Heritage Speaker Competences Differences Language Change and Input Type, Inflected Infinitives

‘International JournalInput and infinitives
of Bilingualism’ in Heritage
• Volume Brazilian
11 • Number 359 – 389|
Portuguese
4 • 2007, | 359
Heritage speaker competence differences,

language change, and input type:
Inflected infinitives
in Heritage Brazilian Portuguese
Jason Rothman
University of Iowa
Acknowledgements
I wish to express my most sincere gratitude to Acrisio Pires whose discussions with me about his
research on this topic enabled and inspired me to write this article. Many of the ideas I pursued
herein were born of these exchanges and are as much, if not more, his as they are mine. I wish to
thank Michael Iverson, Carlos Quicoli, Amanada Owens, and Lisa Dykstra for similar discussions
and helpful suggestions. Equally, I am very grateful to Jennifer Cabrelli whose contacts with BP
heritage speakers in the U.S. enabled me to test more participants to confirm the results found in
a smaller sample size and to Tiffany Judy who has helped immensely in the preparation of this
article. Additionally, I am especially thankful for insightful discussions with Kim Potowski and
Silvina Montrul about issues pertaining to heritage language learners in general and their work on
heritage language learners of Spanish specifically. Any and all errors and oversights are, nonethe-
less, completely my own.
Abstract Key words

It has been argued that colloquial dialects of Brazilian Portuguese (BP) have heritage speakers
undergone significant linguistic change resulting in the loss of inflected
infinitives (e.g., Pires, 2002, 2006). Since BP adults, at least educated ones, inflected
have complete knowledge of inflected infinitives, the implicit claim is that
infinitives
they are transmitted via formal education in the standard dialect. In the
present article, I test one of the latent predictions of such claims; namely,
the fact that heritage speakers of BP who lack formal education in the language change
standard dialect should never develop native-like knowledge of inflected
infinitives. In doing so, I highlight two significant implications (a) that heritage speaker grammars
are a good source for testing dialectal variation and language change proposals and (b) incomplete
acquisition and / or attrition are not the only sources of heritage language competence differences.
Employing the syntactic and semantic tests of Rothman and Iverson (2007), I compare heritage
speakers’ knowledge to Rothman and Iverson’s advanced adult L2 learners and educated native
controls. Unlike the latter groups, the data for heritage speakers indicate that they do not have
target knowledge of inflected infinitives, lending support to Pires’ claims, suggesting that literacy
plays a significant role in the acquisition of this grammatical property in BP.
Address for correspondence

Jason Rothman PhD, Department of Spanish and Portuguese, University of Iowa, Iowa City,
Iowa 52242; Phone: 319 - 541 - 5144; e-mail: < jason-rothman@uiowa.edu >.
‘International Journal of Bilingualism’ The
is ©Kingston Press
International Ltd.of1997
Journal – 2007
Bilingualism
360 J. Rothman
1 Introduction
In a broad sense, the term ‘heritage language’ can describe linguistic acquisition in
many different contexts. Heritage speakers are a subset of all bilinguals, which is to say,
heritage speakers are by definition bilinguals (if one assumes a nonrestrictive definition
of bilingualism in the sense of Haugen (1953)), but not all bilinguals are heritage speakers,
at least in the sense intended here. Importantly, social circumstances constitute the main
variables differentiating heritage speakers from other naturalistic bilinguals. Like all
monolingual and childhood bilingual learners, heritage speakers are exposed naturalisti-
cally to the heritage language; however, this language is by definition a nonhegemonic
minority language within a majority-language environment. Since the heritage language
is the family language used and heard in restricted environments, there are varying
degrees of deterministic consequences for the complete acquisition and / or maintenance
of the heritage language, depending on when and how the societal majority language
is introduced (i.e., simultaneously or successively). Although it is largely accepted that
a heritage language learner need not be a fluent speaker of the heritage language, it is
assumed that a heritage speaker has, to a greater or lesser degree, acquired some level
of proficiency (Valdés, 2001).
Terry Au and colleagues (Au, Knightly, Jun, & Oh, 2002; Oh, Jun, Knightly, & Au,
2003) have demonstrated that exposure to a heritage language as a child has benefits for
phonological acquisition at a later age; however, they claim that there are no residual
benefits for morphosyntax. Recently, Silvina Montrul (2005, 2006, 2007a,b) has chal-
lenged these claims, demonstrating that heritage speakers of Spanish have advantages
over age- and proficiency-matched L2 learners in selective morphosyntactic domains.
This fact makes the linguistic educational needs of heritage learners uniquely different
from the average L2 learner (Lynch, 2003; Potowski, 2005; Roca, 1997; Valdés, 1995, 1997)
in that they are often very proficient speakers of the heritage language but lack functional
literacy, which can have a multitude of linguistic and sociolinguistic consequences.
Beyond the important ramifications that linguistic studies have on heritage language
teaching practices, pioneering work by Montrul (2002, 2005, 2007a,b) and Polinsky
(1997a,b, 2005) in this emerging field has demonstrated that studying heritage language
learners’ linguistic knowledge can also inform linguistic theory, specifically L2 acquisi-
tion theories that stress age as a deterministic variable in ultimate attainment. This article
adds to the claims that studying heritage speakers’ knowledge of specific grammatical
domains can enlighten different areas of linguistic theory. I argue that studying highly
proficient heritage language learners can inform theories of language change and dialectal
variation by providing a test case for the predictions that these theoretical proposals
make. I also argue that studying heritage speaker knowledge of particular grammatical
properties has implications for the role of literacy in the acquisition process and thus
informs acquisition theories as well.
If one takes the position that humans are genetically provided with the general
structure of language and that the acquisition of a primary language has the same
characteristics as any other type of natural growth, then it is reasonable to presume
that language learners never need to become literate speakers to acquire the gram-
matical properties of their native language. Traditionally, this has been the stance of
The International Journal of Bilingualism
Input and infinitives in Heritage Brazilian Portuguese 361
generative linguistics, which envisions the process of language acquisition as a devel-

opmental interaction between the primary linguistic data to which one is exposed and
the language faculty with which we are all hypothesized to be born. Many details aside,
Universal Grammar (UG) delimits linguistic development and fills in the gaps between
the underdetermined input of the environment and the ultimate attainment of particular
language grammars (see Guasti, 2002). However, getting from UG in the initial-state
to the particular language grammar in the steady-state is entirely conditioned upon the
learning of the language-specific lexicon (e.g., Chomsky, 1995, 2000). And so, if accurate,
it seems logical to claim that the only external variable necessary to guarantee linguistic
acquisition is sufficient exposure to input.1
Mere observation provides robust evidence that literacy is indeed not needed in
order to develop grammatical representations. If such were the case, then illiterate adults
would unavoidably have very different grammars than literate adults. This, of course, is
simply not true. In fact, one can acquire multiple languages without being literate in any,
as evidenced by the case of multilingualism in illiterate societies. However, the idea that
literacy does not affect mental grammatical representations oversimplifies the reality
of language learning, which is to say, it idealizes a situation in which every speaker of
any given language is exposed to homogenous input without dialectal variation. Despite
significant syntactic dialectal differences between vernacular and standard dialects,
the fact that all educated learners of a particular language come to know properties
of the standard dialect absent from their colloquial dialect must mean that literacy in
the standard dialect imparts some grammatical knowledge. This is not to suggest that
conscious knowledge of grammatically different properties is necessary for their acquisi-
tion, but rather that formal education provides input of grammatical features missing
from vernacular input. Furthermore, in such cases, literacy can have the added effect of
camouflaging important evidence of language change, since education in the standard
variety preserves grammatical properties that have ceased to exist in colloquial dialects,
properties that are thus not acquired as part of the native dialect. This idea is tested in
the present article via a comparison of knowledge of syntactic and semantic properties
of inflected infinitives between different groups of Brazilian Portuguese (BP) acquirers:
educated native speakers, advanced adult L2 learners, and untutored heritage-language
learners. If it is true that colloquial dialects of BP have lost inflected infinitives (see
Pires, 2002, 2006), being conserved in the standard dialect only, the heritage speakers
would be the only group predicted to receive input in which they are absent, since unlike
monolinguals and tutored adult L2 learners they do not receive formal education in the
standard dialect and presumably have little access to it.
As it relates specifically to heritage language acquisition, the possibility that some
native and heritage speaker competence disparities result from input differences and
unequal formal education opportunities challenges the extent to which it is valid to
suppose that all native / heritage competence differences can be explained by incomplete
acquisition or attrition. While there is little doubt that incomplete acquisition and
attrition exist and are explanatory for many native / heritage differences, it would take
1 I do not intend to ignore that there are important social factors involved in the acquisition of a primary
language.
362 J. Rothman
longitudinal studies that are currently unavailable to really corroborate such assump-
tions, let alone distinguish properly between attrition (what was acquired and then lost),
incomplete acquisition (what was available but simply not acquired) and issues related
to input type (what could not be acquired given its absence in the input).
Taken together, the goals of the present study are: (a) to provide evidence in support
of or against the idea that literacy and formal education can come to bear on underlying
grammatical competence; (b) to confirm or disconfirm the claim that inflected infinitives
have been lost in colloquial BP dialects (Pires, 2002, 2006) and (c) argue that heritage
language acquisition studies must consider the input heritage speakers receive before
making claims of attrition and / or incomplete acquisition to explain competence differ-
ences. This article is structured in the following manner. Section 2 details the predictions
and goals of this project. It also details the relevant syntax and semantics of inflected
infinitives, as well as how this knowledge is assumed to be acquirable within a UG
frame-work. Section 3 discusses previous research. Section 4 describes the methodology,
and the remaining sections present the data and discuss their implications.
2 Purpose and Background

In light of the fact that adult BP speakers reject inflected infinitives as part of their
colloquial dialect and avoid them in favor of alternative finite grammatical forms, Pires
(2002, 2006) with others (e.g., Botelho-Pereira, & Roncarati, 1993; Lightfoot, 1991; but
see Da Luz, 1994; Salles, 2005) have argued that inflected infinitives have died out of
colloquial native dialects of BP (different from the case of European Portuguese (EP),
where they remain in all dialects). However, in light of the fact that inflected infinitives
remain in standard BP, especially in formal writing, the media and high registers of speech
in particular situations (e.g. in school and business), important questions arise regarding
whether inflected infinitives are still acquired as part of BP native grammars, or if they
are only an artifact of literacy / formal education (cf. Pires, personal communication). If
indeed colloquial BP dialects do not have inflected infinitives, then it is reasonable to
assume that children receive little to no input that instantiates them. As a result, there
are several acquisitional predictions whereby each of the following groups should have
indeterminate knowledge of inflected infinitives, contrasting sharply with educated
adult BP speakers (both native and non-native, see Rothman, & Iverson, 2007): (a) adult
native BP speakers who are not educated in standard Portuguese, (b) naturalistic BP L2
learners, (c) children before school-age and (d) BP heritage language adults not educated
in the standard dialect. This article will focus on prediction (d).
Although the answer to this inquiry is an important one insofar as it brings much
to bear on interrelated issues pertinent to formal acquisition, dialectal variation and
language change theorizing, properly addressing and adequately answering it is no small
task. Primarily, the difficulty in determining whether or not inflected infinitives are
acquired as part of native colloquial BP grammars has to do with empirical measurement.
In light of the abovementioned predictions, it perhaps seems easy to test. Firstly, one
could test educated adult native BP speakers to see if they have knowledge of inflected
infinitives. By testing these adult learners, however, we limit ourselves to a retrospective
inference about the path of acquisition based on adult steady-state knowledge. That is,
the mere fact that adult BP learners have determinate knowledge of inflected infinitives
tells us nothing about whether or not they are acquired as part of the native system (i.e.,
outside of schooling). Only in the case that inflected infinitives are not acquired by
these speakers at all will testing adults reveal any significance in this regard, although
it is hardly to be expected that adult speakers educated in standard BP where inflected
infinitives are clearly conserved will not have acquired them at some point.
If formal education is necessary to acquire inflected infinitives, then testing people
not formally educated in the BP standard dialect should reveal significant differences.
By definition, the only monolingual Brazilian speakers not educated in standard BP
are those who are entirely uneducated. In theory, one could test these speakers and
compare their knowledge to educated counterparts. If only the latter group were to have
knowledge of inflected infinitives, one could argue that they are not acquired as part
of the native system. Nevertheless, one cannot ignore the fact that exposure to formal
education is frequently confounded with socioeconomic status and IQ in adults. One
way to separate out these factors would be to test BP children across a variety of ages
from preschool to mid secondary levels.2 Another approach could be to test speakers
who have had clear exposure to higher education, but limited exposure to the target
form, BP heritage speakers, allowing us to test prediction (d).
It is the latter scenario that is tested in this article, and for several reasons. Although
there is a possibility for a role that interference (assuming the parents are themselves
to some extent speakers of the socially dominant language) and variation may play
in the input received by heritage speakers, this is also true for monolingual children.
Ultimately, however, the results confirm that this was not the case. Examining the
predictions for the case of heritage speakers not only addresses similar questions that
investigating BP child acquisition would, but it also has the possible advantage of adding
to the emerging field of heritage language acquisition in general. In the same vein, this
methodology could highlight the resource that heritage speakers are for confirming or
disconfirming theoretical claims about language change and literacy effects on gram-
matical competence. I contend that heritage language learners’ knowledge can hold
the key that unlocks the answer to this inquiry. This is because one of the most salient
differences between highly proficient heritage language learners and age-matched
monolinguals is the lack of formal education (in BP) of the former group only. If literacy
is necessary to acquire inflected infinitives in BP, then the prediction is not only that
preschool-aged children will not have target-like knowledge of inflected infinitives, but
also that heritage speakers, irrespective of age, should have indeterminate3 knowledge.
Moreover, testing adult heritage speakers in comparison to educated adult natives
2 Such a methodology is ideal, since it could precisely pinpoint the moment at which inflected infinitives are
acquired. If it is revealed that children at preschool ages largely have indeterminate knowledge of inflected
infinitives and that they progressively acquire them only after schooling, one has evidence for a determin-
istic role for literacy in grammatical language acquisition, as well as indirect evidence of language change.
Obviously, the opposite observation would disprove such a hypothesis. Pires and Rothman (in progress)
have collected data, which demonstrate the BP monolinguals do not reliably acquire inflected infinitives
until 11 years of age, lending further support to the argumentation in the present article.
3 Unless of course the language of the environment is standard BP, which is more likely to occur in the case
that both parents are first generation speakers of different BP dialects and the standard is spoken as a type
of lingua franca in the family.
364 J. Rothman
and L2 learners not only avoids the confound of education level and familiarity with
formal testing situations, but also avoids confounds related to age, whether these be
general exposure to the world or maturation of linguistic form (Wexler, 1992) since
all participants are adults. This article investigates this prediction by testing heritage
speakers and comparing their knowledge of inflected infinitive properties to data from
Rothman and Iverson (2007), which demonstrated that both educated native controls
and classroom educated advanced L2 learners of BP have robust syntactic and semantic
knowledge of inflected infinitives.
2.1
Inflected infinitives
Crosslinguistically speaking, Portuguese is a rare language in that it has two types of
infinitives (differently from closely related languages such as French, Italian, and Spanish).
Both types are tenseless, yet differentiated via a specification for person / number-Agr
(inflected infinitives) or not (uninflected infinitives). In BP (unlike European Portuguese
(EP), (see Raposo, 1987) only plural forms have corresponding overt morpho-phonological
forms for person / number, as in (1).
(1) Singular Plural
eu sai+ r +Ø nós sai+ r +mos
você vocês
ele sai+ r +Ø eles sai+ r +em
ela elas
‘I /you sg., he, she/we/you pl., they to leave +AGR’
These infinitival forms are not in free variation, which is to say, they have unique
distributions and important syntactic / semantic differences. In some ways, inflected
infinitives act like normal finite clauses, in that they take lexical subjects or null subjects,
unlike uninflected infinitives, as can be seen in (2) as embedded complements of the
factive verb lamentar.
(2) a. Eui lamento eles / pro*i / j não comprarem a casa.
b. *Eu lamento eles não comprar a casa.
c. Eui lamento PROi não comprar a casa.
I regret (pro) / they / PRO not to buy the house
‘I regret (them) not buying the house.’
However, as can be seen in (3) below, inflected infinitives are quite different from
normal finite verbal constructions, yet somewhat similar to uninflected infinitival
clauses in that they never take the complementizer que although they must occur in
embedded clauses. Further distinguishing them from uninflected infinitives is their
ungrammaticality as embedded interrogatives or relative clauses.
(3) a. * É possível que eles chegarem agora.
pro is possible that they arrive-INF-3PPL now
‘It is possible that they arrive now.’
b. É possível que eles cheguem agora.

pro is possible that they arrive-SUBJ now
‘It is possible that they arrive now.’
c. *Eles quererem chegar agora.
Theyi want-INF-3PPL PROi / k arrive-INF now
‘They want them to arrive now.’
d. Eles querem chegar agora.
Theyi want PROi arrive-INF now
‘They want to arrive now.’
e. *Eles chegarem agora.
They arrive-INF-3PPL now
‘They to arrive now.’
f. Eles chegam agora.
They arrive now
‘They arrive now.’
g. Não sabemos que comprar para a festa.
proi (we) not know what PROi buy-INF for the party
‘We do not know what to buy for the party.’
h. * Não sabemos que comprarmos para a festa.
proi (we) not know what proi buy-INF-1PPL for the party
‘We do not know what to buy for the party.’
2.2
Syntactic and semantic properties
There are significant syntactic and semantic differences between inflected and uninflected
infinitives; for example, how they behave with respect to properties of control. Whereas
uninflected infinitives display interpretive properties of obligatory control, inflected
infinitives display properties of nonobligatory control (see Pires, 2001, 2006).
Since inflected infinitives have either a lexical or null subject, their subject must be
disjoint in reference from any DP in the sentence; however, the subject PRO of uninflected
infinitives must have a local c-commanding antecedent in the matrix clause, as can be
seen in (4) and (5).
(4) [O meu pai]i lamenta PROi perder a bolsa.
[The my father]i regrets PROi loose-INF the bag
‘My father regrets having lost the bag.’
(5) [O meu pai]i lamenta pro*i perdermos a bolsa.
[The my father]i regrets pro*i loose-INF-1PPL the bag
‘My father regrets our having lost the bag.’
As can be seen in the ellipsis examples in (6), uninflected infinitives must take a
sloppy reading under ellipsis, whereas inflected infinitives only correspond to a strict
interpretation of the ellipsis site.
366 J. Rothman
(6) a. A Maria lamenta ter ido embora e o Felipe também. ( = Felipe lamenta ter ido embora).
The Mariai regrets PROi / *j have-INF left and Felipe too. (= Felipe regrets
having left)
‘Maria regrets having left and Felipe does too.’
b. A Maria lamenta termos ido embora e o Felipe também. (= Felipe lamenta nós
termos ido embora).
The Mariai regrets proi have-INF-1PPL left and Felipe too. (= Felipe regrets
our leaving)
‘Maria regrets our having left and Felipe does too.’
With uninflected infinitives as in (6a), the elided material can only be interpreted
with a sloppy reading, corresponding to “Felipe regrets his own leaving,” as opposed to
inflected infinitives as in (6b), for which the elided material must be interpreted with a
strict reading of the ellipsis site, corresponding to “Felipe laments our leaving.”
Additionally, there are differences between inflected and uninflected infinitives in
terms of allowing (or not) split antecedents for embedded clause null subjects. Consider
the following sentences in (7):
(7) a. Eui convenci o Leoj PROj / *i + j a compartilhar o sorverte com o João.
Ii convinced the Leoj PROj / *i + j share-INF the ice cream with the João
‘I convinced Leo to share his ice cream with João.’
b. Eui convenci o Leoj proi + j a compartilharmos o sorvete com o
Ii convinced the Leoj proi + j / j +k share-INF-1PL the ice cream with the
João.
João
‘I convinced Leo for us to share our ice cream with João.’
In (7a), PRO does not allow an interpretation where Eu and Leo can form a set
that serves as its antecedent. Conversely, in (7b), the embedded pro must be coreferential
with a set of elements that includes, at the very least, Eu and Leo. Crucially, in light of
the plural Agr-morphology of the inflected infinitive, it may not be coreferential with
Leo only.
2.3
Acquiring inflected infinitives
By all accounts of language acquisition (i.e., not only generative approaches), particular
grammar (PG) properties can only be acquired insofar as there is some type of trig-
gering evidence from the environmental input. Mentioning this fact is of no small
consequence given the present discussion. That is, if indeed vernacular varieties of BP
have lost inflected infinitives, then children (and even adults) exposed exclusively to
these dialects simply have no recourse to acquire them. As a result, such claims are not
only relevant for documenting dialect variation and linguistic change, but also have
acquisitional consequences (such as those discussed in Section 2), making unequivocally
verifiable / falsifiable predictions for specific instances of language acquisition. However,
in noting that speakers must be exposed to input that provides relevant positive evidence
for acquisition to take place, it is in no way suggested that every property of a PG must
be learned in a construction-specific manner, as quite the contrary is assumed to be the
case (e.g., properties derived via clustering).
Within Minimalism, parameters are located within the PG functional lexicon
(Chomsky, 1995, 2000). PG lexicons vary in terms of which functional categories and
features (as well as the type of feature; e.g., interpretable vs. uninterpretable) they
instantiate. In other words, language-to-language differences in feature composition
related to language-specific morphology and functional categories are assumed to be the
locus of parametric difference. Of course, these differences have a number of syntactic
consequences. One such consequence is the possibility or not of inflected infinitives,
which only a few languages, such as Portuguese, Galician, and some dialects of Sardian
have (see Longa, 1994; Uriagereka, 1995).
Raposo (1987) and Quicoli (1988, 1996) demonstrated independently that the
possibility of inflected infinitives is conditioned upon the interaction of a syntactic
parameter (the [+null-subject] of the Null-Subject Parameter) and the positive setting
of a morphological parameter (the Infl-parameter). Both Rapaso and Quicoli propose
that the Infl structure is a morphological parameter of UG that contains values for
Tense and Agr. Verbs can be valued as [±Tense], [±AGR]. In light of Portuguese data,
it is argued that finite Infl is specified for [+Tense] and an infinitival Infl is specified
for [−Tense], independently of Agr. However, in the majority of the world’s languages,
such as English and Spanish, if Infl is finite it is necessarily specified for Agr and the
converse also holds (i.e., when Infl is infinite it cannot be specified for Agr).
Raposo proposed that the positive value of the Infl-parameter in Portuguese
allows for a free choice of [±Tense] in an Infl with Agr. Inflected infinitives, therefore,
obtain from the possibility of having [[−Tense] Agr]. According to Raposo (1987, p. 92),
in the absence of [+Tense], Infl (or Agr in Infl) assigns nominative case to its subject if
it itself is specified for case.4 Verbal Agr is a set of interpretable Φ-features for number,
person, and optionally case, mapped to a morpho-phonological form in null-subject
languages only (Alexiadou & Anagnostopoulou, 1998; Chomsky, 1981). It follows then
that a language with inflected infinitives taking nominative lexical subjects must be a
null-subject language and that the choice of [±Tense] is free of the choice Agr [±case], as
in (8). In pro-drop languages that also have the positive setting of the Infl-parameter (i.e.,
Portuguese and Galician, but not Spanish and Italian), the compilation of features as in
(8d) is possible, and thus allows for infinitives that are inflected for person / number Agr.
(8) a. NP [+Tense] Agr[−c] VP — Chinese finite constructions without agreement
b. NP [+Tense] Agr[+c] VP — Finite constructions
c. NP [−Tense] Agr[−c] VP— ECM constructions
d. NP [−Tense] Agr[+c] VP— Inflected Infinitives
While being a [+ null-subject] language is a necessary condition to have inflected

infinitives, it is hardly a sufficient condition to guarantee them, as evidenced by the fact
4 This claim is motivated by facts of obligatory subject-verb inversion in EP (see Raposo, 1987), although
it should be noted that there are criticisms to some of the details of his proposal and that the word order
restrictions are not the same in BP (see Da Luz, 1994, 1998; Galves, 1991; Madeira, 1995; Pires, 2001).
368 J. Rothman
that the vast majority of pro-drop languages do not have inflected infinitives. And so,
how do acquirers of Portuguese set the Infl-parameter to the positive setting? Pires (2001,
2006) applies Lightfoot’s cue-scanning approach to language acquisition (Lightfoot,
1999), in which the input is argued to be the locus for identifying the purported UG-given
structural cues needed to set any given PG syntax. Although it must be acknowledged
that cues in the sense of Lightfoot (1999) are fundamentally different from the notion of
cues traditionally employed within associationist approaches to language acquisition,
as in MacWhinney and colleagues’ Competition Model (MacWhinney, & Bates, 1989;
MacWhinney, Leinbach, Taraban, & McDonald, 1989), I join Rothman and Iverson
(2007) in assuming a slightly different position in an effort to avoid any terminological
confusion. Similar to Pires’ account 5 and in line with minimalist assumptions, I assume
that learners of Portuguese must acquire the correct feature composition of the different
types of Portuguese inflectional morphology via exposure to them from input in order
to converge on a grammar that permits finite verbal forms, uninflected and inflected
infinitives. Since the positive setting of the Infl-parameter allows for tenseless Agr
morphology, children must learn that Portuguese morphology can encode both tense
and person / number or just person / number features.
A consequence of claiming that colloquial BP dialects do not have inflected infini-
tives is that it effectively claims that BP is negatively valued for the Infl-parameter. As a
result, like Spanish and English, this means that there is only one type of verbal agreement
morphology, which is always specified for tense. If this is true, colloquial BP input does
not provide positive evidence that Agr can be independent of Tense. As a result, children
exposed to colloquial BP only should have no recourse to acquire inflected infinitives.
However, we know that learners educated in the standard dialect (be them native or
advanced adult L2 speakers) come to acquire all relevant syntactic and semantic proper-
ties of inflected infinitives (Rothman, & Iverson, 2007). If the claim that BP colloquial
dialects have lost inflected infinitives is accurate, then formal education in the standard
dialect is the variable that explains their acquisition in adult grammars.
3 Previous research and current direction

To my knowledge, there is only one study that has examined the acquisition of BP
inflected infinitives. Rothman and Iverson (2007) tested adult learners of non-native BP
and compared their knowledge of inflected infinitives to a group of educated BP natives.
In interpretive tasks only, they demonstrated that formally educated adult L2 learners
of BP performed exactly like formally educated native BP speakers, demonstrating clear
knowledge of the syntax and semantics of inflected infinitives. Since inflected infinitives
5 I thank Acrisio Pires for pointing out to me that there are contexts in standard BP in which replacing
an inflected infinitive with an indicative / subjunctive finite form would be really awkward, if not
ungrammatical as in the embedded context of (i) and (ii):
i) [Vocês terminarem esse trabalho] é mais importante do que nós sairmos para almoçar.
Notice that the right dislocated form is fine with a subjunctive (even so, only of the first inflected
infinitive can — although it doesn't have to — be replaced, as in (ii)). Such a replacement would be
extremely awkward in (i):
ii) É mais importante [que vocês terminem esse trabalho] do que nós sairmos para almoçar.
are almost never obligatory, which is to say, there is an equally acceptable finite form
that can be used in its place, Rothman and Iverson decided to focus on interpretive and
grammaticality judgments as opposed to measurements of production.
In light of the data from Rothman and Iverson (2007), we know that formally
educated BP natives and advanced L2 learners acquire the syntactic distribution and
interpretive properties of inflected infinitives. This fact does not automatically invali-
date the claim that colloquial BP dialects do not have inflected infinitives; in fact, it is
perfectly consistent with such a claim. That is, in light of their exposure to the standard
Portuguese dialect, it is not unexpected that educated natives and highly proficient
tutored L2 learners would have knowledge of inflected infinitives (assuming the possi-
bility of postcritical period parameter resetting for L2 learners); however, determining
whether or not (a) inflected infinitives are acquired as part of the native BP system or
(b) if formal education is a deterministic variable is not discernable from their data due
to the level of education in the standard dialect of both groups. As a result, the data
from Rothman and Iverson are neutral with respect to questions of dialectal variation
and language change.
Theoretical proposals of syntax, dialectal variation and language change do not
exist in a vacuum and, therefore, must be parsed against acquisition data. Insofar as they
make predictions under particular acquisition contexts, the tenability of these proposals
can only be authenticated in the case that these predictions are validated by controlled
acquisition data. Although the data from Rothman and Iverson cannot be used to
directly pursue questions of language change and the role of literacy in the acquisition
of inflected infinitives, they can be used as a point of comparison against which data
from heritage speakers could directly address these issues. Employing the same tasks
with heritage language BP speakers who are differentiated from the native group from
Rothman and Iverson by education in standard BP and from their adult L2 group by
timing of acquisition and formal education, we can effectively attend to the question of
whether or not inflected infinitives are only an artifact of the standard dialect as a result
of education in it. And so, this acquisition study will weigh heavily on the tenability of
proposals of BP dialect variation and language change by either confirming the claims
of Pires (2002, 2006) or by not supporting the implications of his claims.
4 Methodology 6
This section details the design and methodology of the present study. The tests used are
the same tests employed by Rothman and Iverson (2007) in an effort to make the heritage
6 It was suggested by a reviewer that due to an inherently unpredictable range of linguistic histories
and competences, traditional / standard experimental designs are perhaps insufficient. This is an
astute observation and worthy of serious consideration. Nevertheless, it is important to highlight
that the heritage speakers in this study are extremely proficient speakers, although this does not
mean that they are as good in reading (ability and comprehension) as age-matched monolinguals.
In part, the patterns observed in this study suggest that such was not a confound (insofar as dispari-
ties were observed only in stimuli that examined knowledge of inflected infinitives specifically).
Future research will benefit from this reviewer’s suggestion, using more oral (elicited production)
and listening measures to test for the same knowledge. It was also suggested that the appropriate
control group should be one of European Portuguese speakers. This is problematic for several
370 J. Rothman
speakers’ data maximally comparable to their native control group and advanced L2
learner group. The first test was a grammaticality judgment task (GJT), which tested
for distributional restrictions on the use of inflected infinitives (e.g., only in embedded
clauses, not after the complimentizer que, not as embedded questions). Using Hornstein’s
(1999) diagnostics for testing properties of obligatory versus nonobligatory control,
the second task tested for knowledge of interpretive restrictions on inflected versus
uninflected infinitives such as possible readings under ellipsis and the (im)possibility
of split antecedent interpretations of embedded null subjects.
4.1
Participants
The heritage language learner group consists of 11 individuals. Although this is a small
sample size, making this study more exploratory, it is a considerable amount of subjects
when one considers the total amount of BP speakers in the U.S. All of the individuals
are highly fluent in BP and none of the individuals received consistent formal education
in Portuguese, except for one who moved back and forth between the U.S. and Brazil
whereby she was in Brazil between the ages of 7 and 13. Both parents of the heritage
speakers are Brazilian and 8 of 11 participants were born in Brazil. In such a case, the
participants moved to the U.S. before the age of eight (range 1 – 8). All of the participants
reported spending a significant amount of time in Brazil with family and a few reported
moving back and forth throughout their childhood / adolescence. Moreover, all of the
participants reported using Portuguese exclusively with their parents and other family
members and all reported using Portuguese on a daily basis. The ages of the participants
ranged from 18 to 25.
These heritage speakers are compared to two groups from Rothman and Iverson:
(a) a group of educated native controls (n = 19) and (b) a group of English advanced adult
learners of L2 Portuguese (n = 17). The age range of all three groups is comparable as
are their levels of education (in a general sense) and their socioeconomic level. The main
linguistic variable differentiating the heritage speakers from the other two groups is a
lack of education in standard BP and exposure to standard BP input.
4.2
Test 1: Grammaticality Judgment / Correction task
The goal of the GJT was to test the participants’ knowledge of the grammatical distribu-
tion of inflected infinitives. This GJT consisted of six sentences types, as in (9) through
(13) (n = 5 each sentence type). Importantly, the directions made the difference between
preference and grammaticality patently clear (via examples not related to inflected infini-
reasons. It ignores the fact that the target grammar is not the grammar of a Portuguese dialect, but
the grammar that is actually attained by adult speakers of Brazilian Portuguese. The most educated
speakers of Brazilian Portuguese master many if not all the properties of the standard grammar
and yet, their standard grammar is clearly distinct from the grammar of European Portuguese.
The latter is not a grammar acquired or learned by any Brazilian speakers (unless perhaps they
grow up in an European Portuguese speaking environment), nor is it the standard grammar that
is learned by educated speakers and used in the media and in formal settings.
tives, involving unparsable word order violations as an example of an ungrammatical

property and poor word choice as an example of preference.
(9) Inflected infinitives as complements of factive matrix verbs
O João lamenta as meninas não terem chegado.
The João regrets the girls not have-INF-3PPL arrived
‘João regrets the girls not having arrived.’
(10) Inflected infinitives as complements of declarative matrix verbs

O Marco afirma não conhecermos a cidade.
The Marco claims not know-INF-1PPL the city
‘Marco claims that we do not know the city.’
(11) Inflected / uninflected infinitives as embedded interrogatives / relative clause

a. * Não sabemos com quem falarmos para maiores informações.
pro not know with whom speak-INF-1PPL for more information
‘We do not know with whom to speak for more information.’
b. Não sabemos quem convidar á festa.

pro not know whom invite-INF to the party
‘We do not know whom to invite to the party.’
(12) Inflected infinitives in matrix clauses

*Eles saírem ao bar nas sextas.
They go out-INF-3PPL to the bar on Fridays
‘They go out to the bar on Fridays.’
(13) Inflected infinitives w/ the complementizer ‘que’

*Quantas pessoas estão aqui? Penso que estarem muitas
How many people are here? pro think that be-INF-3PPL many
‘How many people are here? I think that there are many.’
Sentences (9) and (10) exemplify grammatical uses of inflected infinitives. Sentences
like (11b) are grammatical (as they are in English), however, similar sentences using
inflected infinitives, as in (11a), are ungrammatical because inflected infinitives cannot
be used in relative clause / embedded interrogative contexts. Sentences like (12) and (13)
are all ungrammatical since inflected infinitives must be in embedded contexts although
they never take the complementizer que. When participants deemed a sentence ungram-
matical, they were asked to correct the sentence if they were able to do so. This procedure
ensured that the learner indicated ungrammaticality for the right reason.
4.3
Test 2: Context Sentence Match task
Under the assumption that target knowledge of inflected infinitives must entail knowl-
edge of related interpretive properties, this task tested for awareness of the obligatory
372 J. Rothman
versus nonobligatory properties of control that differentiate uninflected versus inflected

infinitives. Success or failure on this test provides evidence that the learner’s underlying
morphosyntactic system includes or does not include inflected infinitives. However, one
may argue that learners could accomplish this task successfully, even without having
target knowledge of inflected infinitives per se, if they simply realized that the Agr
morphology bounds the subject. As we will see, this was not the case.
This task was a context / sentence-matching task, testing for knowledge of obligatory
sloppy versus strict readings under ellipsis and the possibility (or not) of split antecedent
interpretations of the null subjects of inflected versus uninflected infinitives. A context
was provided followed by two sentences: (a) one that had an inflected infinitive and (b)
one that had an uninflected infinitive. The participants were asked to circle the sentences
that logically corresponded to the context. They were instructed to circle both sentences
if they believed that both were possible. There were four types of context / sentence pairs,
as in (14), of which there were 10 exemplars each (i.e., a total of 40 context matches).
(14) a. Sloppy reading context
Quando o nosso pai morreu a minha irmã chorou em frente de todos. Por isso, ela se
sentia um pouco envergonhada. Mais tarde, ela me disse que estava muito orgulhosa
de mim porque pensou que eu era muito forte. Ela nunca soube que eu tinha chorado
também porque ninguém me viu chorar.
When our father died my sister cried in front of everyone. As a result, she felt a
little embarrassed. Later, she told me that she was very proud of me because she
thought I was very strong. She never knew that I had also cried because nobody
saw me cry.
Which sentence(s) is (are) logical given the context?

i. Eu lamento ter chorado e a minha irmã também.
I regret have-INF cried and the my sister too
‘I regret having cried and my sister does too.’
ii. Eu lamento termos chorado e a minha irmã também.

I regret have-INF-1PL cried and the my sister too
‘I regret our having cried and my sister does too.’
b. Strict reading context

Ontem era o dia da partida de futebol mais importante do ano. Eu pensei que fossemos
ganhar, mas a gente perdeu. Agora estou muito triste e não quero sair. Realmente, eu
não posso acreditar que não ganhamos. A minha namorada está muito triste também
porque agora eu não quero sair de casa.
Yesterday was the most important day for soccer of the whole year. I thought we
were going to win, but we lost. Now I am very sad and I don’t want to go out. Truly,
I just can’t believe that we did not win. My girlfriend is also quite sad because now
I don’t want to leave my house.

i. Eu lamento ter perdido e a minha namorada também.

I regret have-INF lost and the my girlfriend too
‘I regret having lost and my girlfriend does too.’
ii. Eu lamento termos perdido e a minha namorada também.

I regret have-INF-1PL lost and the my girlfriend too
‘I regret our having lost and my girlfriend does too.’
c. Split antecedent w/ PRO?

A Marta e o Roberto eram namorados por 3 anos. Os dois são bons amigos meus. A
semana passada, A Marta soube que o Roberto tinha beijado outra mulher durante a
primeira semana da sua relação. Obviamente a Marta estava muito triste e ela jurou
que nunca mais falaria com ele. Eu não queria que a Marta odiasse o Roberto por
isso falei com ela.
Marta and Roberto were together for three years. Both are my good friends. Last
week, Marta found out that Roberto had kissed another girl during the first week
of their relationship. Obviously, Marta was very upset and she swore that she
would never talk to him again. I did not want Marta to hate Roberto so I spoke
with her.

i. Eu convenci a Marta a perdoar o Roberto.
I convinced the Marta to forgive-INF the Robert
‘I convinced Marta to forgive Robert.’
ii. Eu convenci a Marta a perdoarmos o Roberto.

I convinced the Marta to forgive-INF-3PL the Robert
‘I conviced Marta that we forgive Robert.’
d. Split antecedent w/pro?

A minha melhor amiga, a Joana, não tem muito dinheiro mas precisa mudar fora da
casa dos pais dela. Ela declara que ainda mora com eles para conservar dinheiro. Mas
tem 28 anos e ela precisa ter mais liberdade e independência. Felizmente depois de
dois anos de tentar convencê-la, ela aceitou a minha oferta de alugar um apartamento
comigo.
My best friend, Joana, does not have a lot of money, but she needs to move away
from her parents’ house. She claims that she still lives with them to save money.
However, she is 28 and needs to have more freedom and independence. Luckily,
after two years of trying to convince her, she accepted my offer to rent an apart-
ment with me.

i. Eu convenci a Joana a alugar um apartamento.
I convinced the Joana to rent-INF an apartment
‘I convinced Joana to rent an apartment.’
374 J. Rothman
ii. Eu convenci a Joana a alugarmos um apartamento.

I convinced the Joana to rent-INF-1PL an apartment
‘I convinced Joana that we rent an apartment.’
The contexts provided with the sentences effectively delimit the correct choice,
which is to say, there was only one felicitous sentence for each context. Sentence matches
with uninflected infinitives were expected for contexts like (14a) and (14c) since these
contexts presented a sloppy reading under ellipsis and an environment that precluded
a set-reading of the matrix subject and object as an antecedent respectively. Inflected
infinitive sentences were the logical choice for contexts like (14b) and (14d) since these
contexts presented a strict reading of the ellipsis site and a set reading that included the
matrix subject and object as an antecedent.
5 Results and Discussion

This section is divided into three parts, two of which correspond to the empirical tests
and the third of which corresponds to an overall discussion of the data. Each of the first
two parts is subdivided into three components: (i) a descriptive analysis of the group
results, (ii) a quantitative statistical analysis of the group data which compares the mean
score performance across all groups, and (iii) a look at individual performances. First, a
one-way ANOVA was employed and this was followed up by Tukey pair-wise comparisons
when needed. The statistics were conducted using the mean number correct for each
group. An answer was deemed correct if it was in accord with the theoretical analysis
presented above, which was confirmed by the native control group.
5.1
Task 1
5.1.1
Descriptive analysis
The goal of the grammaticality judgment correction task was to test for knowledge of
the (un)grammaticality of several types of (un)inflected infinitive sentences: declarative
matrix predicates with inflected infinitival complements (DMP), factive matrix predi-
cates (FMP) with inflected infinitival complements, inflected / uninflected infinitives
as embedded interrogatives / relative clauses (InI EI / RC and Inf EI / RC, respectively),
inflected infinitives in matrix clauses (MC InI), and inflected infinitives in embedded
clauses after the complementizer que (InI w / que). The numerical analysis was based on the
average number of sentences accepted in each context (n = 5 for each individual context).
As can be seen in Figure 1, the heritage speaker group performed significantly
differently from Rothman and Iverson’s (2007) native control and advanced L2 learner
groups, who performed strikingly similarly to each other. The relevant comparisons for
this test were intergroup for each context, comparing the native control performance
to both the heritage speakers and the English L2 learner performances, as well as
the latter two groups’ performances to each other. Unlike the other two groups, the
heritage speakers do not demonstrate knowledge of the distribution of inflected versus
noninflected infinitives previously discussed. That is, heritage speakers do not reliably
Figure 1
Results Task 1: Grammaticality Judgment task
DMP = Declarative Matrix Predicate; FMP = Factive Matrix Predicate; InI EI / RC = Inflected
Infinitives as Embedded Interrogatives or Relative Clauses; Inf EI / RC = Uninflected Infinitives as
Embedded Interrogatives or Relative Clauses; MC InI = Inflected Infinitives as Matrix Clauses;
InI w / que = Inflected Infinitives under complementizer que
accept inflected infinitives in embedded contexts with declarative and factive matrix
predicates although they do reliably accept uninflected infinitives in embedded inter-
rogative / relative clause contexts. Also, they do not consistently reject inflected infinitives
used as matrix verbs, after the complementizer que and inflected infinitives in embedded
interrogative / relative clause contexts. In fact, they seemingly accept inflected infinitives
in these ungrammatical contexts more than they do in actual grammatical contexts. I
return to this observation in Section 5.3.
5.1.2
Statistical analysis
One-way ANOVA tests were used to quantify the group data. Statistical analysis revealed
significant differences for all relevant comparisons, except for uninflected infinitives as
embedded interrogatives / relative clauses: DMP (F = 40.1, p <.001); FMP (F = 45.3, p <.001);
InI EI / RC (F = 55.5, p <.001); Inf EI / RC (F = 0.74, p =.48); MC InI (F = 78.6, p <.001); InI
w / que (F = 49.4, p <.001). The fact that there were no significant differences in grammati-
cality judgments for uninflected infinitives as embedded interrogatives / relative clauses
is not surprising since this infinitival use is grammatical in all dialects of Portuguese and
in English. In light of the fact that the ANOVA revealed significant differences across
the three groups for all other contexts, follow-up tests were performed to see where the
significant differences lie. A significant difference was found between the natives and
the heritage learners for all of these contexts (DMP (t = 5.28, p <.001); FMP (t = 5.91,
p <.001); InI EI / RC (t = 6.42, p <.001); MC InI (t = 7.19, p <.001); InI w / que (t = 6.38,
p <.001)); however, there was only a significant difference between the natives and L2
376 J. Rothman
English speakers for inflected infinitives used after que (InI w / que, t = 3.04, p =.008). In
comparing the heritage speakers to the L2 learners we note that they performed statisti-
cally differently from the L2 learners as well on all five of the same sentence types (DMP
(t = 5.86, p <.001); FMP (t = 5.97, p <.001); InI EI / RC (t = 6.25, p <.001); MC InI (t = 6.71,
p <.001); InI w / que (t = 5.11, p <.001)). All of this is summarized in Table 1 below.
Table 1
Statistical results Task 1
DMP FMP InI EI/RC Inf EI/RC MC InI InI w/que

t (f) p t (f) p t (f) p t (f) p t (f) p t (f) p
(ANOVA) 40.1 <.001 45.3 <.001 55.5 <.001 0.74 .48 78.6 <.001 49.4 <.001
NS v. HS 5.28 <.001 5.91 <.001 6.42 <.001 1.03 .31 7.19 <.001 6.38 <.001
NS v. Eng
1.36 .19 0.19 .85 0.58 .57 1.03 .32 1.85 .08 3.04 .008
L2
Eng L2
5.86 <.001 5.97 <.001 6.25 <.001 0.05 .96 6.71 <.001 5.11 <.001
v. HS
NS=native speakers; Eng L2=adult L2 learners; speakersHS = Heritage Speakers; DMP = Declarative Matrix
Predicate; FMP = Factive Matrix Predicate; InI EI / RC = Inflected Infinitives as Embedded Interrogatives or
Relative Clauses; Inf EI / RC = Uninflected Infinitives as Embedded Interrogatives or Relative Clauses; MC
InI = Inflected Infinitives as Matrix Clauses; InI w / que=Inflected Infinitives under complementizer que
It is worth noting that although there is a native / L2 learner difference for one
sentence type, inflected infinitives after the complementizer que (InI w / que), this diver-
gence is quite different than the heritage speaker difference from both groups. The
difference between the L2 learner group and the natives stems from the lack of native
speaker variation, and is not indicative of gross deviation from target-like behavior,
whereas the heritage learners performed very differently from both groups. When the
averages of each group acceptance are compared (natives: 0., English L2: 0.53, heritage
speakers: 3.36), it is evident that the English L2 speakers have an extremely high tendency
(88%) to reject sentences in which inflected infinitives are used after the complementizer
que, but heritage speakers did not (33%).
5.1.3
Individual results
Although the group results demonstrated that heritage speakers seemed to not have
knowledge of inflected infinitives in sharp contrast to the native control and adult L2
learners, analyzing the individual data would allow us to determine whether the heritage
speaker group trend accurately depicted all individual performances, or instead obscured
important variation within the aggregate analysis. As can be seen in Table 2, only two
of 11 individual heritage speakers performed within the range of the native speakers for
all sentence contexts. This is significant as it demonstrates that some of these heritage
speakers acquired inflected infinitives. Heritage Speakers (d) and (h) performed entirely
like the native controls, a point to which I return in Section 5.3.
Table 2
HS individual results Task 1

Native
4.53 4.74 0.11 4.79 0.00 0.00
Average
Native
3-5 4-5 0-1 4-5 0 0
Range
HS a 2 2 3 5 4 4
HS b 1 1 4 4 4 4
HS c 0 2 5 4 5 5
HS d 4 5 0 5 1 0
HS e 2 2 5 5 3 3
HS f 3 3 4 4 4 4
HS g 2 3 5 5 5 4
HS h 5 4 1 5 0 0
HS I 1 2 4 5 4 4
HS j 0 1 4 4 5 5
HS k 1 1 3 5 4 4
5.2
Task 2
5.2.1
Descriptive analysis
This task tested the nonobligatory control properties of inflected infinitives as compared
to obligatory control properties of uninflected infinitives (i.e., obligatory sloppy read-
ings under ellipsis with uninflected infinitives, obligatory strict readings of the ellipsis
site with inflected infinitives, as well as the (im)possibility of split antecedents for null
subjects of embedded uninflected (PRO) and inflected (pro) infinitives). Since each
sentence type allowed only one interpretation and each context clearly corresponded
to one interpretation, answers were deemed incorrect either if the sentence chosen to
match the context did not correspond to the structure that yielded the proper reading
or if both sentences were circled. In the case of the heritage speakers only, many incor-
rect answers were a result of circling both options, which demonstrated that they had
indeterminate knowledge.
For this test, the relevant comparisons made were intergroup for each context,
comparing the heritage speaker performance to both the natives and L2 groups. As
can be seen in Figure 2, the native and L2 groups’ behavior was remarkably similar
although the heritage speaker group’s performance was markedly different from both
groups. Only the natives and L2 learners performed as one would expect of a grammar
that has inflected infinitives. Not surprisingly, however, all three groups correlated
sloppy readings under ellipsis with uninflected infinitives and they did not allow split
antecedent interpretations with PRO, the subject of the uninflected infinitive. This
378 J. Rothman
Figure 2
Results Task 2
Unin w/ellip = Acceptance of Uninflected Infinitives in Sloppy Reading Contexts; Inf w/ellip =
Acceptance of Uninflected Infinitives in Strict Reading Contexts; Inf PRO = Acceptance of Split
Antecedent Readings with Uninflected Infinitives; InI pro = Acceptance of Split Antecedent Readings
with Inflected Infinitives
is expected of all three groups since knowledge of obligatory control properties of

uninflected infinitives is a given, which is to say, uninflected infinitives are part of
every dialect of Portuguese and are the same in English. Conversely, heritage speakers
performed somewhat randomly with respect to deriving strict reading interpretations
of the ellipsis site with inflected infinitives and allowing split antecedent interpretations
for the pro subject of inflected infinitives unlike the native control and L2 groups. This
provides further evidence that only the heritage speaker group does not have inflected
infinitives in its Portuguese grammar.
5.2.2
Statistical analysis
One-way ANOVA tests were used to quantify the group data. Statistical analysis of all
contexts revealed significant differences across the groups in two contexts only, both
of which involved the interpretive properties of inflected infinitives: in ellipsis contexts
(Uninflected (Unin): F = 48.71, p <.001; Inflected (InI): F = 1.14, p =.33); in split antecedent
contexts (w/PRO: F = 0.23, p =.79; w / pro: F = 37.91, p <.001). Therefore, follow-up tests were
only needed for the two inflected infinitive contexts. As Rothman and Iverson (2007)
demonstrated, there were no significant differences in any contexts between the natives
and the advanced L2 learners. And so, these newly revealed intergroup differences must
be between the heritage speakers and at least one, if not both of the other two groups.
Indeed, the heritage speakers performed quite differently than the natives (ellipsis context
(Inf): t = 6.26, p <.001) and split antecedent w / pro: t = 5.87, p <.001) and the L2 learners
(ellipsis context (Inf): t = 5.83, p <.001) and split antecedent w / pro: t = 5.21, p <.001) in
both inflected infinitive contexts. All of this can be seen in Table 3 below.
Table 3
Statistical results Task 2
Inf w/ellip InI w/ellip Inf PRO InI pro

t (f) p t (f) p t (f) p t (f) p
(ANOVA) 48.71 <.001 1.14 .33 0.23 .79 37.91 <.001
NS v. HS 6.26 <.001 1.32 .20 0.05 .96 5.87 <.001
NS v. Eng L2 0.75 0.46 1.31 .20 0.62 .54 1.05 .30
Eng L2 v. HS 5.83 <.001 0.07 .95 0.57 .58 5.21 <.001
HS = Heritage Speakers; Unin w/ellip = Acceptance of Uninflected Infinitives in Sloppy Reading Contexts;
Inf w/ellip = Acceptance of Uninflected Infinitives in Strict Reading Contexts; Inf PRO = Acceptance of Split
Antecedent Readings with Uninflected Infinitives; InI pro = Acceptance of Split Antecedent Readings with
Inflected Infinitives
5.2.3
Individual results
Table 4 below provides the results of the individual performances on Task 2 for the heritage
speaker group, which can be compared against the native speaker range. As was the case
for Task I, the individual results for Task II more or less confirm the applicability of the
group trends to the individual performances. In general, the heritage speakers performed
significantly differently from the range of native speaker individuals. This confirms
that the aggregate trend of having indeterminate knowledge of semantic properties of
inflected infinitives is more or less representative of individual performances within the
group, with the exception of the same two individuals from Task 1: (HL d) and (HL h).
Table 4
HS Individual results Task 2
Inf w/ ellip InI w/ellip Inf Pro InI pro

Native Average 9.68 0.21 0.26 9.63
Native Range 9-10 0-1 0-1 8-10
HS a 5 1 0 6
HS b 6 1 1 4
HS c 7 0 0 5
HS d 9 1 0 10
HS e 6 1 0 7
HS f 5 0 0 6
HS g 4 0 0 4
HS h 9 0 0 9
HS I 3 1 1 4
HS j 7 0 0 6
HS k 6 0 1 5

380 J. Rothman
5.3
Discussion
The purpose of this section is to bring together the results of both tasks and discuss
their possible significance. At first glance, the data from task one are a bit surprising.
While on the one hand it is to be expected that heritage learners would reject felicitous
uses of the inflected infinitive if indeed they do not have inflected infinitives in their
grammar (which was confirmed by their tendency to judge as ungrammatical proper
uses of inflected infinitives), it seems unexpected that they would judge as grammatical
sentences in which inflected infinitives are used ungrammatically. That is, if their
grammar is devoid of inflected infinitives, it is reasonable to anticipate that they would
reject both good and bad uses, but this was not the case. In fact, they largely accepted
inflected infinitives as matrix predicates, after the complementizer que and as embedded
interrogatives or relative clauses. This is especially true at the individual level, which
is to say, when one takes out the native-like performances of heritage Speakers (d) and
(h), this subset aggregate data overwhelmingly show a tendency to accept the inflected
infinitive form as grammatical in these contexts. In Figure 3 below, the averages of this
subgroup, which excludes Speakers (d) and (h) can be seen as they compare to the native,
L2 learner, and the superset heritage speaker groups.
Figure 3
Results Task 1 with Heritage Subgroup

The relevant statistical analyses of this subject heritage speaker group are presented
in Table 5 below.
Table 5
Statistical Results Task 1 (HS Subgroup)

t (f) p t (f) p t (f) p t (f) p t (f) p t (f) p
(ANOVA) 93.4 <.001 106 <.001 227 <.001 0.60 .55 430 <.001 207 <.001
NS v. HS 5.28 <.001 5.91 <.001 6.42 <.001 1.03 .31 7.19 <.001 6.38 <.001
NS
8.83 <.001 10.15 <.001 14.81 <.001 1.17 .27 19.0 <.001 20.5 <.001
v. HS Sub
NS
1.36 .19 0.19 .85 0.58 .57 1.03 .32 1.85 .08 3.04 .008
v. Eng L2
Eng L2
5.86 <.001 5.97 <.001 6.25 <.001 0.05 .96 6.71 <.001 5.11 <.001
v. HS
Eng L2
9.81 <.001 10.22 <.001 13.42 <.001 0.37 .71 16.73 <.001 13.5 <.001
v. HS Sub
HS = Heritage Speakers; HS Sub = Heritage Speaker Subgroup; DMP = Declarative Matrix Predicate;
FMP = Factive Matrix Predicate; InI EI / RC = Inflected Infinitives as Embedded Interrogatives or Relative
Clauses; Inf EI / RC = Uninflected Infinitives as Embedded Interrogatives or Relative Clauses; MC InI = Inflected
Infinitives as Matrix Clauses; InI w / que = Inflected Infinitives under complementizer que
So why would they accept what is wrong on the one hand and judge ungrammatical
what is correct on the other? When one takes a closer look at the contexts and the types
of corrections that the heritage speakers provided a pattern emerges. In the case that the
sentences were rejected, the inflected infinitives were corrected most typically by simply
inserting the complementizer que to the original sentences, although in a few cases the
sentences were rewritten with que accompanied by (in most cases) an appropriate finite
form. In the case an inflected infinitive was incorrectly accepted, it was always in a
context in which a finite form (indicative or subjunctive) would have been grammatical
(as the matrix verb, in an embedded relative clause construction and after que). Taken
together, this pattern seems to indicate that heritage speakers analyze inflected infini-
tival morphology similarly to indicative / subjunctive morphology. That is, it seems that
the identical agreement morpho-phonology that in one case has person / number / case
features only (inflected infinitives) and in the other has person / number / case and tense
features (indicative / subjunctive forms) is being analyzed as the later always. In the case
that the heritage speakers corrected the sentences, this seemed to be motivated not by the
inappropriate use of inflected infinitival morphology itself, but by the lack of an environ-
ment that would permit a finite form (i.e., without the complementizer que). And so, the
contexts in which an inflected infinitive is grammatically impossible become possible if
the inflected infinitival morphology is analyzed as some type of indicative / subjunctive
form. This explains this seemingly unexpected data from Test 1.
The data of Test 2 strengthen the observation from Test 1, which indicates that
the majority of heritage speakers have indeterminate knowledge of inflected infinitives.
382 J. Rothman
While they interpret uninflected infinitives with obligatory control, they clearly have
indeterminate knowledge that inflected infinitives display properties of nonobligatory
control. As was the case in Test 1, however, heritage Speakers (d) and (h) performed
like monolingual natives. Excluding their data, we see that the remaining subgroup of
heritage speakers performed even more at chance levels than the superset aggregate on
the two relevant contexts.
Figure 4
Results Task 2 (HS Subgroup)
Table 6
Statistical Results Task (HS Subgroup)
Inf w/ellip InI w/ellip Inf PRO InI pro

t (f) p t (f) p t (f) p t (f) p
(ANOVA) 98.9 <.001 1.07 .351 0.40 .67 93.1 <.001
NS v. HS 6.26 <.001 1.32 .20 0.05 .96 5.87 <.001
NS v. HS Sub 9.26 <.001 1.17 .27 0.36 .73 11.3 <.001
NS v. Eng L2 0.75 .46 1.31 .20 0.62 .54 1.05 .30
Eng L2 v. HS 5.83 <.001 0.07 .95 0.57 .58 5.21 <.001
Eng L2 v. HS Sub 8.56 <.001 0.07 .92 0.82 .43 9.64 <.001
HS = Heritage Speakers; HS Sub = Heritage Speaker Subgroup; Unin w/ellip = Acceptance of Uninflected
Infinitives in Sloppy Reading Contexts; Inf w/ellip = Acceptance of Uninflected Infinitives in Strict
Reading Contexts; Inf PRO = Acceptance of Split Antecedent Readings with Uninflected Infinitives;
InI pro = Acceptance of Split Antecedent Readings with Inflected Infinitives

For the subset aggregate, deviant responses in general took one of two forms. Most
typically, wrong responses reflected the circling of both choices despite the fact that
the context clearly allowed only for either an inflected or uninflected infinitive. This
behavior indicated confusion on the part of the heritage speaker when one compares
this tendency with their perfect performance on contexts in which uninflected infinitives
were expected. The second type of error, although less common, involved the changing
of the otherwise correct inflected infinitive sentence by inserting que and, in few cases,
by changing the inflected infinitive form to a truly finite morphological form. This latter
type of correction provides further evidence that the heritage speakers analyze inflected
infinitives as some type of finite verbal form. Furthermore, this behavior is revealing
in that the directions did not specify that they could / should correct sentences. In fact,
the directions made it clear that one of the two or both sentences was / were correct. It is
reasonable to suppose that the heritage learners knew that an uninflected infinitive was
not the best choice in light of these contexts; however, without grammatical knowledge
of inflected infinitives they answered either randomly, confused by the sentences that
seemingly had a finite verb without the complentizer que, or attempted to “fix” the
second choice.
The discussion thus far supports the proposal that inflected infinitives are not
acquired by heritage speakers, which is consistent with the hypothesis that inflected
infinitives are no longer available in BP primary linguistic data (Pires, 2002, 2006).
However, if this were the case, then one would not expect there to be individuals in the
heritage speaker group that have determinate knowledge of inflected infinitives. However,
as we have discussed, there were two such heritage speakers. Before abandoning the
hypothesis that BP primary linguistic data does not have inflected infinitives, we should
take a closer look at the linguistic history of these particular individual heritage speakers
to see if something differentiates them from the remainder of the group. A follow up
interview revealed the following linguistic history. Heritage Speaker (d) was born in
Brazil and moved to the U.S. when she was one year old. After spending several years
in the U.S., including preschool, she moved back to Brazil just before 1st grade. She was
then educated in Brazil until the age of 13 when she returned to the East coast of the U.S.
for the rest of her schooling. It is important to note that she classifies herself as English
dominant, but unlike her peers in this study she did receive education in standard BP
between the age of 7 and 13. In fact, she is the only heritage speaker to have received
any type of education in Brazil past the first grade. This information gives us anecdotal
evidence that inflected infinitives are imparted via formal education / exposure to the
standard between the ages of eight (the oldest age at which English was introduced to
any of the heritage speakers after leaving Brazil) and 13 (when this learner left Brazil
for the second time). This also confirms the findings of Pires and Rothman (in progress)
who report data from tests with BP children ranging from 6 –15 in which they pinpoint
11 –12 to be the age at which monolingual BP children have adult-like knowledge of
inflected infinitives.
Heritage Speaker (h) is a different case in that he moved to the U.S. at eight and
thus, received only a 1st grade education in Brazil. It could be argued that this provides
evidence that inflected infinitives are learned immediately via formal education (as
early as 1st grade), but this was not the case for other heritage speakers (n = 2) that also
384 J. Rothman
received a 1st grade education in Brazil. However, it is interesting to note that the parents
of this speaker were from different areas in Brazil where the vernacular varieties are
quite different (the mother was from Curitiba and the father was from Rio de Janiero). In
light of this fact, it is possible that the family language in his home was one of dialectal
leveling, whereby it reflected more the standard BP dialect: the lingua franca of both
parents. In such a case, heritage Speaker (h) was likely exposed to inflected infinitives,
which by all accounts are instantiated in the standard variety.
There is another possible explanation for why these heritage learners do not come
to acquire inflected infinitives, namely because they take BP to be a non-null subject
grammar7. As discussed in Section 2.3, a language can only be positively set for the
Inflection parameter (thus have inflected infinitives) if it is a null-subject language
(Raposo, 1987). It has been argued that BP has been in transition from a pro-drop
language to a pro-retention language, evidenced by the frequency of overt subjects as
compared to Spanish, Italian and even European Portuguese as well as an observed
simplification of the morphological verbal paradigms in many colloquial dialects
(see works in Kato & Negrão, 2000; Roberts, 2007). If heritage speakers have a non-
pro-drop BP grammar, then they must operate under the assumption that BP has one
type of agreement morphology that is always specified for agreement and tense. Since the
heritage speakers are bilingual, English-dominant speakers, it may be the case that it is
most economical for these bilinguals to function with BP as a [-null subject] language (in
the sense of Satterfield, 2003). However, while not impossible, this is unlikely for several
reasons. First, there is no argument that monolingual BP is a non-null-subject language
in the sense that English is, but a hybrid type system in which unmarked subject forms
with defined reference are expressed as full pronouns and null-subjects are conserved
for expressing nonreferential subjects. That means that heritage learners would have
received input (verbal morphology with nominal features and clear cases of pro subjects)
to converge on a grammar that is a null-subject grammar for BP just like monolingual
BP children. Furthermore, such an analysis would not explain the difference in the two
heritage learners that did demonstrate target knowledge of inflected infinitives, nor
could it explain the evidence found in the acquisition patterns of BP children (Pires, &
Rothman, in progress). Future research that independently tests for heritage speaker
knowledge that BP is a null-subject grammar (such as testing for null-subject related
knowledge of Overt Pronoun Constraint restrictions on interpretation) will enable us
to (dis)confirm this alternative account.
6 Conclusions
The goal of modern linguistics is not only to describe, but also to explain, grammatical
systems and how they come to be acquired. As a result, theoretical linguistics cannot
be separated from acquisitional linguistics. Theoretical claims, whether they are about
the structure of grammatical properties, their mental representation, language varia-
tion or change, have implications that must bear out under scrutiny from acquisition
data. All too often this important relationship is functionally overlooked. However,
7 I thank Teresa Satterfield for bringing this to my attention.

any theoretical proposal worthy of serious consideration must match the claims that it
makes for acquisition. And so, acquisition facts gained from empirical evidence can be
used to support theoretical proposals, inform them and / or motivate necessary changes
to them. In the present case, lack of determinate knowledge of inflected infinitives by
heritage language learners of BP as compared to educated natives and adult L2 learners
educated in the standard dialect provides preliminary evidence in support of claims
that colloquial BP dialects have lost inflected infinitives (Botelho-Pereira et al., 1993;
Lightfoot, 1991; Pires, 2002, 2006).
I use the word preliminary since there is another acquisition prediction which would
confirm the conclusions drawn from this data; that is, child acquirers of BP should
also have indeterminate knowledge of inflected infinitives until they receive sufficient
exposure to the standard dialect and linguistic education in formal schooling. This is
true since one could argue that heritage language learners are not the ideal candidates
given issues of cross-linguistic transfer, incomplete acquisition and / or possible linguistic
attrition. Given the fact that these heritage speakers are English-dominant, it is possible
that inflected infinitives have been attrited from their Portuguese grammar. This claim,
however, assumes that inflected infinitives were acquired by them and later lost, which
further assumes that they were exposed with environmental input that sufficiently
provided positive evidence for this property. Although, in such a case, why would two
of the 11 learners have conserved inflected infinitives while the other nine lost them?
Another possibility is that the heritage language learners suffered incomplete acquisition
of particular structures in BP as a result of the introduction of the English grammar.
Again, it makes an implicit assumption that such a property was exemplified in the BP
input to which they had exposure.
I argue that both of these possibilities are unlikely. The first idealizes a situation
in which some structures had been acquired, but due to issues of language use and type
of exposure to a given property is lost. However, none of the learners reported a change
in the type of input or frequency over time. That is, they all use Portuguese everyday,
have received ample input from the same sources throughout their lives and reported
Portuguese as the exclusive family language at home. Moreover, given the frequency with
which they use Portuguese, they have all remained very active bilinguals who cannot
be reliably differentiated from monolingual speakers, at least phonologically. Secondly,
if incomplete acquisition were an issue, one could logically expect to see a difference
in knowledge between those individuals exposed to English as early as one year and
those as late as eight years of age since the introduction of the second grammar was
introduced at different crucial periods of linguistic development. Over half of the 11
heritage speakers (n = 6) were six years or older when English was first introduced, an
age at which the adult native grammar is largely considered to have reached a steady-
state (Guasti, 2002). Interestingly, however, only one of these six demonstrated target
knowledge of inflected infinitives.
Ultimately, heritage language learners may be a fruitful source to test effects of
literacy on the acquisition of particular properties. Such a reality brings to our attention
the fact that language standardization and literacy can shape the grammars of native
speakers who are educated and thus mask important language change in vernacular
varieties. It is a fact that heritage language speakers are likely to receive little input from
386 J. Rothman
the standard dialect insofar as they, by definition, grow up with more functional and less
formal contact with the heritage language. And so, if formal education is a necessary
variable for the acquisition of some properties, particularly those lost in vernacular
varieties and conserved by the standard dialect, it is safe to assume that their grammars
will be devoid of them.
Innovative work by Silvina Montrul (2004, 2005, 2006, 2007a,b) and Maria Polinsky
(1997a,b, 2005) has recognized the importance of examining heritage language grammars
under the assumption that it can inform us on several different fronts. Essentially, they
assume a position of incomplete acquisition or attrition to explain how heritage language
learners’ grammars can be different from monolingual grammars, especially in their
competence of important interpretive differences (e.g., where the indicative / subjunctive
distinction renders a semantic difference in presupposition, see Montrul, 2007a). To say
nothing of the fact that it is virtually impossible to differentiate a posteriori between
attrition and incomplete acquisition for adult learners since in either case one must
render an assumption based on a retrospective reporting of acquisitional circumstances,
claiming that heritage speaker differences must ensue from attrition or incomplete
acquisition ignores another possibility. That is to say, both attritional or incomplete
acquisition accounts idealize a situation in which all of the missing properties were
available to the heritage language learner to acquire in the first place, and for some
reason they were acquired then lost or never acquired at all. This may very well be
the case for some properties (maybe even most), especially properties that are harder
to acquire such as lexical semantics and properties at the syntax-pragmatics interface
(Montrul, 2004; Sorace, 2004; Tsimpli, Sorace, Heycock, & Filiaci, 2004). However,
there is another possibility: namely, that heritage language learners do not acquire
particular properties because they are, simply put, not exposed to certain structures
that are effectively transmitted via formal education. Such a reality would not make
these properties any less grammatical, but merely highlight the fact that literacy can
come to affect grammatical systems, providing learners with input containing properties
that are not exemplified in colloquial dialects. Since the eventual answer to the ques-
tion how is competence determined in heritage speakers must include tenable proposals
accounting for native / heritage competence differences, it is of no small consequence
to highlight the fact that researchers must examine the input type heritage speakers
receive, ensuring that the input unambiguously provides triggers (relevant features) for
the properties under investigation before taking a position that incomplete acquisition
and attrition are the causes of observable differences. In the future, longitudinal studies
of individual heritage learners will be able to control for the necessary variables to be
able to differentiate between these three possible sources for native / heritage competence
differences, most likely revealing that all three possibilities couple together to explain
the full gamut of relevant differences.
The present study has brought attention to the fact that heritage language speakers
are an important, largely untapped source for linguistic inquiry. Not only are they inter-
esting to study for the unique nature of their process of language acquisition, but they
serve as a bridge to investigate the role that literacy can have on linguistic competence as
well as a test case for language change proposals insofar as they enable a disentangling of
variables that are otherwise impossible in educated monolingual speakers. The findings
of the present study provide conditional evidence of language change in BP supporting

Pires’ and others claims and suggest a deterministic role for literacy in the acquisition of
particular grammatical properties. Further research in child BP acquisition of inflected
infinitives will strengthen the implications of this study. Moreover, if literacy is a neces-
sary variable to acquire particular grammatical properties argued to have undergone
linguistic change in colloquial dialects, other studies that examine heritage speakers’
knowledge of similar properties will continue to have implications for dialectal variation
and language change theorizing.
manuscript received: 05.2007

revision received: 10.2007
manuscript accepted: 10.2007
References
ALEXIADOU, A., & ANAGNOSTOPOULOU, E. (1998). Parametrizing Agr: Word Order,
V-Movement and EPP-Checking. Natural Language and Linguistic Theory, 16(3), 491 – 539.
AU, T., KNIGHTLY, L., JUN, S., & OH, J. (2002). Overhearing a language during childhood.
Psychological Science, 13(3), 238 – 243.
BOTELHO-PEREIRA, M. A., & RONCARATI, C. N. (1993). O caso do sujeito em orações
infinitivas introduzidas por “para” no português do Rio. DELTA, 9(1), 15 – 30.
CHOMSKY, N. (1981). Lectures on government and binding. Dordrecht: Foris.
CHOMSKY, N. (1995). The Minimalist Program. Cambridge, MA: MIT Press.
CHOMSKY, N. (2000). New horizons in the study of language and mind. New York: Cambridge
University Press.
DA LUZ, G. A. (1994). O Infinitivo Pessoal: Uma análise do português contemporâneo. Master’s
thesis. UNICAMP.
DA LUZ, G. A. (1998). Inf lected infinitives in Romance languages. Cadernos de Estudos
Lingüísticos, 34, 7 – 17.
GALVES, C. (1991). Inflected infinitives and Agr licensing. Ms., UNICAMP.
GUASTI, M. T. (2002). Language acquisition: The growth of grammar. Cambridge, MA: MIT Press.
HAUGEN, E. (1953). The Norwegian language in America: A study in bilingual behavior. Philadelphia:
University of Pennsylvania Press.
HORNSTEIN, N. (1999). Movement and control. Linguistic Inquiry, 30, 69 – 96.
KATO, M. A. & NEGRÃO, E. (Eds.). (2000). Brazilian Portuguese and the Null Subject Parameter.
Frankfurt: Iberoamericana.
LIGHTFOOT, D. (1991). How to set parameters: Arguments from language change. Cambridge,
MA: MIT Press.
LIGHTFOOT, D. (1999). The development of language: Acquisition, change and evolution. Oxford:
Blackwell.
LONGA, V. M. (1994). The Galician inflected infinitive and the theory of UG. Catalan Working
Papers in Linguistics, 4, 23 – 44.
LYNCH, A. (2003, August). The relationship between second and heritage language acquisition:
Notes on research and theory binding. Heritage Language Journal.
MacWHINNEY, B., & BATES, E. (1989). The crosslinguistic study of sentence processing. New
York: Cambridge University Press.
MacWHINNEY, B., LEINBACH, J., TARABAN, R., & McDONALD, J. (1989). Language
Learning: Cues, or rules? Journal, of Memory, & Language, 28, 255 – 277.
388 J. Rothman
MADEIRA, A. M. (1995). Topics in Portuguese syntax: The licensing of T and D. PhD disserta-
tion. College of London.
MONTRUL, S. (2002). Incomplete acquisition and attrition of Spanish tense / aspect distinctions
in adult bilinguals. Bilingualism: Language and Cognition, 5(1), 39 – 68.
MONTRUL, S. (2004). Subject and object expression in Spanish heritage speakers: A case of
morpho-syntactic convergence. Bilingualism: Language and Cognition, 7(2), 125 – 142.
MONTRUL, S. (2005). Second language acquisition and first language loss in adult early bilinguals:
Exploring some differences and similarities. Second Language Research, 21(3), 199 – 249.
MONTRUL, S. (2006). On the bilingual competence of Spanish heritage speakers: Syntax, lexical-
semantics and processing. International Journal of Bilingualism, 10(1), 37 – 69.
MONTRUL, S. (2007a). Interpreting mood distinction in Spanish as a Heritage language. In K.
Potowski & R. Cameron (Eds.), Spanish in contact policy, Social and linguistic inquiries.
Amsterdam: John Benjamins.
MONTRUL, S. (2007b). On the (incomplete) acquisition of object expression in Spanish: How similar
are L2 learners and Spanish Heritage Speakers. Ms. University of Illinois Urbana-Champaign.
OH, J., JUN, S., KNIGHTLY, L., & AU, T. (2003). Holding on to childhood language memory.
Cognition, 86(3), B53 – B64.
PIRES, A. (2001). The syntax of gerunds and infinitives: Subjects, case and control. PhD disserta-
tion. University of Maryland, College Park.
PIRES, A. (2002). Cuebased change: Inflection and subjects in the history of Portuguese infini-
tives. In D. Lightfoot (Ed.), Syntactic effects of morphological change, (pp.143 – 159). Oxford:
Oxford University Press.
PIRES, A. (2006). The minimalist syntax of defective domains: Gerunds and infinitives. Amsterdam:
John Benjamins.
PIRES, A., & ROTHMAN, J. (in progress). Acquisition of inflected infinitives in native Brazilian
Portuguese: Implications for theories of language change and dialectal variation. Ms.
University of Michigan, Ann Arbor and University of Iowa.
POLINSKY, M. (1997a). Cross-linguistic parallels in language loss. Southwestern Journal of
Linguistics, 14, 1 – 45.
POLINSKY, M. (1997b). American Russian: Language loss meets language acquisition. In W.
Browne et al., (Eds.), Formal approaches to Slavic linguistics, (pp.370 – 407). Ann Arbor:
Michigan Slavic Publications.
POLINSKY, M. (2005). Word class distinctions in an incomplete grammar. In D. Ravid &
H. Shyldkrot (Eds.), Perspectives on language and language development, (pp. 419 – 436).
Dordrecht: Kluwer.
POTOWSKI, K. (2005). La enseñanza del español a los hablantes nativos. Madrid, Spain:
ArcoLibros.
QUICOLI, A. C. (1988). Inflection and parametric variation: Portuguese versus Spanish. Ms.
University of California, Los Angeles.
QUICOLI, A. C. (1996). Inflection and parametric variation: Portuguese versus Spanish. In R.
Freidin (Ed.), Current issues in comparative grammar, (pp. 46 – 80). Dordrecht: Kluwer.
RAPOSO, E. (1987). Case theory and Infl-to-Comp: The inflected infinitive in European
Portuguese. Linguistic Inquiry, 18, 85 – 109.
ROBERTS, I. (2007). Diachronic syntax. New York: Oxford University Press.
ROCA, A. (1997). Retrospectives, advances, and current needs in the teaching of Spanish to
United States Hispanic bilingual students. ADFL Bulletin, 29, 37 – 43.
ROTHMAN, J., & IVERSON, M. (2007). To inflect or not to inflect is the question indeed:
Infinitives in non-native Portuguese. The Journal of Portuguese Linguistics, 6(2).
SALLES, H. (2005). Exceptional case marking in Brazilian Portuguese. In M. Batllori, M. L.
Hernanz, C. Picallo, & R. Roca, (Eds.), Grammaticalization and parametric variation.
Oxford: Oxford University Press.
SATTERFIELD, T. (2003). Economy of interpretation: Patterns of pronoun selection in transitional

bilinguals. In V. J. Cook (Ed.), L2 effects on the L1. Clevedon: Multilingual Matters.
SORACE, A. (2004). Native language attrition and developmental instability at the syntax-
discourse interface: Data, interpretations and methods. Bilingualism: Language and
Cognition, 7, 143 – 145.
TSIMPLI, T., SORACE, A., HEYCOCK, C., & FILIACI, F. (2004). First language attrition and
syntactic subjects: A study of Greek and Italian near-native speakers of English. International
Journal of Bilingualism, 8, 257 – 277.
URIAGEREKA, J. (1995). An F position in Western Romance. In K. E. Kiss (Ed.), Discourse
configurational languages, (pp. 153 – 175). New York: Oxford University Press.
VALDÉS, G (1995). The teaching of minority languages as academic subjects: Pedagogical and
theoretical challenges. Modern Language Journal, 79, 299 – 328.
VALDÉS, G, (1997). The teaching of Spanish to bilingual Spanish-speaking students: Outstanding
issues and unanswered questions. In M. C. Colombi and F. Alarcón (Eds.), La ensenanza
del espanol a hispanohablantes: Praxis y teoria (pp. 8 – 44). Lexington, MA: D.C. Health.
VALDÉS, G. (2001). Introduction. Spanish for Native Speakers, Vol. 1 AATSP professional devel-
opment series handbook for teachers K–12. New York, New York: Harcourt College.
WEXLER, K. (1992). Some issues in the growth of control. In R. K. Larson, S. Iatridou, U.
Lahiri & J. Higginbotham (Eds.), Control and grammar (pp.253 – 295).

Heritage Speaker Competences Differences Language Change and Input Type, Inflected Infinitives

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Heritage Speaker Competences Differences Language Change and Input Type, Inflected Infinitives

Uploaded by

Copyright:

Available Formats

‘International JournalInput and infinitives

Heritage speaker competence differences,

Abstract Key words

Address for correspondence

generative linguistics, which envisions the process of language acquisition as a devel-

2 Purpose and Background

b. É possível que eles cheguem agora.

While being a [+ null-subject] language is a necessary condition to have inflected

3 Previous research and current direction

tives, involving unparsable word order violations as an example of an ungrammatical

(10) Inflected infinitives as complements of declarative matrix verbs

(11) Inflected / uninflected infinitives as embedded interrogatives / relative clause

b. Não sabemos quem convidar á festa.

(12) Inflected infinitives in matrix clauses

(13) Inflected infinitives w/ the complementizer ‘que’

versus nonobligatory properties of control that differentiate uninflected versus inflected

Which sentence(s) is (are) logical given the context?

ii. Eu lamento termos chorado e a minha irmã também.

b. Strict reading context

Which sentence(s) is (are) logical given the context?

i. Eu lamento ter perdido e a minha namorada também.

ii. Eu lamento termos perdido e a minha namorada também.

c. Split antecedent w/ PRO?

Which sentence(s) is (are) logical given the context?

ii. Eu convenci a Marta a perdoarmos o Roberto.

d. Split antecedent w/pro?

Which sentence(s) is (are) logical given the context?

ii. Eu convenci a Joana a alugarmos um apartamento.

5 Results and Discussion

DMP FMP InI EI/RC Inf EI/RC MC InI InI w/que

DMP FMP InI EI/RC Inf EI/RC MC InI InI w/que

is expected of all three groups since knowledge of obligatory control properties of

Inf w/ellip InI w/ellip Inf PRO InI pro

Inf w/ ellip InI w/ellip Inf Pro InI pro

The International Journal of Bilingualism

The International Journal of Bilingualism

DMP FMP InI EI/RC Inf EI/RC MC InI InI w/que

Inf w/ellip InI w/ellip Inf PRO InI pro

The International Journal of Bilingualism

7 I thank Teresa Satterfield for bringing this to my attention.

of the present study provide conditional evidence of language change in BP supporting

manuscript received: 05.2007

SATTERFIELD, T. (2003). Economy of interpretation: Patterns of pronoun selection in transitional

The International Journal of Bilingualism

You might also like