and conflation occurs. The p-sig of P, .
,
is copied into the defective p-sig of v, giving .
. Then the whole
constituent, a projection of v, is labeled with the p-sig of its head, .
The v labeled .
mergers with the subject, a null third absolutive
.
-head. Neither p-sig is defective, so no copying occurs. The whole
constituent is labeled with the p-sig of the head v, .
.
Following Déchaine and Wiltschko (2002), as n's, it is assumed that LAs lack any
functional level that would assign person, number or gender, -features. This lack of a
functional level with -features allows an LA to appear in an intransitive construction,
124
like (110) above, because there is no -feature agreement requirement for the LA to
satisfy, assuming that agreement is required of -heads only. Baker et. al. (2004) argue,
"[i]n such constructions the verb can only agree with the subject, so it is inflected like an
intransitive verb" (154). This allows LAs to saturate the object of a transitive argument
root in an intransitive construction, as in full noun-incorporation.
The applicative construction above in (99), here (113), further illustrates, again
silently suppressing the functional categories above ASP, as there is no further movement
of the morphemes under discussion beyond ASP as discussed in Chapter 3.71 The
derivation for the tree in (113b) is presented in (114).
( 113 ) a.
(
Ø-
I
comp- cut =wood -bt -1acc -3erg
‘He cut me some wood (for me).’
b.
ASP
ASP
I
(Doak 1997:307)
I
vP
I
P
I
P
v
v
I
I
I
APPLP
I
P
APPL
APPL
I
P
I
I
I
I
nI
I
71
McGinnis (1998) and Pylkkanen (2002) claim that the Applicative head is inserted between the v0 which
introduces the external argument and the V0 which represents the core ('root') meaning of the verb and
introduces the internal argument, if any. The same account is adopted by Harley, Tubino-Blanco and
Haugen (2006) for Hiaki Applicative constructions.
125
( 114 ) a. The root
'cut' merges with its complement the LA I
'wood'.
The root's p-sig is defective, and conflation occurs. The p-sig of the LA,
I
is copied into the defective p-sig of the root. The P is labeled with the
p-sig of its head,
I
.
b.
The P labeled
I
merges with an element from the numeration,
[+benefactive] APPL which has a defective p-sig and conflation occurs. The psig of P,
I
, is copied into the defective p-sig of APPL, giving
I
. Then the whole constituent, a projection of v, is labeled with the
p-sig of its head,
I
. Here we note that the APPL head is realized to
the right of the phonological material in P. The assumption here is that there
are two types of affixes: those with defective p-sigs and those without (the
bound pronouns for example). Affixes with defective p-sigs phonologically
align with conflated phonological material in accord with their morphological
affixal feature ([±prefix]). The APPLP is labeled with the p-sig of its head,
I
.
c.
The APPL labeled
I
mergers with the APPL object argument, first
person accusative. Neither p-sig is defective, so no copying occurs. Then, the
whole constituent, a projection of APPL, is labeled with the p-sig of the head
APPL,
I
.
d.
The APPLP labeled
I
merges with an element from the numeration,
a [+transitive] v, . This elements p-sig is defective, and conflation occurs.
The p-sig of the APPLP,
I
, is copied into the defective p-sig of v,
giving
I
. Then, the whole constituent, a projection of v, is labeled
with the p-sig of the head v,
I
.
e.
The v labeled
I
mergers with the subject P, third person
ergative. Neither p-sig is defective, so no copying occurs. Then, the whole
constituent, a projection of v, is labeled with the p-sig of the head v,
.
The direct object raises out to the specifier position of vP to check case.
The vP labeled
I
merges with an element from the numeration, a
I
f.
g.
[+completive] Asp, Ø-. This element's p-sig is defective, and conflation
occurs. The p-sig of the vP,
I
, is copied into the defective p-sig of
Asp, giving
I
. Then, the whole constituent, a projection of Asp,
is labeled with the p-sig of vP,
I
.
126
In this way we can account for the distribution of LAs and their syntactic
construction. No case of instrumental lexical affixation was accounted for. However,
analysis of such constructions would be the same, except that the incorporating LA in
such constructions would function as a modifier of the root to which it joins.
We also see that this analysis has advantages over Baker-style (1988, 1996; Baker
et al 2004) Incorporation in that head-movement is purely a phonological operation here,
rather than the movement of an entire syntactic head (see Chomsky 1995:321, 2001a:37;
and Harley 2004 for discussion of problems with the latter). In the next section it will be
argued further that the account presented here is preferable to Baker et al (2004) on
additional grounds.
4.3. Baker, Aranovich, and Golluscio (2004)
Baker et al (2004) point out that polysynthetic languages, like Cr, treat the phonological
realization of agreement morphology differently cross-linguistically in constructions with
incorporated nouns. The following examples taken from Baker et al illustrate such
phenomena. Languages like Mapudungun (115) and Chukchee (116) do not allow
agreement with incorporated nouns. The incorporated Mapudungun example (115a)
contrasts with the unincorporated example (115b) in that in (115a) agreement
morphology with the incorporated noun would render the construction illicit. In (115b)
we see that the agreement morphology sharing reference with the free standing DP is
necessary. In (115a) subject morphology is required but object agreement morphology is
forbidden. In (115b) both object and subject agreement are required.72
72
Here abbreviations from Baker et al (2004) are employed. They note (139):
The abbreviations used in our glosses include: ABS: absolutive, ADJ: adjectival suffix, APPL: applicative,
ASP: aspect, BEN: benefactive applicative, CAUS: causative, CIS: cislocative, CL: (noun) class marker,
COLL: collective, COMP: complementizer, COND: conditional, DS: dative subject, DUP: duplicative,
ERG: ergative, FACT: factual mood, FEM: feminine, FOC: focus particle, FUT: future, HAB: habitual,
127
( 115 ) a. Ngilla-waka-(*fi)-n
buy-cow-(*30)-IND.1sS
‘I bought a cow.’
b.
Ngilla-fi-
ti waka
buy-30-ind.1S the cow
‘I bought the cow.’
(Smeets 1989:421 in Baker et al 2005:141)
In the next example (116a) we see a construction without an incorporated noun. In
this Chukchee example the data contains the transitive suffix nin. In (116b) we see this
transitive morphology is not present in the construction with the incorporated noun.
( 116 ) a.
b.
tl g-e m tq -m t
kawkaw- k kili-nin
father-ERG butter-ABS
bread-LOC spread-3sS/3sO
‘The father spread the butter on the bread.’
tl g- n
kawkaw- k m tq -kili-g'e
father-ABS
bread-LOC
butter-spread-3sS
‘The father spread the butter on the bread.’
(Baker et al 2004:153)
Languages like Mapudungun (115) and Chukchee (116) can be contrasted with
languages like Southern Tiwa which allows object agreement morphology in
constructions with the incorporated nominal. In (117a) we see that when a noun is
IMP: imperative, IND: indicative, INDEF: indefinite, INSTR: instrumental, INV: inverse voice, LOC:
locative, NE: prenominal particle (Mohawk), NEG: negation, NOM: nominative, NOML: nominalizer, NP:
nonpast, NSF: noun suffix, PART: participle, partitive, PI: past imperfective, PL: plural, POSS: possessor,
PP: past perfective, PRES: present, PROG: progressive, PRT: particle, PUNC: punctual aspect, RE:
repetitive, REFL: reflexive, REP: reportative, SG: singular, STAT: stative, SUBORD: subordinate, VBLR:
verbalizer, _: epenthetic vowel. Glosses for agreement markers include: 3O: 3rd person object agreement;
1sS: 1st person singular subject agreement; 3sS: 3rd singular subject agreement; 3dS: 3rd dual subject
agreement; 3pS: 3rd plural subject agreement; AO, BO, CO, and CS: object and subject agreement for
particular number/gender combinations (Southern Tiwa); ZpS: zoic plural subject agreement (Mohawk).
Agreement is a complex matter in all the languages discussed, and the complexities vary from language to
language in ways we cannot go into here. See the cited sources for details.
128
incorporated there is no change in the agreement morphology when compared to a nonincorporated construction (117b).
( 117 ) a. Ti-seuan-mu-ban
1sS/AO-man-see-past
‘I saw the/a man.’
b.
Seuan-ide
man-sg
ti-mu-ban
1sS/AO-see-past
‘I saw the/a man.’
(Allen, Gardiner, & Frantz 1984 in Baker et all 2004:141)
To account for these apparent differences, Baker et al (2004) propose a deletion
mechanism that not only deletes the phonological features of the lower copy after
movement, but also, in some languages, deletes the person, number, and gender features
(PNG or -features). In languages like Mapudungun and Chukchee, the -features of the
copy left in object position after incorporation are deleted along with the phonological
features. Since there are no -features on the copy in the object position, the verb does
not need to agree with the incorporated nominal in object position. In short, the verb only
needs to agree with the nominal in subject position. In languages like Southern Tiwa on
the other hand, only the phonological features of the copy in object position are deleted,
leaving behind -features. In such languages the verb must agree with the -features in
the object position, so it is inflected transitively (153-54). In this way Baker et al account
for the differences in the two types of languages. This can be illustrated in the following
tree taken from Baker et al.
129
( 118 )
S
NP
I
VP
V
AGR
V
Ni
buy
cow
NP
Ni
cow
{SG, AN, FEM, 3} or {Ø}
(Baker et al 2004:154)
Baker et al (2004) go on to present a third type of language, Mohawk. Mohawk
seems to share features of both types of languages previously discussed: the verb does not
have to agree with the incorporated object, but it must show object agreement of some
kind. In (119) we see the incorporated direct object is third person neuter, the
incorporated version (119a) has the same agreement morphology as the unincorporated
version (119b).
( 119 ) a. Sak ra-[a]tya'tawi-tsher-a-nuhwe'-s
Salk MsS/(NsO)-dress-NOML-Ø-like-HAB
‘Sak likes the dress.’
b.
Sak ra-nuhwe's
ne atya'tawi
Sak MsS/(NsO)-like-HAB NE dress
‘Sak likes the dress.’
(Baker et al 2004:156)
In this example Mohawk appears to pattern with Southern Tiwa. However, Baker et
al note that the third neuter object in Mohawk happens to be a "Ø" form, and so (119a) is
also compatible with treating Mohawk like Mapudungun and Chukchee (156). They go
on to provide the following evidence to illustrate that Mohawk represents a third type of
language.
130
( 120 ) a. Sak ra-wir-a-nuhwe'-s
Sak MsS/(NsO)-baby-Ø-like-hab
‘Sak likes babies.’
b.
Sak shako-nuhwe'-s
ne owira'a
Sak MsS/FsO-like-hab
‘Sak likes babies.’
ne baby
(Baker et al 2004:156)
As Baker et al (2004) note, the agreement morphology in (120a) and (120b) differ,
but the agreement of (120a) is the same as (119a). The conclusion is that verbs in
Mohawk agree with direct objects in constructions with incorporated direct objects, only
the -features of such objects are the default -features of the language. In the case of
Mohawk then, the phonological material is deleted, but only the marked values of the
-features are deleted, leaving behind just the default -features (156). Thus they arrive
at the following typology, proposing a "family of parameters" to account for the crosslinguistic differences based on the facts presented and also on agreement facts related to
WH-traces:
( 121 ) The deletion procedure which applies to copies of moved elements
(i) preserve the PNG features [= -features] of the copy, or
(NI-traces in Southern Tiwa, Mayali)
(ii) delete the marked values of the PNG features of the copy, or
(NI-traces in Mohawk)
(iii) eliminate PNG features entirely.
(NI-traces in Mapudungun, Nahuatl, Chuckchee)
(Baker et al 2004:156)
Next we turn to an account of stem+stem incorporation in Cr to illustrated how a
Conflation account of incorporation is to be preferred over the Baker et al (2004)
analysis.
131
4.4. Stem+stem incorporation in Cr
In this section a preliminary account of stem+stem incorporation, as described in @1
above, will be put forth. Recall that the primary distinction between LA incorporation
and stem+stem incorporation was the presence of connecting morphology,
, and
categorizing morphology such as the nominalizing , both absent in LA incorporation.
Given (88a), repeated here as (122):
( 122 )
*
. ,
)
*
,
3abs- finish-conn-nom-write -m
‘He finished writing.’
It is proposed that the connective
, is actually a -head, as it is the next functional
layer within the DP, unspecified for person, number or agreement, thus allowing
incorporation into a stem. The following tree in (123) illustrates with the relevant steps to
derive the tree in (124).
( 123 )
vP *
P
,
v*
) v
*
)
,
,
*
*
,
*
,
P
,
,
,
n
,
n
v,
,
v,
,
,
,
132
( 124 ) a. The v
(middle) merges with its complement the root
,
'write'.
The v's p-sig is defective, and conflation occurs. The p-sig of the root, ,
is copied into the defective p-sig of the v, giving ,
. The v is labeled with
the p-sig of its head, ,
.
b.
The v labeled ,
merges with an element from the numeration, the
nominalizer , which has a defective p-sig and conflation occurs. The p-sig
,
is copied into the defective p-sig of the n, giving ,
. The n is
labeled with the p-sig of its head, ,
.
c.
The n labeled ,
merges with an element from the numeration, the
-head , which has a defective p-sig and conflation occurs. The p-sig
,
is copied into the defective p-sig of , giving
,
. The P is
d.
labeled with the p-sig of its head,
,
.
The P labeled
,
merges with an element from the numeration, the
root *
'finish', which has a defective p-sig and conflation occurs. The psig
,
is copied into the defective p-sig of the root, giving
. The root is labeled with the p-sig of its head,
*
,
.
The root labeled *
,
merges with an element from the
numeration, v, which has a defective p-sig and conflation occurs. The pig-sig
*
e.
d.
,
of *
,
is copied into the defective p-sig of the v, giving
*
,
. The v is labeled with the p-sig of its head,
*
,
.
The v labeled *
,
merges with an element from the numeration,
the third absolutive subject, ) . Neither p-sig is defective so no copying
occurs. The whole vP is labeled with the p-sig of its head, *
,
.
In this way we account for such constructions in Cr. As mentioned above, it is
assumed that the -head
is unspecified for -features. In Cr therefore, the P is not
required to agree with the verb, and thus need not raise to specifier of v to check features,
and incorporation is possible.73 Next we turn to a comparison of Baker et al’s (2004)
73
It should be noted that Reichard (1938) provides a few examples of nominal incorporation which do not
include the connecting morpheme . Each of these constructions is with the root
'hunt', a root that
cannot "stand on its own" (Doak 1997:287), and
'procure by hunting' which Reichard (1938) notes,
133
analysis and the Conflation analysis of incorporation.
4.5. Conflation over feature deletion
Baker et al 2004 propose the typology in (121) above, repeated here as (125), to account
for the morphological differences seen in the Southern Tiwa, Mohawk, Mapudungun, and
Chuckchee in regards to object incorporation.
( 125 ) The deletion procedure which applies to copies of moved elements
(i) preserve the PNG features [= -features] of the copy, or
(NI-traces in Southern Tiwa, Mayali)
(ii) delete the marked values of the PNG features of the copy, or
(NI-traces in Mohawk)
(iii) eliminate PNG features entirely.
(NI-traces in Mapudungun, Nahuatl, Chuckchee)
(Baker et al 2004:156)
They propose a family of parameters, and a special deletion mechanism (126) the result
of "several possible parameter settings" (156 fn22) to account for the differences in (i),
(ii), and (iii) in (125).
( 126 ) Baker et al's special deletion mechanism
(i) deletes all -features, or
(ii) deletes only marked values of -features, or
(iii) deletes no -features
"incorporates any of the animal names and seems quite free" (642). In the case on
'hunt', the
connective may be present, lost for phonological reasons, or the root may actually be
, with the
element being the remaining phonological content of , having been reduced. Since this root never
"stands on its own" we can't be sure, but it does seem reasonable to put it aside as it is not a standard root, it
cannot stand on its own. In the case of
'procure by hunting', it is a bit more challenging to make the case
that the connective is present, or that it is an exceptional case. It may be the case that
has a
phonologically null counter part in Cr, or that
has special properties that allow it to combine without the
relevant functional category. The exact nature of
incorporation will have to be left for future research,
but it will be assumed that the generalizations presented for noun-incorporation in Cr otherwise hold.
134
Further, Baker et al fail to capture Chomsky’s claim that head-movement is phonological
in nature. A Hale and Keyser (2002) Conflation analysis of incorporation, as seen above
in (@4.4), maintains Chomsky’s claim and does not require a special deletion mechanism
and the requisite parameters Baker et al (2004) propose to constrain it. The crucial
differences are illustrated in (127).
( 127 )
syntactic
head-
special deletion
mechanism
head-movement
phonological
movement
Baker et al
yes
yes
no
Conflation
no
no
yes
Under a Conflation analysis Baker et al's (2004) typology can be described as
follows:
( 128 ) Baker et al incorporation typology redux
(i) NI constructions function like non-incorporating structures in terms of
agreement, verb agrees, full features, or
(NI-traces in Southern Tiwa, Mayali)
(ii) In NI constructions verb requires agreement marking, but agreement
necessarily default, or
(NI-traces in Mohawk)
(iii) NI constructions do not function like non-incorporating structures, do not
require agreement, (no features).
(NI-traces in Mapudungun, Nahuatl, Chuckchee)
From this perspective it seems we need only account for the facts of type (ii)
languages and type (iii) languages, as nothing out of the ordinary seems to be occurring
in type (i) languages. That is, the object is incorporating in all three types of languages,
but in type (i) the syntax is doing nothing out of the ordinary, the only variation is
between conflation and no conflation. In languages of type (ii) and type (iii) the
difference is between no agreement and default agreement. If this were the case, we
135
would only need two parameters to account for the facts discussed so far. The first
parameter would distinguish between type (i) languages, the Southern Tiwa type, and
type (ii) and (iii) languages, Mohawk type and Mapudungun type, respectively.
( 129 ) Parameter 1: Incorporation Variance Parameter
a.
b.
Languages that incorporate show no agreement variation between
constructions with incorporation and constructions without incorporation: full
features are always present, or
(Southern Tiwa, Mayali)
Languages that incorporate show agreement variation between constructions
with incorporation and constructions without: no or default features are
realized in constructions with incorporation.
(Mohawk, Mapudungun, Nahuatl, Chuckchee)
This would account for the differences found between type (i) languages and the
other two types, type (ii) and (iii) languages. A second parameter would be required to
make a distinction between type (ii) and type (iii) languages.
( 130 ) Parameter 2: Incorporated
Parameter
a.
Languages that incorporate and show agreement variation between
constructions with incorporation and constructions without: default
features are realized in constructions with incorporation.
(Mohawk)
b.
Languages that incorporate and show agreement variation between
constructions with incorporation and constructions without: no
features are realized in constructions with incorporation.
(Mapudungun, Nahuatl, Chuckchee)
Cr would appear to be similar to Mohawk in that default agreement features, unspecified
in Cr, are required in cases of incorporation. On the other hand, it has been argued that
LAs have no functional layer providing agreement features, thus incorporation of LAs
136
would be similar to Mapudungun incorporation. Cr would appear to have both type (ii)
and type (iii) incorporation seen in (127) above: Type (ii) for incorporated nominalized
roots and type (iii) for LAs. However, if we assume the traditional view of LA diachronic
origin, LAs deriving from stem+stem incorporation, it would be the case that LAs would
have default agreement features. That is, though reanalyzed as bound morphemes the
-features of the connecting element
, would remain in the reanalyzed element: the
lexical affix. This would account for the fact that LAs, n-heads, incorporate. It would also
explain why no other n-heads or elements may incorporate without the connective
,
and the relevant -features. Thus, it would indeed be the case that default agreement
features, i.e. unspecified -features, are required in all case of incorporation in Cr. In this
way we see that Cr would be a type (ii) incorporating language.
This analysis makes it possible to account for the facts seen above with fewer
parameters and fewer grammatical mechanisms (i.e. special deletion elements). It also
maintains the notion that head-movement is phonological in nature. Naturally, at this
point the parameters proposed must remain hypothetical pending future research
involving a variety of incorporating languages. That is, it must be demonstrated that the
claims here do indeed hold in the languages under discussion other than Cr.
5. Conclusion
This chapter has argued for a formal account of lexical affixation as incorporation that
takes place within narrow syntax. It has been argued that LAs are n-heads, which do not
serve to categorize roots, but rather incorporate into roots. Also, that the incorporation
facts of Cr, both LA and stem+stem, can be accounted for without deviation from
Chomsky's (1995) claim that head-movement is phonological in nature by employing
Hale and Keyser's (2002) Conflation, as Harley (2004) demonstrates. Further, it has been
137
argued that this account of incorporation can be extended to incorporation in other
languages such as Mapudungun, Mohawk, and Southern Tiwa, which all exhibit different
agreement properties. This analysis requires fewer parameters and grammatical
mechanisms than Baker et al (2004), and is thus preferable. However, it will be necessary
for future research on languages such as Mapudungun, Mohawk, and Southern Tiwa,
among other such languages, to confirm these claims.
138
CHAPTER 5
ELEVEN PARTICLES IN COEUR D'ALENE: SOME THEORETICAL CONSIDERATIONS
1. Introduction
In this chapter eleven particles will be presented and analyzed in an attempt to arrive at a
preliminary account of functional projections in the Coeur d'Alene clause. The goal of the
chapter is to propose a hierarchy of clausal functional projections in Coeur d'Alene, in the
spirit of Cinque (1999). The chapter further compares the formalism of Cinque with that
of Rizzi's (1997a) Split CP hypothesis. The claim arrived at is that Cr data fits well with
both a Cinque- and Rizzi-style analysis in terms of the particles discussed. The chapter
relies on the English glosses and descriptions of data found in Reichard (1927-29; 1938)
and Doak (1997).74 The particles to be discussed are presented in Table 17.
74
It should be noted that the conclusions drawn in this chapter must be considered preliminary, pending the
complete analysis of all the Reichard (1927-29) manuscripts and future inquiry.
139
Table 17: Coeur d'Alene particles
TYPE
PARTICLE
Temporal Adverbial
'soon' immediate future (Doak 1997:186)75
'soon' immediate future (Doak 1997:187)
'soon' immediate future (Doak 1997:187)
Sentential Adverbial76
'and' / 'then' discourse/narrative adverbial
'and' / 'then' discourse/narrative adverbial
Mood
irrealis (Reichard 1938:669.777; Doak 1997:188)
Modal
future intentional, permissive, mild request
(Reichard 1938:666-67)
ought, obligation (Reichard 1938:669.780)
was to be but isn't, possibility (Reichard
1939:104)
Aspectual
'used to' terminative (Doak 1997:49)
'always' habitual (Doak 1997:49)
Based on the strict ordering of morphemes found in the data, it will be argued that
the above particles reveal a hierarchy of clausal functional projections in Coeur d'Alene
(henceforth Cr). Further, it will be shown that the structural ordering of the particles
discussed parallels that proposed by Cinque (1999) for a universal hierarchy of functional
projections, with minor modifications.
The chapter is organized as follows. In Section 2 the temporal adverbial particles
and sentential adverbial particles will be presented and analyzed. The three modal
particles are analyzed in Section 3 along with the irrealis. In Section 4, aspect
morphology will be revisited and the two aspectual particles will be discussed. The
interrogative particle is presented in Section 5. Section 6 presents an alternative analysis
75
The exact difference between these morphemes is not clear.
and
should not be confused with standard conjunctions, these are adverbials anchored to
speech time (Cinque 1999:12). Further, it should be noted that they not only occur in narratives but are
frequently employed in discourse (Brinkman pc).
76
140
of the irrealis and interrogative particles in terms of Rizzi's (1997a) split CP hypothesis.
Finally, in Section 7 concluding remarks are presented.
2. Temporal and sentential adverbial particles
As noted in Chapter 1, Cr does not indicate tense with overt morphology other than in the
form of temporal adverbial particles and modal particles that indicate future events. A
given clause in Cr is considered past or present, when no future marking is present,
depending on discourse context (Doak 1997:43). Cr employs three temporal adverbial
particles,
,
, and
may occur with the future Modal
, to indicate immediate future, each of which
. As noted by Comrie (1976) and Palmer (2001), it
is not uncommon for languages that do not employ overt tense morphology to encode
temporal notions via modals and adverbial particles. Cr also employs what Cinque
(1999:12-23) identifies as "sentential adverbials" anchored to speech or narrative time,
the Cr particles
and
, both glossed 'and' and 'then'. First, an analysis of the
temporal adverbial particles is presented followed by a discussion of the sentential
adverbial particles.
2.1. Temporal adverbial particles:
,
, and
Doak (1997) notes that Cr has a number of ways in which to indicate future events.
Among these are the use of the temporal adverbial particles
, to indicate immediate future. Doak notes that
related to
,
, and
, and
are
and are used in similar environments (187). They are all three glossed as
immediate future, 'soon'. Further, it should be noted that these adverbial particles differ in
their semantic content from the irrealis morpheme
which indicates a hypothetical
situation, in that the immediate future morphemes indicate a specific action to be taken
141
once a situation or event occurs (Doak 1997:189). They also differ in semantics from the
future morpheme
which may indicate future permission, a request regarding future
time, or intentionality for a future event (Reichard 1938:666).
As Doak (1997) claims the three temporal adverbial particles appear in the same
environments, the discussion will be limited to
here, with the assumption that the
conclusions arrived at hold for all three particles. In (131) it can be seen that the temporal
adverbial particle precedes the predicate.
( 131 )
4+, -
-
4.,
imfut
pay -dt -2pacc -1serg
‘... I will pay you something ...’
(Reichard 1927-29:cssw159)
In (132) it can be seen that the temporal adverbial particle may also appear twice in a
construction, once preceding a topicalized (adjoined) DP and again preceding the
predicate.77
( 132 )
9#
)
imfut det1 finish+rdp<aug> imfut
loc- give -? -ct -3abs -1serg
‘When I finish (my harvesting) I'll pay ....’
(Doak 1997:187)
That the temporal adverbial particles may appear higher than a topicalized element
and adjacent to the predicate in the same clause, suggest that there are two temporal
adverbial positions in the Cr clause, a "higher" position and a "lower" position. This
follows Cinque's (1999) claim that there are two adverbial positions, one higher in the CP
77
Doak (1997:255) notes that DPs to the left of the predicate may serve as topics. It is not clear, however, if
this is always the case given the non-cofigurational characteristics of Cr. In the discussion that follows it
will be assumed that the forms presented are topics, keeping in mind that they may potentially be adjuncts
attached to TP/IP.
142
and one lower near the vP edge above the DP subject (16). Next we will turn to a brief
discussion of the sentential adverbial particles
2.2. Sentential adverbial particles:
and
, 'and/then'.
and
As noted, Cr employs what Cinque (1999:12-23) identifies as "sentential adverbials"
anchored to speech/narrative time, the Cr particles
'then'.
and
and
, both glossed 'and' and
appear separately (133a) and (133b), and together (134), clause
initially. When the two appear together
appears before
neither Doak nor Reichard provide discussion of
and
. It should be noted that
. Further, in a number of
the narratives Reichard (1927-29) does not provide an English gloss for these forms,
though it is understood that they indicate 'and' or 'then'.
( 133 ) a.
- !
)
!
then 3abs- went det3 Coyote
‘Then Coyote went.’ (lit. ‘Then he went, Coyote.’)
(Reichard 1927-29:cssw11)
b.
*
8''
/P
*
then 2plnom- plural- be.silent
‘Then you were all quiet.’
78
(Reichard 1927-29:cssw174)
Here, vowel lengthening is employed to indicate exaggeration. Also, the morpheme is *
2 <. 2 is
present. The exact nature of this morpheme is not clear. Doak (1997) does not identify it as a plural
marking morpheme in the person marking paradigm, and only presents
as a plural marking element
(55). Reichard (1938) refers to *
as a "verbalizing or demonstrative pluralizing element" (594.389).
Doak (pc) notes that this plural marker is not a pronominal and it is used to indicate a distributive (“each”
or “a piece”; “every one of you”) or collective (“the army”; “the trees”; “you all”) type plural; the suffixal
marker only applies to third persons.
143
( 134 )
)
and then trap.by.surrounding
-ct
‘And then he was surrounded by them.’
-3abs -nte
(Reichard 1927-29:L084)
That
and
appear together suggests that there are two adverbial positions
at the top of the structure both suitable for
conclusion comes from data where
adverbial particle. In such cases
and
or
and
. Further evidence for this
appear with an immediate future
appears higher in the structure than the
temporal adverbial particle. This suggests that there are two types of adverbial heads,
sentential/narrative and temporal, rather than multiple specifier positions, as the two
adverbial types are of a different nature, as Cinque (1999:14) notes.
( 135 ) a.
''
)-
then imft
deitic 3abs- say u u u u
‘Then, soon he said: "u u u u"’
(Reichard 1927-29:cssw102a)
b.
+
555/Q
)
then imfut 3abs- again- loc- say
‘Then, soon again there he will say ...’
That
and
are in higher adverbial positions is further demonstrated by such
examples as the following where
topicalized subject,
79
(Reichard 1927-29:cssw050)
and
appear higher in the structure than the
'Chief', in the following example.
It is not clear the exact nature of the morpheme
'again' / 'back' and the locative , is a sequential
marker, meaning something like "then, in turn", (Doak pc), an analysis of these morphemes will not be
presented here.
144
( 136 )
)
and then det3 Chief
3abs- say
‘And then, Chief, he said ...’
Thus we see that temporal adverbial particles may appear in "higher" adverbial
positions and "lower" positions, (132) above, and that sentential adverbial particles may
appear in "higher" adverbial positions as well. As sentential adverbial particles may
appear together above a topicalized element (134), and may appear above a temporal
adverbial particles (135), it is assumed that Cr has three adverbial positions, two higher
and one lower. In the case of Cr, sentential adverbs appear in the topmost adverbial
position while temporal adverbs occupy the next two adverbial positions (this is
illustrated in (137) below).80 This parallels Cinque's (1999) claim that adverbials may
appear in a higher or lower position within the clause. Thus a preliminary hierarchy can
be proposed to reflect these facts in (137a). These facts are summarized in (137b), where
the highest adverbial positions can be filled by either sentential adverbs (135) and the
lower positions by temporal adverbial particles (132). It should be noted that Cinque
provides no position for topics in his articulated structure; it is included here, and
throughout in a location that accords with the known facts of the Cr clause.
80
Cinque 1999 does not provide a position for sentential adverbials anchored to speech time in the
languages he analyzes as they appear to be much freer than the Cr forms (12-13). As the Cr forms seem
restricted, an adverbial head is posited here as their host.
145
( 137 ) a.
adverbial > adverbial > topic > adverbial > vP
b.
AdvP
AdvPSENTENTIAL
AdvP
AdvPTEMPORAL
TopP
AdvP
AdvPTEMPORAL
vP
Next we turn to distribution of the modal particles to further expand the functional
hierarchy of the Cr clause.
3. Modals:
,
, and
In the discussion that follows it will be demonstrated that there is a fixed order between
adverbial particles and modals. It will also be demonstrated that
and
further
demonstrate a fixed order of functional categories within the Cr clause. It should be noted
that while three modals will be discussed, there are potentially others in the language that
will not be addressed here and left to future research.
3.1.
future intentional, permissive, mild request
Reichard (1938) notes that
indicates future permissive, future mild request, and future
commands (666-667). It may also indicate future intentional, and when doing so often
occurs with the intentional morpheme s- (666). It is these deontic senses of the term that
require the modal analysis, rather than a simple future tense analysis. The fact that
appears after the irrealis
morphology, suggest that
illustrate.
and temporal adverbial particles, and before aspect
is located between TP and ASPP. The following examples
146
( 138 )
temporal adverbial particles before irrealis
-
(
)
imfut
irr
again- loc- 3abs- go
‘Soon he will come.’
( 139 )
(Reichard 1927-29:csssd154)
following the irrealis
J
J I
irr fut loc- barter =face -gen
‘How do you intend to pay?/How will you pay for it?’
(Reichard 1927-29:cossd96)
( 140 )
following a temporal adverb
'
imfut
fut 2nom- int- first- eat
‘Soon you are going to be the first to eat.’
before aspect morphology
( 141 )
(
continuative
8*
)
)
8*
3abs
3abs !
R
‘They thought, whoever gets first …’
In (139), and (140) we see that
adverbial particle
appears before the irrealis
morphology
(Doak 1997:187)
?;5R <
(Reichard 1927-29)81
occurs after the irrealis
and temporal
. In (138) we see the temporal adverbial particle
. Finally, in (141), we see
appears before the aspect
, which is contained within the predicate. This suggests that
, head of
a MODP, is lower in the structure than the irrealis marker and the higher ADVP, but
higher than an ASPP. Further, the data above suggests that the irrealis appears between
81
This example comes from “The Coeur d’Alene Attack”, line 6 of page 2. I thank Ivy Doak for directing
me to it. The analysis is mine.
147
the higher ADVP and the
MODP. Thus, the preliminary hierarchy in (137) can be
expanded as in (143a) and illustrated structurally in (143b). First, however, we can show
that the irrealis is lower in the structure than the topic with the following middle
construction where the topic phrase,
'Grizzly Bear', appears before the
.4
irrealis.
( 142 ) topic,
before irrealis
.4
.4
- *.4
*
then det3 Grizzly.bear irr 1sg.nom- call -mdl
‘Then, Grizzly Bear, I should call.’
(Reichard 1927-29:chkt21)
Now the preliminary hierarchy can be presented with the irrealis included as a
mood element. However, it should be noted that it is not clear exactly where the "lower"
adverbial position is located in the structure. It will be assumed, pending future inquiry,
that the lower adverbial is located above the ASPP and above the resting place of the
pronominal arguments, and below MODP. As this is tentative it will be indicated by the
"≥" symbol.
( 143 ) a. adverbial > adverbial > topic >moodirrealis > modalability/permissive ≥ adverbial ≥
aspect > vP
148
b.
AdvP
ADVSENTENTIAL AdvP
ADVTEMPORAL TopP
TOP
MOODP
MOODIRREALIS
MODALP
MODALABILITY/PERMISSIVE
AdvP
ADVTEMPORAL
ASPP
vP
Next we turn to a discussion of the other two modal elements,
3.2.
and
and
.
: ought and possibility
Reichard (1938) provides an account of both
a sense of obligation (669),
and
.
is described as indicating
indicates possibility or "intention to carry out" but
failure to carry out (670). This
identifies as the aspectual particle
is quite similar in meaning with what Doak (1997)
. The difference seems to be that the modal
indicates a level of possibility that the aspect morpheme does not. I treat these as separate
morphemes here, I do not rule out the possibility that they may be the same morpheme
however.
There are no examples of
and
co-occurring with the morphemes discussed
above with the exception of one example in which
appears with
. This is
illustrated in (144).82
82
Doak (pc) notes that this example Reichard (1938) took from the story of Rabbit and Jackrabbit, and is
not a complete sentence. She notes that this example has S ;Tpreceding in the story, and is translated by
Reichard (1927-29) as "that which I was going to feed you with"; Doak claims the "but now I won't" is at
best contextual. Future research regarding this morpheme should clarify its exact nature.
149
( 144 )
Rposs fut share –t -2acc -1er
‘I was going to share it with you but now I won't.’
(Reichard 1938:667.763)
That
and
can appear together suggests that there are two modal positions.
It will be assumed that there are two modal heads, (though again, this assumption is
subject to future inquiry) one which hosts
and one which hosts
. The proposed
hierarchy can be stated as (145), noting that there is no clear indication of where
'ought' fits in the structure. Thus, for the time being,
'ought' will not be included in
the preliminary hierarchy of functional categories.
( 145 )a.
adverbial > adverbial > topic >moodirrealis > modalpossibility >
modalability/permissive ≥ adverbial ≥ aspect > vP
150
b.
AdvP
ADVTEMPORAL/SENTENTIAL AdvP
ADVTEMPORAL/SENTENTIAL TopP
TOP
MOODP
MOODIRREALIS
MODALP
MODALPOSSIBILITY
MODALP
MODALABILITY/PERMISSIVE
AdvP
ADVTEMPORAL
ASPP
vP
Next we turn to an account of the aspectual particles.
4. Aspectual particles:
Before
and
and
are addressed, a brief review of the aspectual morphology in Cr is
in order. As noted in Chapter 1, there are three aspect affixes which appear within the
predicate. These are the default completive, a null morpheme ')'; the customary
the continuative
. Examples follow.
( 146 ) completive null [) B: A situation has ended
a.
)
1s.nom- comp- smoke
‘I smoked.’
(Doak 1997:83)
; and
151
b.
)
*
comp- call -t -1acc -2erg
‘You called me.’
( 147 ) customary
(Doak 1997:119)
:A situation that is viewed as characteristic of a whole period
rather than of a moment.
a.
1p.nom cust- work
‘We work.’
(Doak 1997:85)
b.
cust- accompany -dt 2p.acc -1p.erg
‘We go with you folks.’
(Doak 1997:115)
:A situation in progress
( 148 ) continuative
a.
*
1s.nom cust- see
‘I am seeing.’
(Doak 1997:106)
Does not occur on transitive stems83
(Doak 1997:106)
b.
Cinque (1999) proposes the following hierarchy for continuative, customary, and
completive.
( 149 ) a.
b.
aspectcontinuative > aspectcustomary > aspectcompletive
ASPP
ASPPcontinuative
ASPP
ASPPcustomary ASPP
ASPPcompletive
83
Doak notes that she has “examples of the continuative occurring with transitive stems; however, they
are usually in subordinate clauses (one speaker seemed able to use this construction in main clauses)” (pc).
152
There is no evidence for this structure in Cr, however, there is no evidence that this
does not exist in Cr either. For clarity the continuative, customary, and completive will
all be assumed to be located in one ASP node. Further, since they never appear together, it
will be assumed that they are generated as ASP-heads in that same structure. The fact that
these morphemes may be internally ordered in terms of Cinque's (1999) hierarchy will be
indicated with the "≥" symbol as follows.
aspectcontinuative ≥ aspectcustomary ≥ aspectcompletive > vP
( 150 ) a.
b.
ASPP
ASPPcontinuative ≥ customary ≥ completive vP
Turning to the aspect particles we see that the habitual
'always', follows the
sentential adverbial particles and precedes the continuative aspect.84
( 151 )
and then always cont- cry -cont
‘And then, he is always crying.’
In the following example we see that the terminative
(Reichard 1927-29:cssd070)
appears before the customary
.
( 152 )
*
;<
;;
‘I used to see it.’
)
.?
;<*
(Doak 1997:49 modified)85
Thus it can be concluded that the two aspect particles
84
and
, appear
Doak (1997) labels as a continuative element that appears in some constructions with the
continuative creating a continuative antipassive (46,106). Interesting as this structure is, a full analysis of
will be left to future research. Doak notes that she has recently been analyzing this % as part of an - +
morpheme, rather than analyzing them as two separate morphemes (pc).
85
Doak (1997:49) does not provide the morpheme by morpheme analysis for this example.
153
higher in the structure than the aspect affixes, customary
, continuative
completive Ø-. Further, it has been shown that the aspect particle
, and
appears below the
sentential adverbial particles. The preliminary conclusion is that there are at least two
ASP nodes, one for the particles
continuative
and
, and the other for the affix customary
,
, and completive Ø-. Again we see that it cannot be demonstrated that
the Cr data conforms exactly to Cinque's (1999) universal hierarchy. Cinque argues that
habitual aspect precedes terminative aspect, but here we have no evidence for this in Cr.
However, Cinque places habitual aspect and terminative aspect above continuative,
durative (customary), and completive which is consistent with what we have seen in Cr.
Therefore employing "≥" to indicate potential ordering, (150) can be expanded to (153).
In (153a) the tentativeness of the morpheme hierarchy is illustrated by placing the two
and
aspect particles
continuative
under a single node, and the aspect affixes customary
,
, and completive Ø- under a separate single node.
( 153 ) a. aspecthabitual ≥ aspectterminative > aspectcontinuative ≥ aspectcustomary ≥ aspectcompletive
> vP
b.
ASPP
ASPPhabitual ≥ terminative
ASPP
ASPPcontinuative ≥ customary ≥ completive
vP
There is no direct evidence for the exact position of realization for the aspect
particles
and
in terms of the other particles discussed above, other than below
the sentential adverbial particles
and
. Therefore, they will be tentatively placed
in the structure in the position predicted by Cinque's (1999) universal hierarchy. Thus,
(145) can be restated as (154), where (154a) indicates the possible hierarchy in terms of
Cinque 1999 for each morpheme and (154b) reflects what is known from the data.
154
( 154 )a.
adverbial > adverbial > topic >moodirrealis > modalpossibility >
modalability/permissive ≥ adverbial ≥ aspect habitual ≥ aspect terminative >
aspectcontinuative ≥ aspectcustomary ≥ aspectcompletive > vP
b.
AdvP
ADVTEMPORAL/SENTENTIAL AdvP
ADVTEMPORAL/SENTENTIAL TopP
TOP
MOODP
MOODIRREALIS
MODALP
MODALPOSSIBILITY
MODALP
MODALABILITY/PERMISSIVE
AdvP
ADVTEMPORAL
ASPP
ASPPhabitual ≥ terminative
ASPP
ASPPcontinuative ≥ customary ≥ completive
vP
Having arrived at a preliminary account of the adverbial particles (both temporal and
sentential), the mood particle, the modals, and finally the aspect particles, we turn to a
discussion of the interrogative and Cinque's MOODSPEECH ACT.
155
5. Interrogative
Reichard (1938) notes the interrogative "stands first in a sentence'" (682). She further
notes that it is often used rhetorically. When it appears preceding the negative
, an
answer of "yes" is expected and when it appears before the sentential adverbial particle
, the answer "no" is expected (668). Following examples illustrate.
( 155 )
+<)
.<
intr 3abs- be.very.long.time
‘Hasn't it been a long time ...’
( 156 )
-
-devel
(Reichard 1927-29:cssw165)
(
intr then 1s.nom int- think
‘Was I to know.’
( 157 )
-
(Reichard 1927-29:L99)
I
intr neg fut int- loc- loc- know =behind -ct
)
-3abs -1sg.erg
‘Am I not to know what it is coming from behind?’
(Reichard 1927-29:cssw22)
Cinque (1999) argues that interrogatives, which are speech act mood elements
under Bybee's (1985) typological system, are speech act mood under his own system also
(53). This places interrogatives at the leftmost edge of the clause in Cinque's hierarchy
(130). Thus, (154) can be restated as (158), where (158a) indicates the possible hierarchy
in terms of Cinque 1999 for each morpheme and (158b) reflects what is known from the
data regarding these particle heads.
156
( 158 ) a. moodspeech act > adverbial > adverbial > topic >moodirrealis > modalpossibility >
modalability/permissive ≥ adverbial ≥ aspect habitual ≥ aspect terminative >
aspectcontinuative ≥ aspectcustomary ≥ aspectcompletive > vP
b.
MOODP
MOODSPEECH ACT AdvP
ADVTEMPORAL/SENTENTIAL AdvP
ADVTEMPORAL/SENTENTIAL TopP
TOP
MOODP
MOODIRREALIS
MODALP
MODALPOSSIBILITY MODALP
MODALABILITY/PERMISSIVE
AdvP
ADVTEMPORAL
ASPPhabitual
ASPP
≥ terminative
ASPP
ASPPcontinuative ≥ customary ≥ completive
vP
Comparing (158b) with (159), we see the Cr facts parallel Cinque's (1999) structure
quite closely, the differences being that a head position has been posited for the sentential
adverbials in Cr which does not appear in Cinque's formulism. Further, a topic position
has been included for Cr; where as, Cinque (225 fn25) acknowledges a topic position
within CP along the lines of Rizzi (1997a) but does not include it in his hierarchy of
functional projections. Also, the ASP heads have been collapsed into one note pending
further inquiry in Cr.
157
( 159 ) Cinque's 1999 Universal hierarchy of clausal functional heads (with relevant
heads only)
MOODP
MOODSPEECH ACT
ADVP
ADV TEMPORAL
MOODP
MOODIREALIS MODAL
MODPOSSIBILITY
MODALP
MODABILITY/PERM
ASPP
ASPHABITUAL
ASPP
ASPTERMINATIVE ASPP
ASPCONTINUATIVE
ASPP
ASPCUSTOMARY
ASPP
ASP COMPLETIVE
ADVP
ADV TEMPORAL vP
It should be noted that a tense position has not been indicated. Cinque (1999)
posits three tense positions, tense past immediately above tense future, both immediately
above mood irrealis, and a third tense position, tense anterior, immediately above aspect
terminative. As Cr does not indicate tense overtly in a tense node, and since discourse
factors determine the past or present nature of a clause, any generalization regarding
tense is rather speculative. However, since it is assumed tense enters an Agree
relationship with the subject of a clause in order to check case (cf. Chapter 3), it is
assumed that Cr minimally has a tense position near the mood irrealis position. Thus, a
158
hierarchy of functional and adverbial heads in the Cr clause is arrived at based on the
ordering of particles identified in Table 17 above, repeated here as Table 18.
Table 18: Coeur d'Alene particles
TYPE
PARTICLE
Temporal Adverbial
'soon' immediate future (Doak 1997:186)
'soon' immediate future (Doak 1997:187)
'soon' immediate future (Doak 1997:187)
Sentential Adverbial
'and' / 'then' discourse/narrative adverbial
'and' / 'then' discourse/narrative adverbial
Mood
irrealis (Reichard 1938:669.777; Doak 1997:188)
Modal
future intentional, permissive, mild request
(Reichard 1938:666-67)
ought, obligation (Reichard 1938:669.780)
was to be but isn't, possibility (Reichard
1939:104)
Aspectual
'used to' terminative (Doak 1997:49)
'always' habitual (Doak 1997:49)
Next we turn to a brief discussion of how the irrealis and interrogative may fit into Rizzi's
(1997a) split CP.
6. Rizzi's split CP
Rizzi (1997a) analyzes the structural representation of the CP as comprising independent
non-V-related material, unlike that in IP. That is, Rizzi claims that the inflectional
properties which C reflects are not encoded in the form of verbal morphology. Rizzi
identifies this complementizer layer as being typically headed by a free functional
morpheme, and hosting various operator elements such as interrogative and relative
pronouns, focalized elements etc., and topics. Importantly, Rizzi articulates the
Force-Finiteness system within CP.
159
Rizzi (1997a) proposes that CP is the "interface between a propositional content
(expressed by IP/TP) and the superordinate structure (a higher clause, or possibly, the
articulation of discourse)" (283). Thus, the CP encodes two types of information:
i.
that oriented toward the supra-ordinate structure (governing
clause or discourse), "Force" (Chomsky's (1995) term) which
expresses the fact that a sentence is a question, a declarative,
an exclamative, a relative, a comparative, or adverbial of a
certain type etc. (283).
ii.
that oriented toward the propositional content (expressed by IP
or VP), "Finiteness" which allows C to express a distinction
related to tense but as Rizzi states it, "more rudimentary
than tense and other inflectional specifications" (284). The core
IP/TP-related characteristics that the complementizer system
expresses are finiteness. The specification for finiteness within
the C system selects an IP/TP system with the familiar characteristics
of finiteness: mood distinctions, subject agreement licensing
nominative case, etc. (283).
More specifically, Rizzi argues that finite forms manifest irrealis/realis distinctions in a
Fin0 (284). Further he posits a head, Force0, within the complementizer "space," which
marks the illocutionary force of the sentence, which is distinct from, and higher than, the
other heads in C. In short, the irrealis
appears to pattern with what Rizzi describes as
a FinP element, as it encodes an irrealis/realis distinction, and the interrogative
appears
to qualify as a ForceP head in Rizzi's articulated structure. Rizzi's articulated CP structure
is illustrated in (160).
160
( 160 ) Rizzi's (1997a:297) articulated CP
FORCEP
FORCE0
TOPP
TOPP0
FOCP
FOCP0
TOPP
TOPP0
FINP
Fin0
IP/TP
Benincá (2001) reanalyzes Rizzi's (1997a) data and proposes an alternate structure to that
in (160), where the lower TopP is raised above the FocP giving the following hierarchy
of elements within the CP.
( 161 )
Force > Topic > Topic > Focus > Finiteness > Infl/Tense
Watanabe (2004) arrives at a similar hierarchy for Ancient Japanese.
( 162 )
Force > Topic > Focus > Finiteness > Infl/Tense
Although Watanabe (2004) argues for only one Topic Phrase, he does argue that
there are two specifiers in the Topic Phrase. This would account for the differences
between (161) and (162), the difference being that under Benincá's analysis, Italian has
two Topic heads adjacent to one another, and Watanabe proposes that Ancient Japanese
had one Topic head with two Topic specifiers.
Above it was shown in (142) that Cr has one topic position between the irrealis
and the interrogative
. Assuming
to be a Finite head and
to be a Force head, Cr
161
would appear to demonstrate a similar structure to Rizzi's (1997a) split CP framework,
modified by Benincá (2001). This can be illustrated in (163).86
( 163 ) Cr "split CP"
FORCEP
FORCE0
TOPP
TOPP0
FINP
FINP0
IP/TP
The facts presented above seem to be well on the way to supporting Rizzi's
(1997a) claim of an articulated structure as seen in (161). However, it is not possible to
ascertain if the facts support Rizzi's claims over Cinque's (1999) claims, or if the two
analyses are just notational variants on one another, as far as the Cr data is concerned.
Cinque claims that his Mood0speech act, the interrogative
here, is a different element than
Rizzi's Force0, although exactly how different is not clarified (84). Further, Cinque's
Mood0irrealis appears lower than any Tense head in the structure, whereas, Rizzi's Fin0,
which encodes irrealis, appears above any Tense head. As Cr has no overt morphological
element indicating a Tense head, it is not clear if the irrealis
is higher or lower than
Tense0. Further, Cinque suggests that Rizzi's account may be on the right track,
especially in terms of Topic and Focus positions, but that the CP has more elements
86
Doak (1997) presents Cr as a Pronominal Argument Language, and thus under the Pronominal Argument
Parameter of Jelinek (2004), it would be assumed that the predicate is located in a Focus position within the
CP. Whether or not the predicate is indeed in a Focus position will have to be determined by future
research. However, that the predicate remains below MODP suggests no raising occurs. That is, under
Jelinek’s hypothesis the entire VP raises to a specifier position of FOCP.
162
cross-linguistically than Rizzi suggests (225 fn25). In short, the facts neither entirely
support, nor contradict either of the distinct structures proposed by Rizzi and Cinque.
7. Conclusions
It has been argued that the strict ordering of particles discussed above reveal a hierarchy
of adverbial and functional heads within the Cr clause. Further, the ordering parallels that
of the universal hierarchy proposed for adverbial and functional heads cross-linguistically
by Cinque (1999). The structure (158b) is presented again here as (164).
163
( 164 ) Proposed hierarchy of adverbial and functional heads in Cr
MOODP
MOODSPEECH ACT AdvP
ADVSENTENTIAL AdvP
ADVTEMPORAL TopP
TOP
MOODP
MOODIRREALIS
TP
T
MODALP
MODALPOSSIBILITY
MODALP
MODALABILITY/PERMISSIVE
AdvP
ADVTEMPORAL
ASPP
ASPPhabitual ≥ terminative
ASPP
ASPPcontinuative ≥ customary ≥ completive
It has further been argued that the data regarding the irrealis
vP
and interrogative
tentatively fit within both a Cinque-type (1999) and Rizzi-type (1997a) analysis. The
conclusion to be drawn from this is that the generalizations that Cinque and Rizzi propose
are on the right track, and that further attempts at applying those generalizations crosslinguistically will yield a better understanding of both, as well as improve our
understanding of universal clause structure.
Finally, while a number of particles were addressed in this chapter, there are a
few, perhaps five, identified by Reichard that have not been discussed and that may fit
within either the mood, modal, or aspect categories. More data regarding these forms,
164
and those discussed above, will surely add to the understanding of Cr adverbial and
functional heads, as well as expand on the generalizations presented here.
165
CHAPTER 6
CONCLUSIONS
1. Introduction
In the first part of this chapter a brief recapitulation of the previous chapters will be
presented. This will be followed by suggestions for future inquiry.
2. Recapitulation
In Chapter 3, considering aspectual elements, transitivizing elements, the root, and bound
pronouns, as the core elements of the basic clause in Coeur d’Alene, it was claimed that
given an example like (165a), (42a) above, the tree in (165b), (42b) above, could be
derivationally accounted for in (166), (43) above, employing the tenets of the Minimalist
Program.
( 165 ) a.
Cr basic transitive
construction
$
cust- accompany -dt
'We go with you.'
-2acc -1p.erg
166
b.
TP
T
ASPP
ASP
vP
P
P
v
$P
$
( 166 ) a. The root
b.
P
'accompany' merges with its complement the P object
'2acc'.
Neither p-sig is defective, no conflation occurs. The P is labeled with the psig of its head,
.
The P labeled
merges with an element from the numeration, the v head
'directive transitive' which has a defective p-sig and conflation occurs.
The p-sig of P,
, is copied into the defective p-sig of v, giving
.
Here we note that the v head is realized to the right of the phonological
material in P. The assumption here is that there are two types of affixes:
those with defective p-sigs and those without (the bound pronouns for
example). Affixes with defective p-sigs phonologically align with conflated
phonological material in accord with their morphological affix feature
c.
d.
e.
([±prefix]). The vP is labeled with the p-sig of its head,
.
The v labeled
mergers with the subject, a '1p.erg' -head. Neither
p-sig is defective, so no conflation occurs.
The object raises to a second specifier position of v, to check case. Neither
p-sig is deficient. The whole constituent is labeled with the p-sig of the head
v,
.
The vP
merges with an element from numeration, the ASP head
,
'customary'. This element's p-sig is defective, and conflation occurs. The
p-sig of the vP,
, is copied into the defective p-sig of ASP, giving
. Then, the whole constituent, a projection of ASP, is labeled with
167
the p-sig of its head,
.
f.
The ASPP merges with an element from numeration, a null T head. Neither
p-sig is defective, and no conflation occurs. The T projects and is labeled
with the p-sig of its head.
g.
The subject checks case via agree with the T head.
It was demonstrated that such an analysis was preferable to a radical argument
drop analysis (of the Baker 1996 type), on the grounds that such an analysis was more
economical, in the sense of Chomsky (1995:367). In addition, it was shown that a bound
pronoun analysis (Jelinek 1984; Bhat 2004) placed Coeur d’Alene (henceforth Cr) as a
language that fit within Neeleman and Szendröi’s 2005, 2006 typological generalizations
of argument drop languages, whereas the radical argument drop account would place Cr
outside the generalizations of the typology. Further, given the unified system of
pronominals under the bound pronoun analysis, it was demonstrated that the bound
pronoun analysis presented far fewer learnability challenges than the hybrid system of
argument structure of the radical argument drop analysis. An appeal to Occam’s Razor
further strengthened the claim that the unified system of argument structure of the bound
pronoun analysis was preferable to the hybrid system of the radical argument analysis.
It was further shown, that an analysis of bound pronouns as -pronouns fit with
Déchaine and Wiltschko’s 2002 -pronoun proposal. This, and an appeal to the Cue
based model of parameters (Lightfoot 1997a, 1997b; Dresher 1999, 2003a, 2003b), led to
an account of the null third person arguments. Finally, under Newmeyer’s (1998) account
of grammaticalization, an account of Cr bound pronoun diachronic origins in agreement
morphology was presented and rejected. Instead, it was argued that an analysis of bound
pronouns as not having their diachronic origins in independent agreement morphology
provided a better account of cross-linguistic variation within the Salishan family.
168
In the following chapter, Chapter 4, an account of lexical affixes in Cr was
presented along with an account of incorporation in Cr. It was shown that lexical
affixation is not an areal phenomenon as might be assumed, but that in fact it appears in a
variety of languages in a variety of geographical locations across the globe. Appealing to
the Hermit Crab hypothesis of grammaticalization (Heath 1998), it was shown that Cr
lexical affixation might not have its diachronic origins in incorporation as argued.
However, after consideration of the proto-Salish facts regarding lexical affixes and the
incorporation facts of stem+stem incorporation, it was claimed that the traditional view of
lexical affixation in Cr, and Salish in general, as being diachronically rooted in
incorporation was on the right track. The conclusion arrived at in terms of lexical affixes
was that they can serve as arguments as, Gerdts 2003 claimed for Halkomelem, and that
lexical affixes in Cr are n-heads.
This chapter also demonstrated that a view of head-movement as Conflation (Hale
and Keyser 2002), as proposed by Harley (2004), could account for the facts of Cr lexical
affix and stem+stem incorporation while maintaining Chomsky’s 1995 notion of
head-movement being phonological in nature. In addition, it was shown that a conflation
analysis of incorporation was preferable to a Baker-style (1988, 1996; Baker, Aranovich,
and Golluscio 2004) analysis. In particular, it was shown that the conflation analysis
required no special deletion mechanism and the family of parameters necessary to
manage it, along with other incorporation facts, as proposed by Baker, Aranovich, and
Golluscio (2004). Instead, two parameters were proposed to account for incorporation
and the various realizations of verbal morphology that accompany incorporation crosslinguistically. Analyzing incorporation as a narrow syntax operation further demonstrated
the advantages of a Distributed Morphology account over a lexical account such as Rosen
(1989).
169
In Chapter 5 a series of particles listed here in Table 19 were analyzed to
determine the hierarchy of function projections in the Cr clause.
Table 19: Coeur d’Alene particles
TYPE
PARTICLE
Temporal Adverbial
‘soon’ immediate future (Doak 1997:186)
‘soon’ immediate future (Doak 1997:187)
‘soon’ immediate future (Doak 1997:187)
Sentential Adverbial
‘and’ / ‘then’ discourse/narrative adverbial
‘and’ / ‘then’ discourse/narrative adverbial
Mood
irrealis (Reichard 1938:669.777; Doak 1997:188)
Modal
future intentional, permissive, mild request
(Reichard 1938:666-67)
ought, obligation (Reichard 1938:669.780)
was to be but isn’t, possibility (Reichard
1939:104)
Aspectual
‘used to’ terminative (Doak 1997:49)
‘always’ habitual (Doak 1997:49)
The strict ordering of these particles, in relation to one another, revealed a
hierarchy of functional and adverbial heads that was quite similar to that of Cinque’s
(1999) universal hierarchy of functional and adverbial heads. Further, comparison of the
Cr data with Rizzi’s 1997a Split CP hypothesis and Cinque’s universal hierarchy of
functional and adverbial heads demonstrated that the two approaches captured
generalizations with further linguistic coverage. Further, the hierarchy arrived at revealed
that there is no apparent movement of the predicate above ASPP in the Cr clause.
In this way an analysis of the basic clause was arrived at, presented here as (167),
with omission of specifiers for illustrative reasons.
170
( 167 )Basic clause in Cr
MOODP
MOODSPEECH ACT AdvP
ADVSENTENTIAL
AdvP
ADVTEMPORAL
TopP
TOP
MOODP
MOODIRREALIS
TP
T
MODALP
MODALPOSSIBILITY MODALP
MODALABILITY/PERMISSIVE
AdvP
ADVTEMPORAL
ASPP
ASPPhabitual ≥ terminative
ASPP
ASPPcontinuative ≥ customary ≥ completive vP
v
$P
$
Thus, it was demonstrated that while Cr may on the surface appear to be quite different
from a language such as English, underlyingly, it can be argued to be quite similar.
Further, as a formal account of underlying Cr clause structure has not been proposed
previously, and no account of functional projections has been put forth, the structure in
(167) can serve as a starting point for discussion of the formal intricacies of the Cr
grammar. In what follows some suggestion are presented for this line of inquiry.
171
3. Future inquiry
A cursory review of Reichard’s 1938 grammar will reveal that a great deal of verbal
morphology was not discussed in this dissertation. Future research should include
discussion of the numerous morphemes Reichard describes as being bound to the
predicate. This should include discussion of the locative and directional morphemes to
determine if they are case marking elements or somehow related to incorporation facts, or
possibly something else entirely. Another area of interest for future inquiry would be a
formal account of the deictic system in Cr. Baht (2004) demonstrates that third person
pronouns in many of the world’s languages are deictic elements, rather than true
pronouns such as first and second person pronouns. It would be interesting, as the third
person is generally null in Cr, to see how Cr fits within Bhat’s typology.
Further, Doak’s 1997 work serves as an excellent starting point for better
understanding the many structures beyond the basic clause. With data from the Reichard
manuscripts a formal account of adjoined clauses and unadjoined clauses could be arrived
at, along with formal accounts of the benefactive and applicative constructions so
elegantly illustrated in Doak’s work. Further, as Gerdts finishes her decade-long work on
the problematic Salishan –m morpheme in Halkomelem, a potential full account of this
morpheme in Cr could be compared with her findings. In short, there is a great deal of
work to be done, and the Reichard manuscripts along with the various other resources
such as Nicodemus’ 1975 grammatical sketch and dictionaries, along with the work Doak
is preparing for the online dictionary, will provide an excellent corpus for investigation.
This work can shed further light on numerous issues within numerous languages
in terms of theoretical inquiry. One area of particular interest is the area of computational
linguistics, as well as the Principles and Parameters approach to linguistic inquiry. In
terms of computational investigations, once the facts of Cr morphotactics are organized
172
in a notation suitable for input into Hulden’s morphological parser, a number of questions
regarding compositionality can be raised and answered. In terms of the Principles and
Parameters approach to linguistic inquiry the Cr data has already shown that elements of
Baker’s (1996) Polysynthesis Parameter, namely incorporation, appear in Cr. In addition
elements of Jelinek’s (2004, 2006) Pronominal Argument Parameter have been seen in
Cr, namely a ban on DPs in argument position. This suggests that languages such as Cr
should be viewed not in terms of macro-parameters, but rather perhaps in terms of microparameters.
173
APPENDIX
Some notes on Coeur d'Alene Phonology
1. Introduction
What follows is a brief description of the phonological inventory of Coeur d'Alene and
some often-occurring phonological phenomena. This description owes much to Doak
1997.
2. Consonants
The Coeur d’Alene inventory comprises forty-two consonants that contrast eleven places
of articulation: labial, alveolar, alveopalatal, lateral, labiovelar, uvular, labio-uvular,
coronal pharyngeal, pharyngeal, labiopharyngeal, and laryngeal. There are six manners of
articulation for the consonants: plain and glottalized voiceless stops and affricates; voiced
stops and affricates; voiceless fricatives; and plain and laryngealized resonants. Coeur
d’Alene lacks the glottalized lateral affricate "U " that occurs in the other Salishan
languages; in Coeur d’Alene this has merged with / T/ (Thompson 1979:706). The Coeur
d’Alene consonants are presented in Figure 1.
T
?
J
V
,
,
,
,T
4
4
*
<
!
7
7
<
!
7
7T
Figure 1 Coeur d’Alene Consonant Inventory
174
3. Vowels
Salishan languages vary in the number of vowels they have from three to five. Coeur
d’Alene has a five-vowel system distinguishing two levels of height87 and two degrees of
backness / / and / / are prototypical high vowels. / / appears to exhibit the greatest
phonetic range: from A;B to AW B5 /X/ and /./ are low back vowels that include elements of
pharyngeal constriction. Schwa is never stressed and occurs as a reduced form of some
unstressed vowels. Schwa also occurs as an inserted, or excrescent, element to break up
consonant clusters. When schwa occurs as an excrescent element its use often varies from
speaker to speaker (Doak 1997:12).
.
X
Figure 2 Coeur d’Alene Vowels
Inventory
4. Syllables
There are four primary core syllables in Coeur d’Alene: V, VC, CV, and CVC:
(1)
87
a.
b.
+
.,
(
‘you are wet’
(V CV CVC)
‘Easter egg’ (VC CVC CV CV-CVC)
(Doak 1997:13)
Doak 1997:12 distinguishes /i/ and /u/ as [+high] vowels and the remaining vowels [-high].
175
Doak notes that the V and CV primary core syllables occur in word initial position in a
few proclitics or prefixes, and that some syllable initial u’s may be analyzable as glides.
The sonority of segments decreases with distance from the nucleus when
consonants are added to the onset of a CV(C) syllable or the coda of a (C)VC syllable.
With S representing segments that are of higher sonority than C, three additional syllables
are: CSV(C), (C)VSC, and CVSSC.
(2)
'wrist' (C-CVC-CC-CVSSC)
Y
(Doak 1997:13)
The resonant m surrounded by consonants, in the previous example, may serve as
a syllable peak. Other syllabic resonant (R1) can be seen in the following forms.
(3)
a.
b.
'I got out of breath'
V
'I got stung'
(C R1 R1-CVC-C88)
(CVC- R1-CVC- R1) (Doak 1997:13)
5. Morphophonolgy
5.1. Vowel lowering
In certain environments Coeur d’Alene vowels / /, / /, and / / lower to [X], [ ], and [.],
respectively. Doak refers to this phenomenon as “vowel harmony.” There are two types
of vowel lowering in Coeur d'Alene, long distance and allophonic. These can be broken
down into two sub-types, regressive and progressive. The following are examples of the
long distance regressive and progressive types, where [X], [ ], and [.] occur preceding
uvulars or pharyngeals as the result of regressive harmony.
(4)
a.
b.
88
, + ,[ . ,
D appears to analyze the final
Z$,
Z $
%, Z
%. , Z
2hat'
2fur coat'
as extra syllabic material in this example.
176
c.
\ .,
Z$
%. ,
Z
2He is tall.'
Doak 1997:30)
Progressive harmony affects stressed suffixal / /, / /, and / / following harmony
roots. Some roots trigger the lowering of suffixal vowels as in the following.
(5)
a.
[
b.
.
c.
9#= #%
Z$ .
\
.
Z$ .
. +.
Z$
$
$ .
Z
2He shot (people).'
Z
. %
2snail'
Z
2Water is full of masmas.'89
(Doak 1997:31)
5.2. /h/ loss
Morpheme initial /h/ is frequently omitted in word initial position before a consonant, or
in a compound:
(6)
a5
b.
c.
'
4.
-
""
""
\
//
4.
//
9 \ //
2from where' (Doak 1997:34)
""
2I love you.' (Doak 1997:34)
'(to) circle' (Reichard 1927-29:L010)
5.3. Coronal sequence reduction
When two coronals occur together, the first is usually dropped. This tends to vary
somewhat and generally applies only to affixes, which tend to be less stable than roots.
Some suffix sequences involving / / and / /, however, are preserved: for example, the
transitive sequence // ) // merges to [ ], and does not reduce to [ ].
(7
$
loc- receive
'He received it.'
)
-d
-t
-3abs -3erg
(Reichard 1927-29:L025)
89
Masmas is a type of vegetable (Doak 1997:31).
177
5.4. / / palatalization
The sequence / - / (where the hyphen marks a morpheme boundary), results in [ ’] :
(8)
Z $
Z
2eating'
(Doak 1997:33)
The reader is directed to Reichard (1938) and Doak (1997; and references therein) for
further discussion of phonological phenomena in Coeur d'Alene.
178
REFERENCES
Ackeman, Peter and Ad Neeleman. 2006. Restricted pro drop in Early Modern Dutch.
manuscript U of Edinburgh. Accessed online 8 August 2006 at
http://www.ling.ed.ac.uk/~packema/publications%20list.htm.
Alexiadou, Artemis and Elena Anagnostopoulou. 1998. Parametrizing AGR: Word order,
V-movement and EPP-checking. Natural Language and Linguistic Theory 16:
491-539.
Allen, Barbara, Donna Gardiner and Donald Frantz. 1984. Noun incorporation in
Southern Tiwa. International Journal of American Linguistics 50:292–311.
Baker, Mark C. 1988. Incorporation: A theory of grammatical function changing.
Chicago: University of Chicago Press.
______. 1996. The Polysynthesis Parameter. New York: Oxford University Press.
______. 2001. The natures of nonconfigurationality. In Handbook of contemporary
syntactic theory. ed Mark Baltin and Chris Collins. Oxford: Blackwell
______. 2002. On zero agreement and polysynthesis. Ms, Rutgers University. Accessed
online 1 June 2006 at http://www.rci.rutgers.edu/~mabaker/papers.html.
Baker, Mark C., Robert Aranovich and Lucía A. Golluscio. 2004. Two types of syntactic
noun incorporation: Noun incorporation in Mapundungun and its typological
implications.
Beninc , Paola. 2001. The position of topic and focus in the left periphery. Current
studies in Italian syntax, ed. Guglielmo Cinque and Giampaolo Slavi 39-64.
Amsterdam: Elsevier.
Bhat, D. N. S. 2004. Pronouns. Oxford: Oxford University Press.
Binnick, R. 1991. Time and the verb. New York: Oxford University Press.
Boas, Franz, and James Teit 1985. Coeur d'Alene, Flathead and Okanogan Indians. Ye
Galleon Press: Fairfield, Washington.90
Butt, Miriam. 2001. Case, agreement, pronoun incorporation and pro-drop in South Asian
languages. Talk held at the workshop on The Role of Agreement in Argument
90
This is a reproduction of the original reprint from the forty-fifth annual report of the Bureau of American
Ethnology, 1927-1928.
179
Structure, Utrecht, August 2001. Accessed on line 6 August 06, at http://ling.unikonstanz.de/pages/home/butt/utrecht01-hnd.pdf
Bybee, Joan L. 1985. Morphology: A study of the relation between meaning and form.
Amsterdam: John Benjamins.
Bybee, Joan L., William Pagliuca and Rever D. Perkins. 1990. On asymmetries in the
affixation of grammatical material. In Studies in typology and diachrony: Papers
presented to Joseph H. Greenberg on his 75th birthday, ed. by William Croft,
Keith Dennings, and Suzanned Kemmer, pp. 1-42. Amsterdam: John Benjamins.
______. 1994. The evolution of grammar: Tense, aspect, and modality in the languages
of the World. Chicago: U of Chicago Press.
Carlson, Barry F. 1973. A grammar of Spokan: A Salish language of Eastern
Washington. University of Hawaii Doctoral Dissertation.
______. 1990. Compounding and lexical affixation in Spokane. Anthropological
Linguistics 32.1-2:69-82.
Chomsky, Noam. 1982. Some concepts and the consequences of the theory of
Government and Binding. Cambridge: MIT Press.
______. 1991. Some notes on economy of representation and derivation. MITWPL #10:
Functional heads and clause structure. ed. by I. Laka and A. Mahijan . MIT
Press: Cambridge.
______. 1993. A minimalist program for linguistic theory. In The view from Building 20,
Kenneth Hale and Samuel J. Keyser (eds), 1-52. Cambridge: MIT Press.
______. 1994. Bare phrase structure. MIT Occasional Papers in Linguistics 5: 1-51.
______. 1995. The minimalist program for linguistic inquiry. Cambridge: MIT Press.
______. 1999. Derivation by Phase. MIT occasional papers in linguistics 18.
______. 2000. Minimalist inquiries: the framework. In Step by step: Essays in honor of
Howard Lasnik. ed by Roger Martin, David Michaels, and Juan Uriagereka, 89155. Cambridge: MIT Press.
______. 2001a. Derivation by phase. In Ken Hale: A life in language, M. Kenstowicz
(ed), 1-52. Cambridge: MIT Press.
______. 2001b. Beyond explanatory adequacy. Occasional Papers in Linguistics 20.
180
Comrie, Bernard. 1976. Aspect. London: Cambridge University Press.
Cinque, Guglielmo. 1999. Adverbs and functional heads. Oxford University Press: New
York.
Davis, Henry. 2000. Remarks on Proto-Salish subject inflection. International Journal of
American Linguistics 66.4:499-526.
Davis, Henry and Lisa Matthewson. 2003. Quasi objects in St'át'imcets. In Formal
approaches to function in grammar, ed. by Andrew Carnie, Mary Willie and
Heidi Harley, 79-106. Amsterdam: John Benjamins.
Déchaine, Rose-Marie and Martina Wiltschko. 2002. Decomposing pronouns. Linguistic
Inquiry 33.3:409-422.
Doak, Ivy G. 1992. Another look at Coeur d'Alene Harmony. International Journal of
American Linguistics. 58.1:1-35
______. 1997. Coeur d’Alene Grammatical Relations. Ph.D. dissertation, University of
Texas at Austin.
, and cognate
Doak, Ivy, and Anthony Mattina. 1997. Okanagan , Coeur d'Alene
forms. International Journal of American Linguistics 63.3:334-61.
Dresher, B. Elan Dresher. 1999. Charting the learning path: Cues to parameter setting.
Linguistic Inquiry 30:1 27-67.
______. 2003a. Meno’s paradox and the acquisition of grammar. In Living on the edge:
28 papers in honor of Johnathan Kayne (Studies in Generative Grammar 62) ed
Stefan Ploch. Berlin: Mouton De Gruyter. 7-27. Accessed online 10 July 2006 at
http://www.chass.utoronto.ca/~dresher/publications.html.
______. 2003b. Contrast and asymmetries in inventories. In Asymmetry in Grammar,
Volume 2: Morphology, Phonology, Acquisition. ed Anna-Maria di Sciullo.
Amsterdam: John Benjamins. 239-257.
Embick, David and Rolf Noyer. 2005. Distributed Morphology and the
Syntactic/Morphology interface. Manuscript U of Pennsylvania.
Fleck, David. 2006. Body-part prefixes in Matses: Derivation or noun incorporation?
International Journal of American Linguistics. 72.1: 59-96.
Galloway, Brent. 1993. A grammar of Upriver Halkomelem. University of California
Press: Berkeley.
181
Gerdts, Donna. 1998. Incorporation. In The handbook of morphology, ed. Andrew
Spencer and Arnold M. Zwicky, 84-100. Oxford: Blackwell Publishers.
______. 2003. The morphosyntax of Halkomelem lexical suffixes. International Journal
of American Linguistics 69.4:345-56.
Haegeman, Liliane and Tabea Ihsane 2001. Adult null subjects in the non-pro-drop
languages: Two diary dialects. Language Acquisition 9:4 329-346.
Hagège, Claude. 1978. Lexical suffixes and incorporation in Mainland Comox. Forum
Linguisticum 3.5:7-71.
______. 1980. On incorporation in universal grammar (further comments on a previous
article). Forum Linguisticum 4.3:241-45.
Hale, Ken. 2000. Remarks on the syntax of the Navajo verb. In Working papers on
endangered and less familiar languages 3: Diné bizaad naalkaah: Navajo
language investigations, ed. Fernald and Hale 55-96. Boston: MIT Press
Hale, Ken and Samuel J. Keyser. 2002. Prolegomenon to a theory of argument structure.
Cambridge, Mass.: MIT Press.
Halle, Morris, and Alec Marantz. 1993. Distributed Morphology and the pieces of
inflection. In The view from Building 20: Essays in linguistics in honor of Sylvain
Bromberger, ed. Kenneth Hale and Samuel Jay Keyser, 111–176. Cambridge,
Mass.: MIT Press.
Harley, Heidi. 2004. Merge, conflation, and head movement: The First Sister Principle
revisited. In Proceedings of NELS 34, (eds) Keir Moulton and Matthew Wolf.
UMASS Amherst:GLSA. Accessed on 9 November at
http://dingo.sbs.arizona.edu/~hharley/PDFs/HarleyNELS2003Paper.pdf
Harley, Heidi and Rolf Noyer. 1999. State-of-the-Article: Distributed Morphology. Glot
International 4.4, 3-9.
Harley, Heidi and Mercedes Tubino-Blanco and Jason Haugen. 2006. Applicative
constructions and suppletive verbs in Hiaki. manuscript. University of Arizona.
Haspelmath, Martin, Matthew S. Dryer, David Gil and Bernard Comrie (eds) 2005. The
world atlas of language structures. Oxford: Oxford University Press.
Haugen, Jason. 2004. Issues in comparative Uto-Aztecan morphosyntax. PhD
dissertation, the University of Arizona.
182
______. 2006. Denominal verbs and noun incorporation: Uto-Aztecan evidence for a
unified syntactic account. manuscript University of Arizona.
Heath, Jeffrey. 1998. Hermit crabs: Formal renewal of morphology by phonologically
mediated affix substitution. Language 74.4:728-759.
Heine, Bernd. 1990. The dative in Ik and Kanuri. in Studies in Typology and Diachrony:
Papers Presented to Joseph Greenberg on His 75th birthday, ed by William
Croft, Keith Denning, and Suzanne Kemmer, pp. 129-149. Amsterdam: John
Benjamins.
______. 1993. Auxiliaries: Cognitive Forces and Grammaticalization. New York:
Oxford University Press.
Heine, Bernd, Ulrike Claudi, and Friederike Hünnemeyer 1991. Grammaticalization: A
Conceptual Framework. Chicago: University of Chicago Press.
Heine, Bernd and Tania Kuteva. 2002. World lexicon of grammaticalization. Cambridge
University Press:New York.
Hill, Jane. 2003. Subject number agreement, grammaticalization, and transitivity in
Cupe o verb construction. ed. by Andrew Carnie, Mary Willie and Heidi Harley,
207-226. Amsterdam: John Benjamins.
Hoji, Hjime.1998. Null object and sloppy identity in Japanese. Linguistic Inquiry 29:1
127-152.
Holmes, Philip and Ian Hinchliffe. 2003. Swedish: A comprehensive grammar. 2nd
edition. London: Routledge.
Holmberg, Anders. 2005. Is there little pro? Evidence from Finnish. Linguistic Inquiry
36:4 533-564.
Holmberg, Anders and Urpo Nikanne. 2002. Expletives, subjects, and topics in Finnish.
In Subjects expletives and the EPP. ed Peter Svenonius. Oxford: Oxford
University Press 71-106. Accessed online 5 March 2006 at
http://www.ncl.ac.uk/elll/research/publications?staff=31637
Hornstein, Norbert, Jairo Nunes, and Kleanthes K. Grohmann. 2005. Understanding
Minimalism. Cambridge: Cambridge University Press.
Huang, C.-T. James. 1984. On the distribution and reference of empty pronouns.
Linguistic Inquiry 15:4 531-574
183
Jaeggli, Osvaldo and Ken Safir. 1989. The null subject parameter and parametric theory.
In The null subject parameter. ed. Osvaldo Jaeggli and Ken Safir. Kluwer:
Dordrecht.
Janda, Richard D. 1980. On the decline of declensional systems: the overall loss of OE
nominal case inflections and the ME reanalysis of -es as his. Papers from the 4th
International Conference on Historical Linguistics, ed. by Elizabeth C. Traugott,
Rebecca Labrum, and Susan Shapherd, pp. 243-252. Amsterdam: John
Benjamins.
_____. 1981. A case of liberation from morphology into syntax: the fate of the English
genitive-marker –(e)s. In Syntactic change, ed. By Brenda D. Johns and David R.
Strong. University of Michigan: Ann Arbor. 59-114
Jelinek, Eloise. 1984. Empty categories, case, and configurationally. Natural Language
and Linguistic Theory 2:39-76.
_____. 2004. The Pronominal Argument parameter. manuscript, University of Arizona.
_____. 2006. The Pronominal Argument parameter. Agreement and Argument structure.
ed by P. Ackeman. Oxford University Press: Oxford.
Jelinek, Eloise, and Richard A. Demers. 1994. Predicate and pronominal arguments in
Straits Salish. Language. 70.4:697-736.
Kinkade, M. Dale. 1963. Phonology and morphology of Upper Chehalis: II. International
Journal of Linguists 29.4
_____. 1998. Origins of Salishan lexical suffixes. Papers for the 33rd International
Conference on Salish and Neighboring Languages, ed Dawn Bates pp. 266-95.
Kiss, Katalin. 1995 Introduction. In Discourse configurational languages, (ed). by
Katalin Kiss. Oxford University Press: Oxford.
Koster, Jan. 1997. Anaphora and the uniformity of grammar. In Atomism and binding, ed.
by Hans Bennis, Pierre Pica, and Johan Rooryck, 235-250. Foris: Dordrecht.
Kroeber, Paul. 1999. The Salish language family: Reconstructing syntax. U of Nebraska
Press: Lincoln.
Kroch, Anthony, John Myhill, and Susan Pintzuk. 1982. Understanding do. Chicago
Linguistic Society 18, 282-294.
184
Kuipers, Aert. 1974. The Shuswap language: Grammar, texts, dictionary. Mouton: The
Hague.
_____. 2002. Salish etymological dictionary. University of Montana Occasional Papers in
Linguistics 16.
Lasnik, Howard. 1999a. Minimalist analysis. Oxford: Blackwell
_____. 1999b. On feature strength: three minimalist approaches to overt movement.
Linguistic Inquiry 30:1 197-218.
_____. 2000. Syntactic Structures Revisited: Contemporary lectures on classic
transformational theory. Cambridge: MIT Press.
Lai, I.-J. Sabdra. 1998. The grammar and acquisition of Secwepemctsín. Masters thesis,
University of British Columbia, Vancouver.
Lightfoot, Douglas J. 1997a. Catastrophic change and learning theory. Lingua 100. 171192.
______. 1997b. Shifting triggers and diachronic reanalysis. In Parameters of
morphosyntactic change. ed Ans van Kemenade and Nigel Vincent. Cambridge:
Cambridge University Press. 253-272.
______. 2005. Can the lexicalization/grammaticalization distinction be reconciled?
Studies in Language 29.3:583-615.
Longobardi, Giuseppe. 1994. Reference and proper names: A theory of N-movement in
syntax and Logical Form. Linguistic Inquiry 25:609–665.
Mattina, Anthony. 2001. Okanagan sentence types: A preliminary working paper.
In UBCWPL 6: The thirty-sixth international conference on Salish and
neighboring languages 215-236, eds. Leora Bar-el, Linda Tamburri Watt and Ian
Wilson.
______. 1973. Colville grammatical structure. Ph.D. dissertation. Distributed
by the University of Hawaii
Mattina, Nancy. 1996. Aspect and category in Okanagan word formation. Ph.D.
dissertation, Simon Fraser University.
Mithun, Marianne. 1984. The evolution of noun incorporation. Language 60.4:847-894.
185
______. 1997. Lexical affixes and morphological typology. In Joan Bybee, John Haiman,
and Sandra A. Thompson (eds) Essays on language function and language type:
Dedicated to T. Givón. Amsterdam: John Benjamins. 357-371.
Neeleman, Ad and Kriszta Szendröi. 2005. Radical pro drop and the morphology of
pronouns. manuscript Utrecht, accessed at
http://www.phon.ucl.ac.uk/home/kriszta/#Papers 5 July 2006.
______. 2006. Radical pro drop and the morphology of pronouns. manuscript Utrecht,
accessed at http://www.phon.ucl.ac.uk/home/ad/2.html 29 July 2006.
Newmeyer, Frederick J. 1998. Language form and language function. MIT Press:
Cambridge.
Nicodemus, Lawrence G. 1975. Snchitsu’umshtsn: The Coeur d’Alene Language Vol I &
II. Spokane: University Press.
O’Donnell, Megan. (2004). Kaska (Northern Athabaskan) Pronominal Affixes and
Clitics. In Coyote Papers XIII: University of Arizona working papers dedicated to
the Indigenous Languages of the Americas. (Eds) Tania Granadillo, Megan
O’Donnell, and Stacey Oberly. Tucson, AZ. 41-74.
Oku, Satoshi (1998a) A theory of Selection and Reconstruction in the Minimalist
Perspective, PhD dissertation. University of Connecticut.
Palmer, F. R. 2001. Mood and Modality. (2nd edition) Cambridge University Press:
Cambridge.
Panagiotidis, Phoevos. 2003. Empty nouns. Natural Language and Linguistic Theory
21:2. 381-432.
Pensalfini, Rob. 2004. Towards a typology of configurationality. Natural Language &
Linguistic Theory 22:2 359-408.
Pylkkanen, Liina. 2002. Introducing Arguments. PhD. Dissertation, Massachusetts
Institute of Technology.
Raposo, E. and J Uriagereka. 1990. Long-distance case assignment. Linguistic Inquiry
21 505-538.
Reichard, Gladys A. 1939. Stem-list of the Coeur d’Alene Language. International
Journal of American Linguistics 10:92-108.
186
______. 1938. Coeur d’Alene. Handbook of American Indian Languages 3:515-707, ed.
Franz Boas. New York: J. J. August.
______. 1927-1929. Unpublished field notes and manuscripts.
Rizzi, Luigi. 1997a. The fine structure of the left periphery. Elements of grammar. ed. by
Liliane Haegeman, 281-337. Kluwer: Netherlands.
______. 1997b. A parametric approach to comparative syntax: properties of the
pronominal system. In The new comparative syntax. ed Liliane Haegeman. New
York: Longmans 268-285.
______. 1986. Null objects in Italian and the theory of pro. Linguistic Inquiry 17:501517.
Roberts, Ian. 1997. Comparative syntax. New York: Arnold
Rohrbacher, Bernard. 1994. The German languages and the Full Paradigm: A theory of
V to I raising. PhD dissertation university of Massachusetts at Amherst.
Rosen, Sara. 1989. Two types of noun incorporation: A lexical analysis. Language 65
294-317.
Saito, Mamoru. 2003. Ellipsis and pronominal reference in Japanese clefts. In NELS 33:
323-345. Accessed online 3 July 2006 at
http://www.people.fas.harvard.edu/~ctjhuang/Course_LSA222/readings/Saito.200
4.cleft.pdf.
Sapir, Edward. 1911. The problem of noun incorporation in American languages.
American Anthropologist 13:250-82
Satterfield, Teresa. 2003. Economy of interpretation: Patterns of pronoun selection in
transitional bilinguals. In Effects of the second language on the first. Ed Vivian
Cook. Clevedon: Multilingual Matters. 214-233. Accessed online 3 August 2006
at http://www-personal.umich.edu/~tsatter/.
Sato, Yosuke. 2006. Ellipsis and LF recycling: An East Asian perspective. manuscript
University of Arizona.
Sato, Yosuke, and Jason Ginsburg. Forthcoming. A new type of nominal ellipsis in
Japanese. MITWPL. MIT: MIT Press.
Saunders, Ross and Philip W. Davis. 1975. Bella Coola referential suffixes. International
Journal of American Linguistics. 41.4:355-368.
187
Silva-Villar, Luis. 1998. Subject positions and the roles of CP. In Romance linguistics:
Theoretical perspectives, ed Armin Schwegler, Bernard Tranel, and Myriam
Uribe-Etxebarria, 247-270. Philadelphia: John Benjamins.
Sloat, Clarence. 1966. Phonological redundancy rules in Coeur d'Alene. Ph.D.
dissertation University of Washington.
______. 1968. A skeleton key to Reichard's Coeur d'Alene transcriptions.
Anthropological Linguistics 10:5.
______. 1972. Vowel harmony in Coeur d'Alene. International Journal of American
Linguistics. 38.2:34-39 check
Smeets, Ineke. 1989. A Mapuche grammar. Leiden: University of Leiden dissertation.
Speas, Margaret. 1993. Null arguments in a theory of economy and projection. UMOP
17, GLSA:Amherst.
______. 2004. Economy and the representation of Null Arguments. Agreement and
Argument structure. ed by P. Ackeman. Oxford University Press: Oxford.
Accessed online at http://people.umass.edu/pspeas/pubs.html 8 October 2005 at
http://people.umass.edu/pspeas/pubs.html.
Tomioka, Satoshi. 2003. The semantics of Japanese null pronouns and its cross-linguistic
implications. In The interfaces: Deriving and interpreting omitted structures, ed
K. Schwabe and S. Winkler. Amsterdam: John Benjamins 321-339.
Vainikka, Anne and Yonata Levy. 1999. Empty subjects in Finnish and Hebrew. Natural
Language and Linguistic Theory 17:613-671.
Vogt, Hans. 1940. The Kalispel language. An outline of the grammar with texts,
translations, and dictionary. Oslo: Det Norske Videnskaps-Akademi i Oslo.
Watanabe, Akira. 2004. Topic-focus Articulation in Old Japanese. manuscript, University
of Tokyo.
Williams, Edwin S. 1997. Blocking and anaphora. Linguistic Inquiry 28.4:577-628.
Wiltschko, Martina. 2006. The structural deficiency of (certain) incorporated nouns:
Implications for form and meaning. Paper presented at Noun Incorporation and its
Kind, University of Ottawa.
Wunderlich, Dieter. 1996. Constructing inflectional paradigms. Paper presented at the
Workshop on Inflection, Vienna, 15 February.
188
Zushi, Mihoko. 2003. Null arguments: The case of Japanese and Romance. Lingua 113:
559-604.
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement