***LATEST NEWS***
April 11, 2008
I successfully defended my dissertation thesis
titled The Acquisition of Verbal Inflectional Morphology
in Child Grammar in a Variability Model of Early
Morphosyntactic Development: A Biolinguistic
Perspective.
The final version in a bound book form should be
available within a few weeks or so.
*******************************************************
I'm writing a doctoral dissertation on the topic of verbal
inflectional morphology and morphosyntax in child grammars,
particularly child English, Slovenian, and Spanish,
trying to combine the tools and evidence from formal
(generative) linguistic theory, cognitive neuroscience,
cognitive psychology, and computational language
modeling in the explanation of the omission of early
tense/agreement marking. My working title is
"The Acquisition of Verbal Inflectional Morphology
in Child Grammar in a Variability Model of Early Morphosyntactic Development".
In my work, I argue that even a little fraction of
"grammar" such as tense and agreement requires
a substantial cooperation not just among linguists
("interfaces"), but also crossing the boundaries of
linguistics, teaming up with cognitive psychology
and neuroscience. Hence, most of my argumentation
will be reminiscent of investigations into language
from a "biolinguistic" perspective, voiced some five
decades ago among scholars such as Noam Chomsky
and George Miller, who advocated that linguistics
be part of a more interdisciplinary endeavor that
has become to be known as "the cognitive sciences".
April 11, 2008
I successfully defended my dissertation thesis
titled The Acquisition of Verbal Inflectional Morphology
in Child Grammar in a Variability Model of Early
Morphosyntactic Development: A Biolinguistic
Perspective.
The final version in a bound book form should be
available within a few weeks or so.
*******************************************************
I'm writing a doctoral dissertation on the topic of verbal
inflectional morphology and morphosyntax in child grammars,
particularly child English, Slovenian, and Spanish,
trying to combine the tools and evidence from formal
(generative) linguistic theory, cognitive neuroscience,
cognitive psychology, and computational language
modeling in the explanation of the omission of early
tense/agreement marking. My working title is
"The Acquisition of Verbal Inflectional Morphology
in Child Grammar in a Variability Model of Early Morphosyntactic Development".
In my work, I argue that even a little fraction of
"grammar" such as tense and agreement requires
a substantial cooperation not just among linguists
("interfaces"), but also crossing the boundaries of
linguistics, teaming up with cognitive psychology
and neuroscience. Hence, most of my argumentation
will be reminiscent of investigations into language
from a "biolinguistic" perspective, voiced some five
decades ago among scholars such as Noam Chomsky
and George Miller, who advocated that linguistics
be part of a more interdisciplinary endeavor that
has become to be known as "the cognitive sciences".
Dominik Rus:
Dissertation work
Dissertation work
In order to achieve that, however, I have to submit only to certain types of evidence and reject
many others. Specifically, I argue against many (what I believe are entirely) "ad-hoc" (and
descriptive) linguistic principles and operations with respect to early grammars (e.g., the
Tense Omission operation, the Uniqueness Checking Constraint, the Minimize Violations
Constraint, etc.).
I embrace an innate property of Language as the most minimal one, i.e., referring solely to a
widely-assumed architecture with hierarchy, binarity, locality, and Merge (the concatenation
operation), linking it to what Hauser, Chomsky, and Fitch (2002) call "the faculty of Language
in a narrow sense" (FLN). I show that most of early tense and agreement learning is sensitive
to (and hence procedes at) the interfaces, particularly as morphology-syntax mapping rather
than syntax per se ("biology" as has been often argued by nativists). Hence, I argue, that kids
do not omit tense/agreement because of "biology" since "biology" in my sense is present from
the start (the "syntax-before-morphology" hypothesis, voiced by quite a few researchers).
On the other hand, I also reject many claims coming from "general cognitive" explanations of
early language learning (e.g., "constructivist") and take only those pieces of evidence that I
believe have been luminary for the properties in question (particularly the effects of saliency
and frequency as well as the idea of "gradual learning" rather than instantaneous
convergence to adult grammar).
I additionally present evidence on language processing and the generation (production) of
functional categories in the neurolinguistic tradition and the evidence on language modeling,
weaving both of them together with developmental linguistic facts into an explanatory model
of early tense and agreement marking (see below for details).
My dissertation committee consists of:
Dr. Donna Lardiere (Georgetown; Ph.D. Applied Linguistics, Boston University), co-director
Dr. Charles Yan (Univ. of Pennsylvania; Ph.D. Computer Science, MIT), co-director
Dr. Héctor Campos (Georgetown; Ph.D. Romance Linguistics, UCLA), committee member
Dr. Michael Ullman (Georgetown; Ph.D. Brain & Cognitive Sciences, MIT), committee member
The following problem statement is the official (short) Hypothesis/Problem Statement that I
have filed with Georgetown University (see below for the longer version):
While acquiring early verb morphosyntax, children are amazing learners yet prone to errors.
This study explores how and why this is the case. By viewing a language acquisition process
as a combination of deterministic and probabilistic mechanisms, the study addresses early
tense and agreement marking in child Slovenian, English, and Spanish by using the tools and
empirical findings from a generative linguistic theory, formal theories of early morphosyntax,
and developmental psycholinguistic accounts of early verb learning, supplemented with the
evidence on computational modeling of the learning of verb morphosyntax and neurolinguistic
evidence on the processing and generation of functional categories.
Please note that some details of my hypothesis have been constantly changing since my first
thoughts on the topic.
Below you can check out my Orals Exam Problem Statement (with full bibliography), followed
by the official Dissertation Proposal, followed by the most current abstract.
Any comments and questions are welcome (dr59 [at] georgetown [dot] edu). This is still VERY
MUCH work in progress.
My orals exam problem statement with complete suggested bibliography:
Reconciling Mother Nature and Momma Nurture: Towards a Variability Model of the
Development of Tense and Agreement in Child Grammars
[submitted on April 13 2006 and defended with High Pass on May 17 2006]
My dissertation proposal (long version):
Reconciling Mother Nature and Momma Nurture: The Acquisition of Tense and Agreement in
Child Grammar in a Variability Model of Early Morphosyntactic Development
[submitted on November 14 2006]
My most recent "dissertation abstract":
The Acquisition of Tense and Agreement in early morphosyntax: A biolinguistic perspective
[written in January 2007, and slightly updated in February 2007 for a Biolinguistic
Investigations conference in Santo Domingo, Dominican Republic]
My most recent paper, basically a very condensed version of the dissertation:
The Learning at the Interfaces: Tense and Agreement in a Variability Model. A View from a
Biolinguistic Perspective
[in preparation; to be published in the Georgetown University Working Papers in Theoretical
Linguistics Vol 7, 2007]
Other Projects
(A) [2007- ] The Acquisition of LOCATIVE MARKING in 2-3 year-olds
(with Ashley Fidler)
KEY WORDS: locative marking, case marking, adpositions, linguistic complexity,
cognitive (relational) complexity, perceptual biases, goal/source/state path
distinctions, child English, child Hungarian, child Slovenian
Our Research Statement
Ashley and I have recently started investigating the acquisition of locative markers where it
has been claimed that there appears to be an interaction between linguistic and conceptual
development.
Conceptual factors, such as the early perceptual bias towards GOAL over SOURCE paths are
important predictors of the accuracy of early locative use. Linguistically, moreover, the
differential complexity of locative expressions may influence their development.
Agglutinating case markers and adpositions have been argued to be relatively less complex,
which should lead to more accurate early production. Complex locative markers which
combine adpositions and lexically-selected syncretic case markers, on the other hand, should
be used less frequently in early child language. Thus, research into locative marking must
cross the boundary between linguistic and cognitive development in order to tease apart the
influence of these and other potentially contributing factors.
Assuming this line of reasoning, linguistic complexity should have a noticeable impact on
locative marking in young children cross-linguistically. Very little research, however, has
investigated this hypothesis. Using naturalistic production data from child Slovenian,
Hungarian, and English, then, we are seeking to address the role of linguistic complexity in
relation to conceptual complexity in the locative marking of 2-year-olds. These languages are
good candidates for such an investigation in that they have similar locative semantics, but
diverse linguistic means of expressing these relationships. Slovenian uses complex
morphosyntactic marking which combines prepositions and lexically selected case markers
with a lot of syncretism, Hungarian marks locatives using either agglutinating case markers
with no morphological agreement or postpositions, and English makes use of prepositions of
varying complexity.
We are investigating the following relation types in three corpora: IN, ON, AT, UNDER,
BETWEEN, NEXT TO, IN FRONT OF, and BEHIND. Where they are reliably distinguished, path
distinctions (GOAL vs. STATE vs. SOURCE) are also being assessed. Specifically, we are
looking at how linguistic complexity independently contributes to productive locative marking
in 2-year-olds, particularly the production of Hungarian postpositions, which seem to develop
pretty rapidly and with few errors (though gradually!) and Slovenian preposition + case
amalgams, which seem to be very inaccurate in early development with preposition being
frequently omitted (though nominals almost always carry correct case-marking!).
(B) [2004-2006] The Acquisition of Imperatives in 1-2 year-olds
(some parts of the project with Pritha Chandra)
KEY WORDS: imperatives, acquisition of imperatives, acquisition of finiteness, early
root nonfinites, root infinitives, root infinitive-analogues, very early child Slovenian
Summary and Results
After my theoretical study on the morphosyntax of Slovenian imperatives, Pritha and I carried
out a couple of studies on the acquisition of imperative clause types in children who are just
beginning to speak. We investigated the morphological, syntactic, and semantic properties of
imperatives, with a special emphasis of relating them to the study of finiteness and early root
nonfinites.
Studies on many child languages have shown that most children start off with a root
infinitive (RI) stage, i.e., using an infinitive form where adult contexts use a finite (tense- and
agreement-marked forms). It has been shown that languages such as Catalan, Spanish,
Italian and Slovenian (generally, null subject languages with rich morphology) are exceptions.
This, of course, raises a question of why some languages lack an RI stage, and, more
broadly, if there is a universal RI/RI-analogue stage in early grammars.
Salustri & Hyams (2003) [henceforth S&H] argue that in early Italian, it is the imperative that
represents an RI analogue. S&H state that null subject Non-RI child languages exhibit an
early imperative stage when the frequency of imperatives is extremely high when compared
to other verb forms, as well as to the frequency of imperatives in the target language. Child
imperatives in these languages lack TP/AgrP projections and have the same semantic
interpretation of RIs (in terms of event semantics as well as the realis/irrealis semantic
opposition). Hence, the “imperative as an RI-analogue” hypothesis (IRIAH) provides a
“universal” explanation of children’s syntax and semantics of early root nonfinites and their
analogues.
In our studies, we investigated root nonfinites and imperatives in very early child
Slovenian (age 1;3 - 2;0; mean age: 1;7; MLU 1;4 – 2;31; mean MLU: 1.94), trying to shed light
on the IRIAH. Slovenian makes an extremely interesting case in terms of imperatives since (i)
the imperative paradigm exhibits a separate conjugation paradigm with [person] and [number]
agreement, not overlapping with any other finite or nonfinite verb form in any mood, (ii) in
syntactic literature, Slovenian imperatives have always been claimed to be finite clauses in
both matrix and embedded contexts (Sheppard & Golden 2000, 2002; Rus 2005), and (iii)
Slovenian is a null subject second position (Wackernagel) clitic language with raised
imperative verbs preceding the clitics.
Our data showed an extremely high number of imperatives (56% of all verb forms) and Bare
Participles (16.3% of all verb forms), with practically no RIs (<1%). However, we found that
imperatives did not lack the TP/AgrP projection(s) because (i) they showed perfect [person]
and [number] agreement (99.7 % corr.), and (ii) they showed structures with object
scrambling and post-imperative clitic placement, which in the present minimalist paradigm
require a full clause structure, contra S&H’s “missing/eliminated functional heads” account.
Conceptually, we also argued that we need not posit an RI-analogue, and that whatever
biological learning and/or processing mechanism is responsible for the differences in early
root nonfinites (Rizzi 2002), early grammars seem to opt for one (or a couple of ) “salient”
verb form(s) besides the indicative finite verb, and that form can be an infinitive (“true RIs” as
in German or Dutch), an imperative and/or a bare participle (“Non-RIs” as in Italian or
Slovenian), or even a bare stem (as in English or Inuktitut).
[to download papers from this project click here]
(C) [2003-] The Acquisition of Finiteness and Early Root Nonfinites in Child Slovenian
KEY WORDS: acquisition of tense, aspect, and agreement, acquisition of finiteness,
development of verb paradigms, early root nonfinites, acquisition of verb
morphosyntax, very early child Slovenian
Summary and Results
Most of the recent child syntax acquisition literature has focused on the acquisition of
finiteness. It has been shown cross-linguistically that children between the ages of 2 and 3
produce extremely few agreement errors when they supply tense and/or agreement markers
(Hyams 2003), but frequently omit finite morphology, producing infinitives (R(oot) I(nfinitives)
Hoekstra & Hyams 1998; Pierce 1992; Platzack 1992; Radford 1990, 1995; Rizzi 1994, etc.) or
Bare Verb stems (BVs) (Wexler 1994, 1998) in finite contexts. Interestingly, children acquiring
Italian (Guasti 1994), Spanish (Torrens 1995), or Catalan (Torrens 1995) do not go through an
RI or BV stage, producing utterances with finite verbs with (near-)perfect agreement.
In this project I wanted to assess early verb acquisition in Slovenian children who are just
beginning to speak. On the basis of naturalistic production data from Kranjc (1999), I showed
that very early child Slovenian (age 1;3 – 2;0; mean age: 1;7; mean MLU: 1.94) falls in the
latter group with no RIs (<1% in our data) or BVs (unattested) and with 98% correct subject-
verb agreement.
Specifically, I showed that early child Slovenian exhibits a full range of functional projections
above the lexical layer (contra Gagarina 2002 for child Russian; Radford 1990, 1995 for child
English; or Varlokosta, Vainikka, and Rohrbacher 1998 for child Greek, for example) in
constructions containing both finite verbs as well as imperatives.
Then I noticed one property of child Slovenian, which, I argued, is a common property of all
child grammars of null subject languages with rich agreement – B(are) P(articiples), i.e.,
active past participles unsupported by auxiliaries, which had been studied very little. While
Slovenian children do not go through an RI stage, almost 40% of all verbal utterances
(excluding imperatives) in my data are utterances containing BPs.
BPs are ungrammatical in adult Slovenian where the corresponding construction involves
either the so-called L(ong) H(ead) M(ovement) construction (due to the Wackernagel second-
position clitic effect) or the past/future tense construction with an overt subject. Compared to
LHM constructions, the child utterances generally only seem to lack the auxiliary be. The
(number/gender) subject-participle agreement on past participles shows almost no errors at
all, just like the verbal paradigm for finite verbs and imperatives.
[to download papers from this project click here]
(D) [2003-2005] Participial Constructions in Child Slovenian
(with Pritha Chandra)
Recently it has been proposed that RIs and bare participles in child grammars should be
treated analogously as 'root non-finites' (RNFs) (Varlokosta, Vainikka, & Rohrbacher [VVR]
1996, 1998). This claim was based on Greek, a language that lacks infinitives. Greek children
usually opt for non–auxiliary participial constructions, with perfective meaning (‘bare
perfectives’).
Pritha and I argued that bare participles are full clauses, containing a phonologically null
auxiliary (and a pro subject). We supported our claim by providing data from very early child
Slovenian (mean age: 1;7; mean MLU: 1.94), a language with no RIs, but extremely rich in
past participles. We found that these are almost always ‘bare’, i.e., without the auxiliary be.
However, what is most challenging for all RNF accounts is the manifestation of full agreement
morphology (person/gender/aspect) on participles, i.e., just like in the verbal paradigm for
finite verbs, where the agreement is practically 100% correct.
Bare participles are ungrammatical in adult Slovenian, the corresponding adult construction
involving long head movement or the past/future tense with overt subjects. The difference
between child Slovenian and its adult counterpart reduces to the lack of auxiliary in the
former.
Using Chomsky's minimalist assumptions that T hosts tense and phi features, we argued that
Slovenian children do not lack a T head, and that bare participles are not analogous to bare
perfectives. The child Greek –i form, traditionally assumed to be a subjunctive form (and
homophonous with 3SG in subjunctive and future clauses), is reported to be frequently
overgeneralized with non-3SG subjects and appropriately used with 3SG agreement only later
in the course of language development when agreement-related properties emerge. However,
in child Slovenian the participial forms are not invariable, always showing correct subject-
participle agreement.
We also noted that VVR’s minimal projection markedness hierarchy cannot hold and that
Slovenian children do not truncate above VP although they have an option of choosing an
infinitive over a participle.
Our final observation was with regard to the semantics of bare participles. We argued that
these in Slovenian do not correspond to Greek bare perfectives in having an irrealis/modal
interpretation, nor do they obey the eventivity constraint, a usual property of RIs (Hoekstra &
Hyams 1998).
[to download papers from this project click here]
Colleagues and professors I've been influenced by most and/or have worked/collaborated
with most in graduate school even though I may disagree with certain things they say :)
(listed in alphabetical order):
HÉCTOR CAMPOS Spanish/Linguistics, Georgetown University
STEPHEN CRAIN Linguistics, Univ of Maryland, now Macquarie University (Australia)
PRITHA CHANDRA Linguistics, University of Maryland
JOHN DRURY Neuroscience, Georgetown University, now McGill University (Canada)
ASHLEY FIDLER Linguistics/Psychology, Georgetown University
MARIA GUASTI Psychology, University of Milan Bicocca (Italy)
NINA HYAMS Linguistics, University of California at Los Angeles
DONNA LARDIERE Linguistics, Georgetown University
HOWARD LASNIK Linguistics, University of Maryland
JEFF PARROTT Linguistics, Georgetown University
JULIAN PINE Psychology, University of Liverpool (UK)
ROSALIND THORNTON Linguistics, University of Maryland, now Macquarie University
MICHAEL ULLMAN Neuroscience/Psychology/Linguistics, Georgetown University
CHARLES YANG Linguistics, University of Pennsylvania
RAFFAELLA ZANUTTINI Linguistics, Georgetown University
many others. Specifically, I argue against many (what I believe are entirely) "ad-hoc" (and
descriptive) linguistic principles and operations with respect to early grammars (e.g., the
Tense Omission operation, the Uniqueness Checking Constraint, the Minimize Violations
Constraint, etc.).
I embrace an innate property of Language as the most minimal one, i.e., referring solely to a
widely-assumed architecture with hierarchy, binarity, locality, and Merge (the concatenation
operation), linking it to what Hauser, Chomsky, and Fitch (2002) call "the faculty of Language
in a narrow sense" (FLN). I show that most of early tense and agreement learning is sensitive
to (and hence procedes at) the interfaces, particularly as morphology-syntax mapping rather
than syntax per se ("biology" as has been often argued by nativists). Hence, I argue, that kids
do not omit tense/agreement because of "biology" since "biology" in my sense is present from
the start (the "syntax-before-morphology" hypothesis, voiced by quite a few researchers).
On the other hand, I also reject many claims coming from "general cognitive" explanations of
early language learning (e.g., "constructivist") and take only those pieces of evidence that I
believe have been luminary for the properties in question (particularly the effects of saliency
and frequency as well as the idea of "gradual learning" rather than instantaneous
convergence to adult grammar).
I additionally present evidence on language processing and the generation (production) of
functional categories in the neurolinguistic tradition and the evidence on language modeling,
weaving both of them together with developmental linguistic facts into an explanatory model
of early tense and agreement marking (see below for details).
My dissertation committee consists of:
Dr. Donna Lardiere (Georgetown; Ph.D. Applied Linguistics, Boston University), co-director
Dr. Charles Yan (Univ. of Pennsylvania; Ph.D. Computer Science, MIT), co-director
Dr. Héctor Campos (Georgetown; Ph.D. Romance Linguistics, UCLA), committee member
Dr. Michael Ullman (Georgetown; Ph.D. Brain & Cognitive Sciences, MIT), committee member
The following problem statement is the official (short) Hypothesis/Problem Statement that I
have filed with Georgetown University (see below for the longer version):
While acquiring early verb morphosyntax, children are amazing learners yet prone to errors.
This study explores how and why this is the case. By viewing a language acquisition process
as a combination of deterministic and probabilistic mechanisms, the study addresses early
tense and agreement marking in child Slovenian, English, and Spanish by using the tools and
empirical findings from a generative linguistic theory, formal theories of early morphosyntax,
and developmental psycholinguistic accounts of early verb learning, supplemented with the
evidence on computational modeling of the learning of verb morphosyntax and neurolinguistic
evidence on the processing and generation of functional categories.
Please note that some details of my hypothesis have been constantly changing since my first
thoughts on the topic.
Below you can check out my Orals Exam Problem Statement (with full bibliography), followed
by the official Dissertation Proposal, followed by the most current abstract.
Any comments and questions are welcome (dr59 [at] georgetown [dot] edu). This is still VERY
MUCH work in progress.
My orals exam problem statement with complete suggested bibliography:
Reconciling Mother Nature and Momma Nurture: Towards a Variability Model of the
Development of Tense and Agreement in Child Grammars
[submitted on April 13 2006 and defended with High Pass on May 17 2006]
My dissertation proposal (long version):
Reconciling Mother Nature and Momma Nurture: The Acquisition of Tense and Agreement in
Child Grammar in a Variability Model of Early Morphosyntactic Development
[submitted on November 14 2006]
My most recent "dissertation abstract":
The Acquisition of Tense and Agreement in early morphosyntax: A biolinguistic perspective
[written in January 2007, and slightly updated in February 2007 for a Biolinguistic
Investigations conference in Santo Domingo, Dominican Republic]
My most recent paper, basically a very condensed version of the dissertation:
The Learning at the Interfaces: Tense and Agreement in a Variability Model. A View from a
Biolinguistic Perspective
[in preparation; to be published in the Georgetown University Working Papers in Theoretical
Linguistics Vol 7, 2007]
Other Projects
(A) [2007- ] The Acquisition of LOCATIVE MARKING in 2-3 year-olds
(with Ashley Fidler)
KEY WORDS: locative marking, case marking, adpositions, linguistic complexity,
cognitive (relational) complexity, perceptual biases, goal/source/state path
distinctions, child English, child Hungarian, child Slovenian
Our Research Statement
Ashley and I have recently started investigating the acquisition of locative markers where it
has been claimed that there appears to be an interaction between linguistic and conceptual
development.
Conceptual factors, such as the early perceptual bias towards GOAL over SOURCE paths are
important predictors of the accuracy of early locative use. Linguistically, moreover, the
differential complexity of locative expressions may influence their development.
Agglutinating case markers and adpositions have been argued to be relatively less complex,
which should lead to more accurate early production. Complex locative markers which
combine adpositions and lexically-selected syncretic case markers, on the other hand, should
be used less frequently in early child language. Thus, research into locative marking must
cross the boundary between linguistic and cognitive development in order to tease apart the
influence of these and other potentially contributing factors.
Assuming this line of reasoning, linguistic complexity should have a noticeable impact on
locative marking in young children cross-linguistically. Very little research, however, has
investigated this hypothesis. Using naturalistic production data from child Slovenian,
Hungarian, and English, then, we are seeking to address the role of linguistic complexity in
relation to conceptual complexity in the locative marking of 2-year-olds. These languages are
good candidates for such an investigation in that they have similar locative semantics, but
diverse linguistic means of expressing these relationships. Slovenian uses complex
morphosyntactic marking which combines prepositions and lexically selected case markers
with a lot of syncretism, Hungarian marks locatives using either agglutinating case markers
with no morphological agreement or postpositions, and English makes use of prepositions of
varying complexity.
We are investigating the following relation types in three corpora: IN, ON, AT, UNDER,
BETWEEN, NEXT TO, IN FRONT OF, and BEHIND. Where they are reliably distinguished, path
distinctions (GOAL vs. STATE vs. SOURCE) are also being assessed. Specifically, we are
looking at how linguistic complexity independently contributes to productive locative marking
in 2-year-olds, particularly the production of Hungarian postpositions, which seem to develop
pretty rapidly and with few errors (though gradually!) and Slovenian preposition + case
amalgams, which seem to be very inaccurate in early development with preposition being
frequently omitted (though nominals almost always carry correct case-marking!).
(B) [2004-2006] The Acquisition of Imperatives in 1-2 year-olds
(some parts of the project with Pritha Chandra)
KEY WORDS: imperatives, acquisition of imperatives, acquisition of finiteness, early
root nonfinites, root infinitives, root infinitive-analogues, very early child Slovenian
Summary and Results
After my theoretical study on the morphosyntax of Slovenian imperatives, Pritha and I carried
out a couple of studies on the acquisition of imperative clause types in children who are just
beginning to speak. We investigated the morphological, syntactic, and semantic properties of
imperatives, with a special emphasis of relating them to the study of finiteness and early root
nonfinites.
Studies on many child languages have shown that most children start off with a root
infinitive (RI) stage, i.e., using an infinitive form where adult contexts use a finite (tense- and
agreement-marked forms). It has been shown that languages such as Catalan, Spanish,
Italian and Slovenian (generally, null subject languages with rich morphology) are exceptions.
This, of course, raises a question of why some languages lack an RI stage, and, more
broadly, if there is a universal RI/RI-analogue stage in early grammars.
Salustri & Hyams (2003) [henceforth S&H] argue that in early Italian, it is the imperative that
represents an RI analogue. S&H state that null subject Non-RI child languages exhibit an
early imperative stage when the frequency of imperatives is extremely high when compared
to other verb forms, as well as to the frequency of imperatives in the target language. Child
imperatives in these languages lack TP/AgrP projections and have the same semantic
interpretation of RIs (in terms of event semantics as well as the realis/irrealis semantic
opposition). Hence, the “imperative as an RI-analogue” hypothesis (IRIAH) provides a
“universal” explanation of children’s syntax and semantics of early root nonfinites and their
analogues.
In our studies, we investigated root nonfinites and imperatives in very early child
Slovenian (age 1;3 - 2;0; mean age: 1;7; MLU 1;4 – 2;31; mean MLU: 1.94), trying to shed light
on the IRIAH. Slovenian makes an extremely interesting case in terms of imperatives since (i)
the imperative paradigm exhibits a separate conjugation paradigm with [person] and [number]
agreement, not overlapping with any other finite or nonfinite verb form in any mood, (ii) in
syntactic literature, Slovenian imperatives have always been claimed to be finite clauses in
both matrix and embedded contexts (Sheppard & Golden 2000, 2002; Rus 2005), and (iii)
Slovenian is a null subject second position (Wackernagel) clitic language with raised
imperative verbs preceding the clitics.
Our data showed an extremely high number of imperatives (56% of all verb forms) and Bare
Participles (16.3% of all verb forms), with practically no RIs (<1%). However, we found that
imperatives did not lack the TP/AgrP projection(s) because (i) they showed perfect [person]
and [number] agreement (99.7 % corr.), and (ii) they showed structures with object
scrambling and post-imperative clitic placement, which in the present minimalist paradigm
require a full clause structure, contra S&H’s “missing/eliminated functional heads” account.
Conceptually, we also argued that we need not posit an RI-analogue, and that whatever
biological learning and/or processing mechanism is responsible for the differences in early
root nonfinites (Rizzi 2002), early grammars seem to opt for one (or a couple of ) “salient”
verb form(s) besides the indicative finite verb, and that form can be an infinitive (“true RIs” as
in German or Dutch), an imperative and/or a bare participle (“Non-RIs” as in Italian or
Slovenian), or even a bare stem (as in English or Inuktitut).
[to download papers from this project click here]
(C) [2003-] The Acquisition of Finiteness and Early Root Nonfinites in Child Slovenian
KEY WORDS: acquisition of tense, aspect, and agreement, acquisition of finiteness,
development of verb paradigms, early root nonfinites, acquisition of verb
morphosyntax, very early child Slovenian
Summary and Results
Most of the recent child syntax acquisition literature has focused on the acquisition of
finiteness. It has been shown cross-linguistically that children between the ages of 2 and 3
produce extremely few agreement errors when they supply tense and/or agreement markers
(Hyams 2003), but frequently omit finite morphology, producing infinitives (R(oot) I(nfinitives)
Hoekstra & Hyams 1998; Pierce 1992; Platzack 1992; Radford 1990, 1995; Rizzi 1994, etc.) or
Bare Verb stems (BVs) (Wexler 1994, 1998) in finite contexts. Interestingly, children acquiring
Italian (Guasti 1994), Spanish (Torrens 1995), or Catalan (Torrens 1995) do not go through an
RI or BV stage, producing utterances with finite verbs with (near-)perfect agreement.
In this project I wanted to assess early verb acquisition in Slovenian children who are just
beginning to speak. On the basis of naturalistic production data from Kranjc (1999), I showed
that very early child Slovenian (age 1;3 – 2;0; mean age: 1;7; mean MLU: 1.94) falls in the
latter group with no RIs (<1% in our data) or BVs (unattested) and with 98% correct subject-
verb agreement.
Specifically, I showed that early child Slovenian exhibits a full range of functional projections
above the lexical layer (contra Gagarina 2002 for child Russian; Radford 1990, 1995 for child
English; or Varlokosta, Vainikka, and Rohrbacher 1998 for child Greek, for example) in
constructions containing both finite verbs as well as imperatives.
Then I noticed one property of child Slovenian, which, I argued, is a common property of all
child grammars of null subject languages with rich agreement – B(are) P(articiples), i.e.,
active past participles unsupported by auxiliaries, which had been studied very little. While
Slovenian children do not go through an RI stage, almost 40% of all verbal utterances
(excluding imperatives) in my data are utterances containing BPs.
BPs are ungrammatical in adult Slovenian where the corresponding construction involves
either the so-called L(ong) H(ead) M(ovement) construction (due to the Wackernagel second-
position clitic effect) or the past/future tense construction with an overt subject. Compared to
LHM constructions, the child utterances generally only seem to lack the auxiliary be. The
(number/gender) subject-participle agreement on past participles shows almost no errors at
all, just like the verbal paradigm for finite verbs and imperatives.
[to download papers from this project click here]
(D) [2003-2005] Participial Constructions in Child Slovenian
(with Pritha Chandra)
Recently it has been proposed that RIs and bare participles in child grammars should be
treated analogously as 'root non-finites' (RNFs) (Varlokosta, Vainikka, & Rohrbacher [VVR]
1996, 1998). This claim was based on Greek, a language that lacks infinitives. Greek children
usually opt for non–auxiliary participial constructions, with perfective meaning (‘bare
perfectives’).
Pritha and I argued that bare participles are full clauses, containing a phonologically null
auxiliary (and a pro subject). We supported our claim by providing data from very early child
Slovenian (mean age: 1;7; mean MLU: 1.94), a language with no RIs, but extremely rich in
past participles. We found that these are almost always ‘bare’, i.e., without the auxiliary be.
However, what is most challenging for all RNF accounts is the manifestation of full agreement
morphology (person/gender/aspect) on participles, i.e., just like in the verbal paradigm for
finite verbs, where the agreement is practically 100% correct.
Bare participles are ungrammatical in adult Slovenian, the corresponding adult construction
involving long head movement or the past/future tense with overt subjects. The difference
between child Slovenian and its adult counterpart reduces to the lack of auxiliary in the
former.
Using Chomsky's minimalist assumptions that T hosts tense and phi features, we argued that
Slovenian children do not lack a T head, and that bare participles are not analogous to bare
perfectives. The child Greek –i form, traditionally assumed to be a subjunctive form (and
homophonous with 3SG in subjunctive and future clauses), is reported to be frequently
overgeneralized with non-3SG subjects and appropriately used with 3SG agreement only later
in the course of language development when agreement-related properties emerge. However,
in child Slovenian the participial forms are not invariable, always showing correct subject-
participle agreement.
We also noted that VVR’s minimal projection markedness hierarchy cannot hold and that
Slovenian children do not truncate above VP although they have an option of choosing an
infinitive over a participle.
Our final observation was with regard to the semantics of bare participles. We argued that
these in Slovenian do not correspond to Greek bare perfectives in having an irrealis/modal
interpretation, nor do they obey the eventivity constraint, a usual property of RIs (Hoekstra &
Hyams 1998).
[to download papers from this project click here]
Colleagues and professors I've been influenced by most and/or have worked/collaborated
with most in graduate school even though I may disagree with certain things they say :)
(listed in alphabetical order):
HÉCTOR CAMPOS Spanish/Linguistics, Georgetown University
STEPHEN CRAIN Linguistics, Univ of Maryland, now Macquarie University (Australia)
PRITHA CHANDRA Linguistics, University of Maryland
JOHN DRURY Neuroscience, Georgetown University, now McGill University (Canada)
ASHLEY FIDLER Linguistics/Psychology, Georgetown University
MARIA GUASTI Psychology, University of Milan Bicocca (Italy)
NINA HYAMS Linguistics, University of California at Los Angeles
DONNA LARDIERE Linguistics, Georgetown University
HOWARD LASNIK Linguistics, University of Maryland
JEFF PARROTT Linguistics, Georgetown University
JULIAN PINE Psychology, University of Liverpool (UK)
ROSALIND THORNTON Linguistics, University of Maryland, now Macquarie University
MICHAEL ULLMAN Neuroscience/Psychology/Linguistics, Georgetown University
CHARLES YANG Linguistics, University of Pennsylvania
RAFFAELLA ZANUTTINI Linguistics, Georgetown University
