The role of language processing

in language acquisition

Colin Phillips and Lara Ehrenhofer

Department of Linguistics, Maryland Language Science Center,
University of Maryland

Language processing research is changing in two ways that should make it more
relevant to the study of grammatical learning. First, grammatical phenomena are
re-entering the psycholinguistic fray, and we have learned a lot in recent years
about the real-time deployment of grammatical knowledge. Second,
psycholinguistics is reaching more diverse populations, leading to much
research on language processing in child and adult learners. We discuss three
ways that language processing can be used to understand language acquisition.
Level 1 approaches (“Processing in learners”) explore well-known phenomena
from the adult psycholinguistic literature and document how they play out in
learner populations (child learners, adult learners, bilinguals). Level 2
approaches (“Learning effects as processing effects”) use insights from adult
psycholinguistics to understand the language proficiency of learners. We argue
that a rich body of findings that have been attributed to the grammatical
development of anaphora should instead be attributed to limitations in the
learner’s language processing system. Level 3 approaches (“Explaining learning
via processing”) use language processing to understand what it takes to
successfully master the grammar of a language, and why different learner
groups are more or less successful. We examine whether language processing
may explain why some grammatical phenomena are mastered late in children
but not in adult learners. We discuss the idea that children’s language learning
prowess is directly caused by their processing limitations (‘less is more’:
Newport, 1990). We conclude that the idea is unlikely to be correct in its
original form, but that a variant of the idea has some promise (‘less is
eventually more’). We lay out key research questions that need to be addressed
in order to resolve the issues addressed in the paper.

Linguistic Approaches to Bilingualism 5:4 (2015), 409–453. doi

10.1075/lab.5.4.01phi issn 1879–9264 / e-issn 1879–9272 © John Benjamins
Publishing Company
410 Colin Phillips and Lara Ehrenhofer
1. Introduction

Language processing research is changing in at least two ways that should make
it more interesting to people whose concerns involve successful and
unsuccessful learning of grammars. First, grammatical phenomena are re-
entering the psycholinguistic fray after a period of partial exile. We have learned
a lot recently about the real-time deployment of grammatical knowledge.
Second, psycholinguistics is reaching more diverse populations, such as child
and adult learners, in addition to the college-age native speakers who are the
staple diet of psycholinguists. This is partly because experimental tools have
become cheaper. But little of the research depends on recent technological
advances, and the key to progress depends mostly on identifying new questions.
Our aim here is to give an (opinionated) sketch of how language processing
should be harnessed in the service of understanding language learning. We do
not want to simply generate a lot of complicated data by throwing new(ish)
experimental measures at our favorite learner population. Rather, we want to use
an understanding of real time grammatical computation to understand how and
when learning succeeds in young children, and why it is generally less successful
in later learners. As psycholinguists, we welcome the increased attention from
language acquisition experts in what we do. But we hope that this will not draw
attention away from the central problems that should be driving language
acquisition research.
We see at least three ways that language processing can be used in the
service of understanding language acquisition. They vary in the degree to which
they engage with the central questions of the field.
Level 1 approaches: Processing in learners. Research at this level explores
wellknown phenomena from the adult psycholinguistic literature, documents
how they play out in different populations (child learners, adult learners,
bilinguals), and asks whether there are differences. For example, do children
resolve syntactic ambiguities in the same way that adults do? Most existing
research on language processing in learners falls into this category. It addresses
the abilities of language learners, but it does not directly address questions about
learners’ ultimate successes and failures. We survey key findings in this area in
Section 3.
Level 2 approaches: Learning effects as processing effects. Research at this level
draws on insights from adult psycholinguistics to understand the language
proficiency of learners. It asks whether findings that have been attributed to
incomplete grammatical learning should instead be attributed to limitations in
the learner’s language processing system. In Section 4 we explore one area
where findings about adult real time grammatical computation illuminate
otherwise mysterious aspects of children’s learning profiles.
 The role of language processing in language acquisition
411
Level 3 approaches: Explaining learning via processing. Research at this level
seeks to understand what it takes to successfully master the grammar of a
language, and what language processing abilities are needed to reach that level.
It seeks to understand whether language processing limitations might be to
blame for cases of incomplete learning and for critical periods or maturational
constraints. This type of work directly addresses the learning problem. This is
the focus of Section 5, where we discuss the seductive but counter-intuitive idea
that children’s cognitive limitations are the cause of their success (‘less is more’:
Newport, 1990). We argue that this idea is unlikely to be correct in its original
form. But we also suggest that a variant of the idea has some promise, taking
into account some key recent findings about the details of language processing
mechanisms. The key connection that we explore involves a possible link
between predictive processing mechanisms and the learner’s ability to test
detailed hypotheses about the language.
We focus here on the learning of sentence-level grammatical phenomena,
because we think it is the area where the most progress has been made in
understanding child language processing. Sentences are extended in time, so
useful insights about the time course of syntactic processing can be gained using
relatively coarse-grained and child-friendly measures. In other domains of
language, our key questions involve processes that operate on shorter time scales
and develop at earlier ages, when children are less easy to work with.
In order to discuss language processing in learners, we first need to outline
what (we think) we know about the language processing mechanisms that
support grammatical computation in adult native speakers. This population
provides the gold standard for rapid use of grammatical knowledge.

2. Language processing essentials for learners

2.1 Preliminaries: What we’re trying to explain

Language processing research is not all about ways of confusing speakers using
garden path sentences or hard-to-process structures such as object relative
clauses. Much of the language comprehension literature is concerned with
quantifying difficulty (e.g., Gibson, 1998; Levy, 2008). But for our purposes the
goal is not to explain what is easy or hard, but to understand what
representations are constructed in real time, how comprehenders figure out
which representations are possible, and how tightly those representations are
constrained by a speaker’s grammatical knowledge. One tradition in
psycholinguistics assumes that comprehenders build rough-and-ready
representations that bear only a loose relation to what their grammar sanctions
412 Colin Phillips and Lara Ehrenhofer
(e.g., Ferreira & Patson, 2007; Townsend & Bever, 2001). That is worrying for
language learners, if true, as it suggests that learners might systematically
misperceive crucial input for learning. Fortunately, we think that there is good
evidence that comprehenders quickly build grammatically accurate
representations. Many studies show that comprehenders make effective use of
their syntactic and semantic knowledge (e.g., Frazier, 1999; Kazanina et al.,
2007; Phillips, 2006; Traxler & Pickering, 1996; for review see Phillips et al.,
2011). In electrophysiological studies, rapid sensitivity to grammatical errors is
so commonplace that it is news only when speakers fail to quickly notice a
violation of a linguistic constraint (e.g., Wang et al., 2012; Xiang et al., 2009).
We can characterize the representations that are built during comprehension
at multiple grains of analysis. In addition to standard linguistic analyses, we can
describe how the representations are encoded at a finer level of detail (as objects
in short-term memory or as neural encodings), how grammatical information is
used in memory access operations and how we do all of this fast enough to
comprehend words at a rate of 3–5 words per second. These issues are certainly
relevant to learners.

2.2 Accurate parsing

Comprehenders need to assign the intended structure and meaning to sentences

incrementally, taking advantage of the grammatical constraints of the language.
Each word or phrase in a sentence structure forms relations with other words and
phrases in the sentence, e.g., selection/thematic relations, agreement relations,
coreference relations, scope relations. Therefore, each word introduces a set of
requirements that need to be satisfied, either by accessing an item in memory, or
by waiting for some future item in the sentence (agreement relations, thematic
relations, scope relations, focus relations, etc.). These various requirements call
for a comprehension system that can identify the right type of item at the right
location in memory, without being distracted by similar items in structurally
inappropriate locations, and can manage multiple demands simultaneously.
The more we learn about language processing, the clearer it becomes that
successful comprehension requires sophisticated time and resource management.
A new word appears roughly every 200–400 milliseconds, yet the time needed to
carry out sensory, phonological, lexical, syntactic, and semantic processing is
longer than the duration of each word. Hence many different processes must
operate in parallel, without interfering with each other. Hence, parsing is a
highly skilled process that requires rapid switching between sub-tasks.
2.3 Reanalysis

If the comprehender rapidly parses the input, then he will likely need to make
revisions along the way, since language is filled with temporary ambiguities.
 The role of language processing in language acquisition
413
Successful revision requires recognizing the problem, diagnosing how it could
be repaired, inhibiting the initial parse, plus any interpretive commitments that
arose from that parse, and building a new parse consistent with the input. This
task is not always easy: when an initial parse fails, the cue that the parse failed
provides few clues to what went wrong and how to fix it. Reanalysis is
something that adults can often do, but they are not especially good at it (Ferreira
& Henderson, 1991; Sturt, 2007). Even when successful re-parsing occurs, new
beliefs that were created based on the mis-parse are not retracted (Christianson et
al., 2001; Sturt, 2007).
Reanalysis abilities are, in principle, important for language learners, as they
are necessary for correctly apprehending the input. Therefore, when we find that
children are especially bad at reanalysis, this raises the danger that they
systematically misperceive the properties of the language around them. But the
fact that we observe reanalysis difficulties in the lab is no guarantee that they
play an important role in real life. The situations that we concoct in the lab could
be very rare in the wild. An important task for understanding the relation
between language processing and language acquisition is to figure out when and
where reanalysis abilities are needed in everyday language use.

2.4 Prediction

There is a lot of current interest in psycholinguistics in how good adults are at

predicting upcoming material. Our main goal here, however, is to examine how
these abilities might be essential for language learning. Our group’s recent work
has led us to find that we might not be as good as we thought at using rich
linguistic constraints in prediction. We suspect that this might be very relevant
for learning.
Predictive processes encompass category predictions that are very reliable,
e.g., the determiner the signals that a noun is coming soon, and some that are
probabilistic, e.g., the verb whisper is often but not always followed by a
complement clause, and lexical predictions that vary greatly in their reliability,
e.g., the sequence the prize that the athlete … is likely to be continued with the
verb won.
The notion that comprehenders are ‘active’ participants in the parsing
process is best known from research on unbounded dependencies (‘filler-gap’
dependencies; Fodor, 1978; Frazier & Flores D’Arcais, 1989; Phillips & Wagers,
2007; Stowe, 1986). But it is now clear that active comprehension is a pervasive
property of the mature comprehension system (DeLong et al., 2005; Dikker et
al., 2009; Lau et al., 2006; Staub, 2006; van Berkum et al., 2005), possibly
reflecting the engagement of production mechanisms (Federmeier, 2007;
Phillips, 1996; Pickering & Garrod, 2007).
414 Colin Phillips and Lara Ehrenhofer
Predictive mechanisms are likely important for the robustness of language
understanding, such as in noisy settings or when listening to an unfamiliar
accent. Prediction may also play a key role in learning: if a learner accurately
parses and interprets a sentence, then he has evidence that the sentence is
possible in the target language. But if a learner can use his current knowledge of
the language and the context to predict how the sentence will unfold, then the
comparison of what he expects with what actually occurs could provide valuable
additional information. The idea that predictions can provide a key source of
information for language learners has been explored in a number of learning
models (e.g., Chang et al., 2006; Elman, 1993), and we suspect that this area
may be even more fruitful, based on recent findings about the successes and
failures of predictive mechanisms, to which we turn next.

2.5 Fast and slow predictions

We now briefly dive into the weeds of electrophysiology, as we suspect that

some recent findings might hold valuable clues for understanding language
learning mechanisms.
The most widely known generalization about the electrophysiology of
language is very likely wrong. In 1980 Kutas and Hillyard reported that
semantically anomalous sentences like I drink my coffee with cream and socks
elicit a characteristic ERP response that came to be known as the N400 (Kutas &
Hillyard, 1980). Other types of linguistic anomalies did not elicit the same
response. This finding is reliable, but its interpretation is less clear. An early
view was that the N400 reflects the difficulty of combinatorial semantic
interpretation, but this view has fallen out of favor among ERP specialists, based
on a wealth of further evidence.
A widespread view in the current ERP literature is that the N400 reflects
lexical access mechanisms, specifically the ease of recognizing a word in context
(Deacon et al., 2004; Kutas & Federmeier, 2000; Lau et al., 2008; van Berkum,
2009). Studies using magnetoencephalography (MEG) suggest that the N400 is
generated in posterior temporal brain areas that are commonly associated with
lexical processing (Halgren et al., 2002; Helenius et al., 1999; Lau et al., 2008;
Pylkkänen et al., 2007). N400 amplitudes are modulated by variables typically
associated with lexical access, such as word frequency and priming (Federmeier
& Kutas, 1999; Holcomb et al., 2002; Kutas et al., 2006; Lau et al., 2013; van
Petten & Kutas, 1990).
Although many studies have shown that N400 amplitude is tightly linked to
a word’s cloze probability, i.e., the probability that the word is used in that
context in an untimed sentence completion task, some important studies have
shown that N400 amplitudes reflect superficial lexical associations between the
words in a sentence, rather than cloze probability (Fischler et al., 1983; Urbach
 The role of language processing in language acquisition
415
et al., 2008). For example, in the sentence A robin is not a {bird | tree} the word
bird, which is closely associated with robin elicits a smaller N400 than tree, even
though the completion with bird creates a false statement (Fischler et al., 1983).
This further suggests that the N400 reflects the degree to which the current word
is expected in context. But it also raises the puzzle of why lexical expectations
are sometimes based on low-level associative relations, and sometimes based on
more sophisticated use of the semantic and pragmatic context (Nieuwland &
Kuperberg, 2008). Recent work in our group, led by Wing Yee Chow and Shota
Momma, points towards an answer to this puzzle.
Many different sources of information affect a word’s cloze probability in
context: word associations, argument roles, discourse context, etc., and we know
little about how these are combined to generate predictions about upcoming
words. We have been able to isolate the contribution of specific predictive cues,
and to determine how quickly they can be exploited, using verb-final sentences
in English, Chinese, and Japanese. In (1a–b) the sentence-final verb is highly
unlikely in both sentence contexts, but for different reasons. In (1a) the verb is
unlikely simply because evicting events don’t involve landlords and cats. The
thematic roles do not matter. In (1b) the verb is unlikely specifically because of
the thematic roles of the two arguments, which are the reverse of their canonical
configuration. We found that these two constraints had different impacts on
N400 amplitudes. When we manipulated cloze probability via lexical
associations, as in (1a), this had strong and immediate effects on N400
amplitudes. But when we manipulated cloze probability via argument role
reversals, as in (1b), there was no effect on N400s (Chow et al., 2015).
(1) a. Which cat did the landlord evict?
b. Which landlord did the tenant evict?

In other studies, using Mandarin Chinese, we found that cloze manipulations

based on role reversals do affect N400s, but only when there is additional time
between the arguments and the verb (Chow et al., submitted). We found a
similar effect in Japanese: manipulating case information on nouns makes a big
difference to the plausibility and probability of the verb, e.g., canonical (2a) vs.
reversed (2c), or canonical (2b) vs. reversed (2d). But the case manipulation had
no impact on the N400 to a lexically related verb when the verb closely followed
the noun. The case manipulation did elicit a clear N400 effect when the verb was
delayed by an additional 400ms (Momma et al., 2015).
(2) a. hachi-ga sasu b. sakana-o tsuru Canonical
[Link] sting [Link] catch

c. hachi-o sasu d. sakana-ga tsuru ‘Reversed’

[Link] sting [Link] catch
416 Colin Phillips and Lara Ehrenhofer
Importantly, in all of these studies speakers consistently noticed the anomaly, as
reflected in P600 effects. This shows that they did notice the problem, but failed
to quickly take account of the argument role information in anticipating likely
and unlikely upcoming verbs.
These results suggest that different cues affect linguistic predictions on
different time scales, and that these different effects can be studied. One
suggestion for why argument role information has delayed effects on prediction
is that it is difficult to directly probe memory representations based on complex
cues, involving either multiple cues or relational cues (such as “agent of”). For
example, the memory query “what type of events involve landlords?” might be
easier to carry out than the memory query “what type of events involve landlords
as patients?” Therefore it is probably too coarse-grained to ask whether a given
learner group is good or bad at linguistic prediction. This is relevant to the
broader concerns of this article, since the ability to rapidly generate predictions
based on complex cues may be important for learning about complex
contingencies in language.

2.6 Learning relevance: Input, intake, and missed opportunities

Based on this brief survey, we can imagine at least 3 ways that a learner’s
language processing abilities could aid or hinder learning. Each of these is
relevant to the examples that follow.
First, a learner could simply fail to assign a parse to an input sentence,
perhaps because it is too complex or arrives too quickly. This would slow but not
actively impede learning, as it would not lead the learner to incorrect
generalizations.
Second, a learner could systematically mis-parse a certain type of input
sentence, e.g., due to parsing biases and/or reanalysis failure. This type of failure
is potentially more disruptive, as it could lead a learner to see evidence for
structural generalizations that are not present in the target language. There is a
long tradition of worrying about how informative the input is to learners
(‘poverty of the stimulus’ arguments: Berwick et al., 2011; Lidz et al., 2003;
Perfors et al., 2011; Phillips, 2013; Pullum & Scholz, 2002) and whether the
input contains misleading information due to speech errors (Newport et al.,
1977). But even if the input is both informative and impeccable, there is a danger
that learners might themselves introduce errors, as their ‘intake’ might diverge in
systematic ways from the input (cf. Omaki, 2010; Omaki & Lidz, 2014).
Third, a learner who successfully parses an input sentence could extract
more or less information from that sentence, depending on his ability to predict
ahead of time how the sentence will unfold. A learner who can make more
sophisticated predictions about what will be said when may be able to learn more
complex contingencies about the language.
 The role of language processing in language acquisition
417
3. Level 1 accounts: Processing in learners

Research at our first level of analysis takes findings about language processing
in adult native speakers and asks whether child or adult learners perform
similarly. This research does not directly address learning questions, but it is a
critical first step in understanding the relevance of language processing for
learning.

3.1 Analysis and reanalysis in children

We know surprisingly little about children’s ability to assign detailed and

accurate analyses to incoming sentences. Our main source of evidence is the
meanings they assign to sentences they hear. But although children do relatively
well at understanding intended meanings, we know less about how they arrive at
these meanings, and whether they use the same syntactic and semantic
representations as adults. Comparing children’s interpretations of ambiguity —
both true and fake — lets us ask whether children access all and only the
meanings that adults allow.
‘Fake’ ambiguities are cases where the adult grammar allows only one
interpretation, but where an alternative interpretation might be available if the
child violates a grammatical constraint. Many studies suggest that preschoolers
are good at using grammatical constraints to avoid fake ambiguities (e.g.,
Conroy et al., 2009; Crain & Thornton, 1998; de Villiers & Roeper, 1995). But
we find other cases where children seem to violate simple grammatical
constraints to arrive at non-adultlike interpretations. For example, studies on
children’s interpretation of non-canonical word orders suggest that they might
systematically ignore casemarker cues, e.g., reversing agent and patient roles in
[Link] chased [Link] (MacWhinney et al., 1985), or misinterpreting passive
sentences (Bever, 1970; Huang et al., 2013; Maratsos et al., 1985; Pinker et al.,
1987).
In the domain of true ambiguities, much attention has been devoted to simple
prepositional phrase (PP) attachment ambiguities, as in Put the frog on the
napkin in the box, where the PP on the napkin could be understood as the goal
argument of the verb put or as a modifier of the direct object the frog. Adults are
biased toward the argument parse of the PP (Rayner et al., 1983; Tanenhaus et
al., 1995). Children show the same bias, but more strongly (Trueswell et al.,
1999). Children are able to access the alternative, modifier parse of the PP under
some circumstances (Snedeker & Trueswell, 2004), but they are less able than
adults to integrate multiple cues that might lead them to that alternative parse
(Engelhardt, 2014; Hurewitz et al., 2000; Snedeker & Trueswell, 2004; Weighall
et al., 2008).
418 Colin Phillips and Lara Ehrenhofer
In studies on ambiguous wh-questions like (3), we found that children show
the same bias as adults to construe the wh-phrase where with the first verb
(Omaki et al., 2014). The bias is based on order/timing, rather than on structure
or plausibility, as shown by the fact that preferences reversed between English
(3a) and Japanese (3b), which have very different word orders. English speakers
took the question to be about the location of telling, Japanese speakers took it to
be about the location of catching. Children’s bias for the first-verb interpretation
was so strong that we worried that they might systematically fail to notice the
alternative parse, leading to a mistaken view of the grammar of their language. 1
Omaki (2010, Chapter 6) examines this danger in detail: he examines a corpus of
child-directed wh-questions and asks whether the distributions in children’s
input would be seriously distorted once filtered through a child parser (the
‘intake’). He concludes that distortion would indeed occur, but that the truly
ambiguous input sentences are rare enough that the child would be saved by
unambiguous evidence. But such arguments need to be assessed on a case-by-
case basis, in many linguistic domains.
(3) a. Where did Lizzie tell somebody that she was going to catch butterflies?
b. Doko-de Yukiko-chan-wa chouchou-o tsukamaeru-to itteta-no? where-at
Yukiko-dim-top butterfly-acc catch-comp was-telling-Q
“Where was Yukiko telling someone that she will catch butterflies?” In more
subtle cases of true or fake scope ambiguity, there are concerns that children
may undershoot or overshoot what the target grammar allows. In cases of true
scope ambiguity children typically show strong biases for one interpretation,
generally the same interpretation that adults favor (Lidz & Musolino, 2002),
though not always (Goro & Akiba, 2004; Unsworth et al., 2008), though these
biases can be overcome with strong contextual support (Gualmini, 2008; Lidz,
2014; Viau et al., 2010). More worryingly, we find cases where children allow
scope interpretations that are disallowed in the adult language (Goro, 2007).
These findings are especially troubling for learning theories, as they suggest that
children could mistakenly perceive evidence for grammatical possibilities that
are not available in the target language.
The clearest finding about child sentence processing is that children are very
poor at reanalysis (Trueswell et al., 1999). Children’s difficulty with reanalysis
has been confirmed many times in subsequent research (Choi & Trueswell,
2010; Kidd et al., 2011; Lassotta et al., submitted; Weighall, 2008). Therefore, a
rather robust generalization about child parsing is that children’s first
interpretation is their only interpretation. (This is also a convenient property of

1 . De Villiers et al. (2008) report apparently conflicting results from a large sample of
around 1000 English-speaking children, who show a strong bias for a long-distance construal
of the question How did the boy say he hurt himself? However, this finding is based on a
single test sentence, and Omaki et al. (2014) show that the bias is specific to the use of bare
say as the main clause verb.
 The role of language processing in language acquisition
419
children’s parsing, as it sometimes allows researchers to learn about children’s
initial parsing choices using simple off-line measures.)
It is widely assumed that children’s difficulty with reanalysis is not a
languagespecific limitation, but instead reflects more general limitations in
children’s cognitive control abilities (Novick et al., 2005; Mazuka et al., 2009).
Children show broad difficulties in revising initial action plans, likely due to the
delayed maturation of the frontal lobe, the primary home of executive function
abilities (Davidson et al., 2006). As such, this aspect of children’s sentence
processing abilities is maturational in nature, and should be unrelated to
language proficiency in general.
For detailed reviews of child sentence processing research at this level, see
Snedeker (2013) and Omaki & Lidz (2014).

3.2 Analysis and reanalysis in L2ers

We can ask the same questions about adult learners that we asked about children:
do they analyze incoming sentences accurately, and how well can they recover
from misanalyses? Both of these are relevant to the question of what evidence
learners actually perceive (‘intake’) about the target language.
Reanalysis appears to be less of a problem for adult learners than for
children, consistent with the view that children’s reanalysis difficulties are
related to language-independent maturational constraints. We know more about
the accuracy of adults’ than children’s parses of input sentences, though the
findings are complicated by the fact that adult learners are more heterogeneous
than child learners.
Our knowledge of how accurately L2 comprehenders parse incoming
sentences owes much to the Shallow Structure Hypothesis (Clahsen & Felser,
2006), which argues that L2 learners rely on lexical and conceptual associations
to guide interpretation, rather than using structural cues to build detailed and
accurate syntactic representations. SSH struck a nerve in the L2 community,
which succeeded in eliciting a host of supporting and contradictory evidence.
The evidence for or against SSH generally involves demonstrations of L2ers’
sensitivity or insensitivity to structural cues that native speakers attend to in
parsing, rather than evidence that L2ers are more strongly guided by lexical and
conceptual associations. For example, in the domain of wh-dependency
formation, Marinis et al. (2005) argued that L2ers do not process long-distance
filler-gap dependencies with the same profile found in native speakers (Gibson
& Warren, 2004). One limitation of findings like these is that it is relatively easy
to fail to replicate L1 findings in the noisier data of L2ers, especially when those
findings depend on statistical interactions. Arguing against this, Omaki and
Schulz (2011) and Aldwayan et al. (2010) claim that L2ers parse wh-
420 Colin Phillips and Lara Ehrenhofer
dependencies in a nativelike fashion, based on online sensitivity to island
constraints. (Felser et al. (2012) present similar findings, though they arrive at a
different conclusion.) Other studies have made similar arguments involving
L2ers’ sensitivity to constraints on anaphora (Rodriguez, 2008). In the domain of
case and agreement processing, findings are highly varied, ranging from the
claim that L2ers are highly insensitive to inflectional information (Jackson &
Dussias, 2009; Jiang, 2004) to the claim that they can achieve native-like
proficiency (Hopp, 2010).
Less is known about adult learners’ susceptibility to grammatically illicit
dependencies during parsing. In subject-verb agreement, where native speakers
show fleeting interference effects (‘attraction’) from structurally inappropriate
nouns, e.g., the errors that classification software make, it seems that L2
comprehenders are no less susceptible to error (Tanner et al., 2012). In closely
related domains where native speakers fail to show attraction effects, such as the
processing of reflexive pronouns (Dillon et al., 2013), there is some evidence
that L2ers do show attraction effects (Felser & Cunnings, 2012). This suggests
that L2ers may be more susceptible to mis-parses than native speakers, but this is
less of a concern than it is for children (see Section 3), because L2ers are better
reanalyzers than children. When mis-parses occur, they are probably less
disastrous for adult L2 learners than for child L1 learners: there is little evidence
in the adult L2 parsing literature that reanalysis poses special problems for
learners (Hopp, 2015; Jacob & Felser, 2015; Pozzan & Trueswell, 2013; Roberts
& Felser, 2011), and we presume that reanalysis difficulties therefore have less
of an impact in L2 parsing than in child learners.

3.3 Prediction in children

Predictive mechanisms are most useful for learners if they are able to act on
diverse predictive cues, preferably combinations of cues. Predictions are most
useful when they are generated in advance of the external input. Therefore, in
assessing what is known about prediction mechanisms in child and adult
learners, we need to look beyond the binary question of whether learners are able
to use predictions, to the specifics of what they are able to do, and how quickly.
In these terms, relatively little is known at present. For child and adult learners
alike, there is evidence of some facilitation from prediction, but also evidence
that they are not as good as native speaking adults. This is unsurprising.
A number of studies with children have used demonstrations of anticipatory
looks in a simple visual scene as evidence of predictive linguistic processing.
This includes studies where verbs trigger looks to likely direct objects in 2–10-
year-olds (Borovsky et al., 2012; Mani & Huttig, 2012; Nation et al., 2003), and
where gendermarked determiners in Spanish facilitate looks to pictured objects
in 3-year-olds (Lew-Williams & Fernald, 2007). However, these involve very
 The role of language processing in language acquisition
421
simple relations in simple situations containing a small number of visual targets,
and so we should be cautious in generalizing from these findings.
In a tougher test of predictive processing involving wh-dependencies, 5-
year- olds lagged behind adults (Atkinson et al., 2013). Following a story,
children listened to wh-questions such as Can you tell me what Emily was eating
a cake with?, or yes-no questions such as Can you tell me if Emily was eating
the cake with a fork? Although the wh-question queried the instrument, the main
interest in the study was the looks to the direct object (the image of the cake)
upon hearing the verb eating. Increased fixations to the cake in the wh-question
relative to the yes-no question provided evidence of initial mis-interpretation of
the question as a direct object question. Adults showed this pattern at the verb,
but children did not. This suggests that although children have basic predictive
abilities at an early age, they are less good at acting quickly on more complex
cues. In the wh-question setting, the predictive fixations depend on retrieving the
wh-phrase, combining its interpretation with the verb and then directing looks to
relevant objects in the display.
Clearly, we still know relatively little about children’s predictive abilities.
Much more needs to be learned in this area, especially if we hope to link these
abilities to hypothesis testing and recovery from overgeneralization in learning.

3.4 Prediction in L2ers

There is little work that directly compares prediction in child and adult learners.
It is easier to directly compare adult learners with adult native speakers (for
review see Kaan, 2014).
In a follow-up to their study with Spanish-speaking children, Lew-Williams
and Fernald (2010) found that adult learners of Spanish were unable to use
gender on a determiner to anticipate an upcoming noun; this limitation persisted
in highly proficient Spanish learners (Grüter et al., 2012). This suggests a
possible case where child parsing abilities might outstrip adults. However,
caution is needed, as other studies have found evidence that gender can be used
as a predictive cue by L2 comprehenders (Dussias et al., 2013; Hopp, 2013), and
it is hard to know whether the observed difficulties reflect differences in
predictive mechanisms or differences in the way that grammatical gender is
represented by child and adult learners.
An ERP study by Martin et al. (2013) also found evidence of reduced
predictive processing in adult learners, using a design adapted from DeLong et
al. (2005). A sentence like It was a windy day so the boy decided to go out to
fly… creates a strong expectation for a continuation with the noun kite.
Therefore the determiner a is more expected than the determiner an (as in … an
airplane). DeLong and colleagues showed that adult native speakers detect
422 Colin Phillips and Lara Ehrenhofer
evidence against their prediction already at the determiner an. But Martin and
colleagues found no corresponding sensitivity to the mismatch at the determiner
in L2 learners of English, despite their flawless knowledge of the determiner
forms. A subsequent study by the same group using gender marking in learners
of Spanish suggests that L2ers may be better able to exploit predictive cues that
also exist in their native language (Foucart et al., 2014).
We already saw that children’s predictive processes fell short of adults’ in
processing wh-questions, as reflected in a lack of active wh-dependency
formation effects (Atkinson et al., 2013). We know of no direct comparison with
L2 speakers, but a number of L1-L2 comparisons of active wh-dependency
formation in adults suggest that L2ers process wh-dependencies in an adultlike
fashion (Omaki & Schulz, 2011). Therefore, we find evidence of limitations in
L2ers’ prediction abilities, but we do not find clear evidence that L2ers lag
behind children in their prediction abilities.

3.5 Summary

Level 1 studies are most informative about learning when they provide clues to
what information learners can extract from the sentences that they hear. For this
reason, we focused primarily on the questions of whether learners build accurate
analyses of incoming sentences, and whether they have the predictive abilities
that could help them to extract additional information from the input. The
clearest finding is that children are not good at reanalysis, and this makes it more
important that their initial parses be correct. We find mixed evidence on the
grammatical accuracy of children’s and adults’ parses. Some evidence suggests
that they do very well, while other evidence points to serious risk of misanalysis.
It is not yet clear how often the parsing failures found in the lab occur in real-life
situations. Meanwhile, prediction is poorly understood in learners, in part
because it is treated monolithically. Learners can exploit some predictive
information, but they struggle in more complex cases.
The clearest generalization from the Level 1 studies should be the following:
children’s language processing abilities give them no obvious advantage over
adult learners.

4. Level 2 accounts: Learning effects as processing effects

In this section we present a more extended example of a Level 2 account, which

uses sentence processing findings to explain phenomena typically attributed to
stages in children’s grammatical development. We review striking parallels
between independent literatures in language acquisition and language
processing, though the effects operate on different time scales. We found that
 The role of language processing in language acquisition
423
children’s nonadultlike interpretations of anaphors correspond to interpretations
that adults fleetingly consider but then reject during on-line comprehension. This
fits with the generalization from Section 3 that adults’ first interpretation is
children’s only interpretation.

4.1 Starting point: Universal vs. language-specific constraints on backwards

anaphora
Our starting point is a surprising finding about an under-studied constraint on
anaphora in Russian, which creates a minimal contrast with English. English
allows backwards anaphora (4a), but not when the pronoun c-commands the
name (4b). The Russian counterparts are both unacceptable. (4a) and (5b) reflect
a crosslinguistically robust constraint, Principle C, which prevents a pronoun
from taking an item that it c-commands as its antecedent (Chomsky, 1981). In
contrast, (5a) is ruled out by a more specific constraint that applies only to
specific combinations of subordinators, aspect, and thematic roles (Avrutin &
Reuland, 2004; Kazanina, 2005). We refer to the Russian constraint as the poka-
constraint, based on the subordinator that most strongly gives rise to the effect.2

(4) a. While hei was reading the book, Poohi ate an apple.
b. * Hei ate the apple while Poohi was reading the book.

(5) a. * Poka oni chital knigu, Poohi s’el yabloko.

while he was [Link] the book Pooh [Link] the apple
b. * Oni s’el yabloko, poka Poohi chital knigu.
he [Link] the apple while Pooh was [Link] the book
Young children show mastery of the contrast as early as they can be tested,
using Truth Value Judgment Tasks (TVJT) in 3–6-year-olds (Crain & McKee,
1985; Crain & Thornton, 1998), and preferential looking tasks in children as
young as 30 months (Lukyanenko, Conroy, & Lidz, 2014). This has been offered
as compelling evidence that Principle C, a cross-linguistically robust constraint,
is known innately.
Nina Kazanina tested Russian children’s interpretations of sentences like (5)
(Kazanina & Phillips, 2001). Her findings were clear: 6-year-olds performed
similarly to adults, disallowing coreference in both sentence types. But the
youngest Russian children, aged 3 years, performed remarkably similarly to the
Englishspeaking 3-year-olds in Crain & McKee’s study, allowing coreference in
(5a), but not in (5b). 4–5-year-olds showed an intermediate profile.
These findings have an attractive interpretation in terms of grammatical
development: children master universal constraints at a very early age, but
2 . Related effects can be seen in null subject languages, reflecting constraints on the use of
overt pronouns. But Russian is not a null subject language.
424 Colin Phillips and Lara Ehrenhofer
languageparticular constraints develop later. But we were also troubled by how
Russian children could ever learn the constraint, and why it should take them 5–
6 years to do so. Deficiencies in children’s knowledge about verbal aspect
appeared not to be the problem (Kazanina & Phillips, 2007). Possibly, the poka-
constraint is learned late because the relevant evidence in the input is very
sparse, but if so, we should encounter a lot of variability in when children master
the constraint: some children might take 6 years to hear enough relevant
evidence, others might never hear enough. Second, we were troubled by the fact
that Russian children need to move from a more permissive to a less permissive
grammar: they need to somehow figure out that coreference is not actually
possible in sentences like (4a). The answer to these questions became clearer
once we learned more about how adults process these constructions.

4.2 Principle B (pronouns) in children

Principle B effects in children present an interestingly different puzzle.

Informally, Principle B prevents pronouns from taking an antecedent in the same
clause. One of the most impressive cases of developmental findings influencing
theoretical linguists is the finding that preschoolers allow interpretations that
violate Principle B in sentences with referential antecedents (6a), but adhere to
Principle B in sentences with quantificational antecedents (6b) (Chien & Wexler,
1990; Thornton & Wexler, 1998).
(6) a. Mama Bear washed her.
b. Every bear washed her.

Our interest in this area was triggered by the claim that the Quantificational
Asymmetry in the Delay of Principle B Effect (DPBE) is an experimental
artifact, due to poorly matched materials (Elbourne, 2005). We designed new
experiments to try to respond to this challenge, but the results were surprising.
Once we more closely matched the quantificational and referential conditions,
the quantificational asymmetry did indeed go away. But, to our surprise, our
improvements also got rid of the DPBE. Children allowed non-adultlike
interpretations in only around 12% of trials (Conroy et al., 2009).
Puzzled by the fact that children in our study had performed so well, we
analyzed around 30 previous studies on DPBE, revealing an interesting picture.
Across studies, there was wide variation in the proportion of non-adultlike
interpretations, ranging from 10% (Kaufman, 1988) to a high of 82% (McKee,
1992). This variability lay beyond measurement error, and it appeared to be
linked to experimental design issues. Experiments that made a non-adultlike
interpretation less accessible tended to elicit higher error rates from children.
Importantly, instead of concluding that we had hit upon a great design that
revealed children’s true abilities, we took the variability to indicate that in
 The role of language processing in language acquisition
425
applying Principle B, children are very fragile. This is especially striking in
McKee (1992), a study with one of the highest rates of non-adultlike
interpretations in the DPBE literature. Tellingly, in that experiment, the pronoun
interpretation that would reflect adherence to Principle B was barely relevant in
the example scenario. Most interestingly, when McKee re-ran the experiment in
Italian, using clitic pronouns instead of full pronouns, children allowed the non-
adultlike interpretation in only 15% of trials. This is consistent with other
evidence that children do not show a DPBE with clitic pronouns (Baauw et al.,
1997; Varlokosta, 2000). So, whereas children show great fragility in
implementing Principle B with full pronouns, they showed very little fragility in
implementing Principle B with clitic pronouns, a fragility not found in the
literature on children’s mastery of Principle C. In all three cases we see evidence
that the adult constraint is in place, but children’s fragility in applying this
knowledge is quite selective. This is reminiscent of a similar profile that we have
observed in our studies on adult parsing.

4.3 Principle C effects in adult parsing

In independent work with adults, we were interested in the effects of (preferably

obscure) grammatical constraints on real-time language comprehension,
motivated by an interest in the nature of grammatical knowledge and its relation
to moment-by-moment processes (Phillips, 1996; Phillips & Lewis, 2013; Lewis
& Phillips, 2015). Principle C turns out to be a parade case of grammatical
faithfulness in language processing.
We can test the online effects of Principle C by creating instances of
backwards anaphora in English, and asking whether comprehenders temporarily
consider an interpretation where the pronoun corefers with an NP that should be
excluded by Principle C.3 We can test where people attempt coreference using a
gender mismatch paradigm: in any structural position where speakers attempt
coreference, manipulation of the gender match between the pronoun and the
potential antecedent should lead to processing disruption (van Gompel &
Liversedge, 2003). We found gender mismatch effects in positions that satisfy
the constraint (7a), but not in positions that violate the constraint (7b). We found
the same contrast across three self-paced reading experiments involving different
syntactic configurations (Kazanina et al., 2007), and the finding appears to be
3 . Some readers may be concerned that Principle C is a tenuous or epiphenomenal
constraint (Ambridge et al., 2014). Objections come from cases where instances of
supposedly acceptable backwards anaphora are rated as little better than Principle C violations
(Gordon & Hendrick, 1998), or from cases where apparent Principle C violations are clearly
acceptable (Harris & Bates, 2002). We do not think that these facts undermine the
grammatical generalization, though they certainly illustrate the importance of careful
discourse controls in studies of cataphora. See Kazanina (2005) for detailed discussion.
426 Colin Phillips and Lara Ehrenhofer
robust across languages and labs (Aoshima et al., 2009; Fedele & Kaiser, 2014;
Pablos et al., submitted; Rodriguez, 2008; Yoshida et al., 2014).
(7) a. While she was taking classes full time, Kathryn|Russell
was working two jobs…
b. She was taking classes full time while Kathryn|Russell was
working two jobs…

Interpretations that violate Principle C appear to simply not occur to people,

even when we probe using time-sensitive measures. In more technical terms:
interpretations that violate Principle C are not merely illicit, they are beyond the
generative capacity of the language system.
The child-adult parallel that we saw with Principle C is just the first among a
list of similar parallels that we have observed.

4.4 Principle A (local reflexives)

4–5-year-old children perform well in tests of the locality constraint on

reflexives like herself or themselves, which requires (roughly) that a reflexive
find an antecedent in the same clause (Principle A: Chomsky, 1981). Some
studies have shown relatively high rates of illicit interpretations (Chien &
Wexler, 1990), but those errors reflect a failure to attend to gender marking
(himself, herself). Once that problem is taken care of, children perform quite well
(Zukowski et al., 2008).
We also know a lot about the real-time effects of Principle A in adults,
thanks to a recent debate about the generality of “retrieval interference effects”
in parsing (Dillon, 2014). We know that speakers frequently make errors in
producing subject-verb agreement, incorrectly agreeing with a structurally
inappropriate noun (‘agreement attraction’), as in (8) (Bock & Miller, 1991;
Eberhard et al., 2005; Vigliocco & Nicol, 1998). In comprehension, speakers fail
to notice the same agreement errors, which has been attributed to noisy memory
retrieval mechanisms: in attempting to retrieve a feature-matching noun from the
subject position, comprehenders may mistakenly retrieve a feature-matching
noun from a different position (Wagers et al., 2009). Interpreting a reflexive
should be similar, requiring retrieval of a feature-matching noun from the subject
position. In general, however, we do not find corresponding ‘reflexive attraction’
effects. When agreement and reflexive attraction are carefully matched (9), this
gives rise to strong agreement attraction effects, but no reflexive attraction
effects (Dillon et al., 2013). This is consistent with numerous other findings of a
lack of attraction effects with reflexives (Clackson et al., 2011; Clifton et al.,
1999; Nicol & Swinney, 1989; Sturt, 2003; Xiang et al., 2009).4

4 . Here we focus on facilitatory attraction effects in ungrammatical sentences, which are

consistently found in agreement processing, and which provide the clearest evidence of
 The role of language processing in language acquisition
427
(8) a. * The key to the cabinet(s) are on the table.
b. * The runner(s) who the driver see every day never forget to wave.
(9) * The new manager [who oversaw the middle manager(s)] … …
apparently were dishonest about the company’s profits.
… apparently doubted themselves on most major decisions.

Principle A has parallel effects in children and adults: children successfully

avoid illicit interpretations in off-line tasks of acceptability and interpretation;
adults successfully avoid illicit interpretations in on-line measures.

4.5 Principle B (pronouns) in adults

We already saw that preschool-aged children seem to know Principle B, but

show extreme fragility in implementing the constraint, and we find something
similar in adult on-line processing. Well, almost. As in children, studies on
Principle B in adult parsing certainly show varied and conflicting results. Adults
successfully avoid pronoun interpretations that violate Principle B (Clifton et al.,
1997; Lee & Williams, 2006; Nicol & Swinney, 1989), but in other studies they
do not (Kennison, 2003; Runner et al., 2006). The best known of these studies
(Badecker & Straub, 2002) used a variant of the gender match paradigm outlined
above, testing whether the presence of two nouns that matched the gender of the
pronoun would lead to competition and hence slowdown in reading times, even
when one of the nouns is a clausemate of the pronoun, making it an illicit
antecedent (10). They found clear slowdowns (“multiple match effect”) in their
first two experiments, and concluded from this that Principle B does not remove
potential antecedents from consideration during parsing.
(10) John thought that {Bill | Beth} owed him another chance to solve the problem.

Based on these past results, it appears that Principle B neatly fits our pattern.
Children’s off-line interpretations show sensitivity to the constraint, but in a
fragile fashion. Adults on-line interpretations show similar fragility. Our main
reservation here is that studies in our group attempted to replicate and
manipulate the presence of fleeting on-line Principle B violations, with no
success. Across seven studies, including one that was a direct copy of Badecker

retrieval of an inappropriate antecedent. We know of only two studies that found facilitatory
effects with reflexives. Cunnings & Felser (2013) report such effects in a low memory-span
subgroup of participants. Parker & Phillips (2014) replicated the standard no-attraction
finding in materials where the structurally appropriate NP mismatches the reflexive in just one
feature, but found strong facilitatory attraction effects when the structurally appropriate NP
mismatched the reflexive in two features, e.g., gender + number. Some studies have found
small inhibitory interference effects in grammatical sentences (Chen et al., 2012; Kush &
Phillips, 2014; Patil et al., 2011).
428 Colin Phillips and Lara Ehrenhofer
& Straub’s materials and procedures, we consistently failed to find the multiple
match effect that would suggest on-line Principle B violations (Chow, Lewis, &
Phillips, 2014). In light of this difficulty in eliciting the fleeting Principle B
violations, we must proceed with caution. But, looking at previous results, it
seems that adults and children consider the same illicit interpretations, except
that in adults those interpretations are evident only in rapid on-line measures.

4.6 Russian Backwards Anaphora (again)

Given the child-adult parallels in other constraints on anaphora, it is worth

returning to the Russian poka-constraint, to find out how Russian adults process
the constructions where young children showed non-adultlike interpretations.
Our study places the child findings in a new light (Kazanina & Phillips, 2010).
Russian adults read sentences in which a potential antecedent for a pronoun
appeared in three different configurations: one where coreference is possible for
adults (11a), one where coreference is impossible due to Principle C (11b), and
one where coreference is impossible due to the poka-constraint (11c).
(11) a. Posle togo kak oni procital knigu, Ivani s’el jabloko. [no constraint] after he
read book Ivan ate apple
“After he read the book, Ivan ate an apple.”
b. Oni cital knigu, poka Ivani el jabloko. [Principle C]
he read book while Ivan ate apple

“He read a book while Ivan ate an apple.”

c. Poka oni cital knigu, Ivani s’el jabloko. [poka-constraint]

while he read book Ivan ate apple

“While he read the book Ivan ate an apple.”

The reading profiles for (11a–b) were as in English: there was a gender
mismatch effect in (11a), reflecting an attempt to resolve coreference, but no
corresponding effect in (11b), indicating that Russians do not attempt
interpretations that violate Principle C. But in (11c), where Russian adults also
disallow coreference, we found a (delayed) gender match effect, i.e., slower
reading times when the gender of the name matched the pronoun. We take this to
mean that Russian adults do initially consider the illicit coreference relation, and
that the effect of the poka-constraint kicks in later, leading to disruption in the
gender-match condition.
If Russian adults temporarily consider interpretations that violate the
pokaconstraint, and then quickly inhibit them, perhaps young Russian children
do the same thing, except that they fail to inhibit the illicit interpretation. This
would suggest that the developmental pattern that we saw from ages 3–6 might
 The role of language processing in language acquisition
429
reflect changes in the children’s ability to inhibit inappropriate parses, rather
than changes in their grammar per se.

4.7 Additional antecedent effects in reconstruction (reflexives)

The final stop in this survey involves a case where children’s interpretations
diverge from adults in a way that it is hard to explain unless we invoke language
processing limitations. We have already seen that English-speaking children and
adults implement the locality constraint on reflexives rather well. An interesting
twist on the locality constraint arises when a reflexive is embedded inside a
larger phrase, such as a noun phrase (12) or a predicate phrase (13). When the
phrase containing the reflexive occupies its canonical position, interpretive
possibilities are exactly as expected: only a clausemate antecedent is possible
(12a, 13a). But a surprising contrast emerges when the phrase containing the
reflexive undergoes wh-fronting. NP-fronting creates an additional antecedent
for the reflexive (12b). It can still be interpreted as referring to the subject of the
embedded clause (the ‘reconstruction’ interpretation), but it now can also be
understood as referring to the subject of the main clause. But fronting a predicate
does not have the same effect: in (13b) the fronted reflexive still only allows the
embedded clause subject as its antecedent. The contrast is interesting fodder for
linguistic analysis (Heycock, 1995; Huang, 1993), but the effects in language
acquisition are even more striking.
(12) a. Johni knew that Billj saw the picture of himself*i,j.
b. Johni knew which picture of himselfi,j Billj saw.

(13) a. Johni knew that Billj was very proud of himself*i,j.

b. Johni knew how proud of himself*i,j Billj was.

In a TVJT using sentences like (14) Leddon and Lidz (2006) found that children
gave adultlike interpretations, with one exception. In (14b) they correctly
allowed only the embedded clause antecedent, and in (14a) they also allowed the
main clause antecedent. But in (14a) they surprisingly rejected the embedded
clause antecedent, which we might expect to be the more ‘basic’ interpretation.
Since it is unlikely that those children have a grammar in which reflexives
disallow local antecedents, the most natural interpretation of these findings is
that children adopt the first available grammatical interpretation of the reflexive
and stick with it, even when additional antecedents appear. In the NP-fronting
sentence (14a) the first grammatical interpretation of the reflexive takes the main
clause subject as an antecedent. In the predicate-fronting sentence (14b) the main
clause subject is not a grammatical antecedent, and therefore the subordinate
clause subject is the first antecedent that they consider.
430 Colin Phillips and Lara Ehrenhofer
(14) a. Miss Cruella figured out which picture of herself Janie put up.
b. Mr Monkey figured out how proud of himself Andy was.

Omaki et al. (2007) probed adults’ word-by-word interpretation of similar

sentences and found that adult parsing respects the contrast between NP fronting
and predicate fronting. That is perhaps unsurprising in light of the child findings,
but it is not so straightforward to see the effect of the contrast. Using a gender
mismatch paradigm, it is easy to show that upon reaching the fronted reflexive in
(12b) adults immediately interpret it as coreferring with the main clause subject,
since that is the only potential antecedent at that moment in time. However, in
our initial study we found a similar gender mismatch effect in both the NP-
fronting and predicate-fronting conditions, suggesting that on-line interpretation
ignored the contrast in acceptability judgments. Further investigation revealed
that adult comprehenders were not ignoring the grammatical constraint, but were
making full use of available possibilities.
It is, in fact, grammatically possible for a reflexive in a fronted predicate to
corefer with the main clause, because the reflexive may take the embedded
clause subject as its antecedent, and that subject could be a pronoun that corefers
with the main clause subject (15).
(15) Johni figured out how proud of himselfi hei must have seemed to hisi boss.
Therefore, comprehenders who encounter the reflexive in (15) might be willing
to interpret it as coreferring to the main clause subject, because of the possibility
of a subsequent pronoun, and this might lead them to anticipate a pronoun. The
same expectation for a pronoun should not arise in NP-fronting examples like
(12b), where the reflexive can directly link to the main clause subject. We tested
this in two further experiments. In the first, we simply gave participants sentence
fragments that extended as far as the reflexive, e.g., John knew {which picture of
| how proud of } himself…, and asked them to provide a written completion.
Pronoun completions were far more common in the predicate-fronting conditions
(82%) than in the NP-fronting conditions (50%). The second experiment used
self-paced reading to test what comprehenders expect to see after they have
already encountered a fronted reflexive that matches the main clause subject. In
NP-fronting conditions (16a) they should have no specific expectations, and we
found no gender mismatch effect to the embedded subject. But in predicate-
fronting conditions (16b) they should expect to encounter a subsequent matching
noun, and we found a gender mismatch effect following the embedded clause
subject.
(16) a. {Patrick | Rachel} found out which story about {himself | herself} the
alcoholic …
b. {Patrick | Rachel} knew how proud of {himself | herself} the
alcoholic … Thus, children show knowledge of the contrast between
 The role of language processing in language acquisition
431
argument fronting and predicate fronting, but they exhibit one
surprising gap in the interpretations that they allow. That gap is
naturally linked to independent limitations in children’s ability to revise
initial parses. Meanwhile, adult on-line processing also shows evidence
of the contrast between argument fronting and predicate fronting.

4.8 Summary: Adult on-line studies reveal the source of child errors

The moral of this section is that looking at real-time processes in adult native
speakers reveals a lot about learners’ linguistic abilities. Using five constraints
on anaphora as a test case, we saw that adults’ first interpretation corresponds to
children’s only interpretation. Much evidence suggests that children have great
difficulty in revising their initial parses, which may be due to independent
limitations in their cognitive control abilities (Omaki & Lidz, 2013; Snedeker,
2013; Trueswell et al., 1999). The contrasts between children’s mastery of
different constraints on anaphora would be mysterious if we tried to understand
them as a case of uneven grammatical development. Thanks to the link with
adult on-line parsing profiles, the challenge for grammatical learning theories
changes. Rather than needing to explain why children show severe delays in
learning some constraints on anaphora, we instead are left with the task of
explaining the uneven profile in the on-line effects of grammatical constraints in
adults (cf. Phillips et al., 2011).

5. Level 3 accounts: Explaining learning via processing

This level aims to go beyond documenting how language processing gives rise to
specific effects during child or adult learning and instead ask how language
processing contributes to an understanding of learning success in children, or
ultimate attainment in adult learners. This is not straightforward, since the most
salient finding about language processing in learners is that they are limited in
various ways and prone to errors. Therefore we minimally want to understand
why the learners’ processing limitations do not jeopardize their chances of
success. If we are particularly optimistic, then we might hope that their
processing abilities could somehow contribute to explaining their learning
outcomes, including an understanding of why children outperform adults.
Unlike the previous section, which was built around a synthesis of clearly
related empirical findings, this section is more speculative: we discuss some
ways in which language processing findings might potentially contribute to an
understanding of learning outcomes. The key elements here involve observations
432 Colin Phillips and Lara Ehrenhofer
about the age at which children outperform adults, and the possible role of rich
predictive processing mechanisms.

5.1 Where is language processing relevant to learning?

The more information the learner is able to extract from the input, the more
successful he is likely to be, and language processing is key here.
As a first step, the learner minimally needs to be able to accurately analyze
sentences that he encounters in the target language. Without this, it is difficult to
see how learning could ever succeed. As we have seen, there are a number of
ways that learners might frequently misanalyze the input. How much this matters
depends on how learning proceeds. If learners rely on key encounters with
informative examples in their language input, and are relatively insensitive to
distributional information, then they might be relatively unaffected by mis-
parses. But if learners instead rely on observing distributional patterns in the
input, then systematic mis-parsing potentially wreaks havoc with learning, by
changing the distributions. Distributional learning is widely assumed to be
important in many learning theories (e.g., Lidz & Gagliardi, 2014; Pearl &
Sprouse, 2013; Perfors et al., 2011; Saffran, 2003; Yang, 2002), and it is
probably the only viable option in cases where the learner must retreat from an
over-general hypothesis.
Various of the studies reviewed above highlight the potential dangers of
systematic mis-parsing, which can lead to discrepancies between “input” and
“intake” (Omaki & Lidz, 2014). It is currently unclear how much of a danger
this presents for learners. It is one thing to mis-parse sentences in controlled
laboratory settings, and entirely another to systematically mis-parse crucial
experience in everyday life. To assess the seriousness of the danger, we need to
know what distributional information about sentences is crucial for learners,
whether mis-parsing is a real danger in the uncontrolled settings where children
naturally encounter those sentences, and how much the distributional
information in the input is therefore likely to be distorted (see Omaki (2010) for
further discussion).
As a second step, language learners might be able to extract valuable
additional information from their experience if they can compare predicted input
with actual input. A learner who simply passively analyzes incoming sentences
can recognize whether he is able to parse each sentence as it arrives, but this is
informative only when he encounters a sentence that he cannot parse. In contrast,
a more active learner who generates expectations as a sentence unfolds,
predicting upcoming categories and words, potentially gains a lot more
information about the language input. If the active learner not only makes
predictions, but tracks their source, then he can use confirmed and disconfirmed
predictions as feedback that either reinforces or weakens learned generalizations,
 The role of language processing in language acquisition
433
allowing retreat from overgeneralizations. In other words, a predictive parser is
a hypothesis testing device.
An active learner that acts on complex cues to generate predictions could
potentially learn about complex contingencies in the language. But in order to do
that, the learner must first be able to recognize and integrate complex cues in
order to initiate predictions. This ability may be beyond the reach of beginning
learners, and certainly beyond the reach of early child learners. We have seen in
Section 3.1 that children have difficulty integrating multiple cues to ambiguity
resolution (Engelhardt, 2014; Snedeker & Trueswell, 2004; Trueswell et al.,
1999), and so it is unlikely that they can use complex cues to generate
predictions.
Also, an active learner can use predictions to generate feedback only if he
does so quickly enough to ‘get ahead of’ the input. We currently know little
about how quickly different types of predictions can be generated, but as we saw
in Section 2.5, we are starting to learn that some are generated only slowly.
One clear obstacle to drawing a close connection between parsing and
learning is that children are more successful language learners than adults,
despite not being especially good parsers. So we should not want to rely too
heavily on language processing abilities for ensuring learning success. But of
course this is at odds with a rather well known claim in the language acquisition
literature, namely that children are better language learners precisely because of
their cognitive limitations (Newport, 1990). We now investigate this claim,
which seems rather implausible, based on what we now know about how
children’s limitations impact their language processing.

5.2 Less is more. Maybe.

According to the ‘less is more’ proposal, adults fail because they are too smart
(Newport, 1990), and learners with limited resources should enjoy better
outcomes (Elman, 1993). This idea has attracted a great deal of attention, but
surprisingly little critical discussion. A strength of Newport and Elman’s claims
is that they are backed up by computational simulations, but a closer look at the
models shows that they largely avoid the problems that most concern us here.
Newport’s argument is supported by a computational model of
morphological learning. Goldowsky and Newport (1993; henceforth GN) present
a model of how learners figure out the correspondence between morphemes and
meanings in complex words that they encounter. For example, the learner must
discover that the word unhelpful has three morphemes, and what the
meaning/function of each is (17), i.e., the model learns pairwise
correspondences.
434 Colin Phillips and Lara Ehrenhofer
(17) un help ful neg
‘aid’ adj

For each word that it encounters, the model considers multiple potential
correspondences. As it accumulates more experience, it builds up in memory a
matrix of potential correspondences, and it uses this to identify reliable
correspondences. Its main task is to separate true correspondences, e.g., un=neg
from spurious correspondences, e.g., ful = neg + aid. GN show that the model’s
performance is limited by the large number of spurious correspondences that it
stores, and that performance improves when the model’s memory is limited, so
that it stores only one or two potential correspondences per trial.
Rohde and Plaut (2003) is one of very few studies to engage with the
specifics of the GN model. They argue that GN made an unwarranted
assumption in their model’s retrieval mechanism, and that correcting this yields
a model that fares better with unlimited memory.
A more important question about the GN simulations is whether their
assumptions are applicable to the challenge of learning syntax and semantics. In
particular, their model takes as its starting point a perfect encoding of the form
units and meaning units that need to be learned, e.g., the morphemes and the
syntactic/semantic chunks in (17) are exactly the right ones, and the model’s
only task is to figure out how forms and syntax/semantics align. (The model also
considers that combinations of form units might be linked to particular
meanings, but this does not greatly affect the model’s task.) Giving the model a
limited memory imposes a bias for encoding correspondences of approximately
the right grain size. Similarly, other recent demonstrations of ‘less is more’
effects in word segmentation models succeed because the model is forced to
make plausible assumptions about the size of the target lexicon (Hitzcenko &
Jarosz, 2014; Pearl et al., 2010; see Perfors (2012) for useful discussion of the
relation between memory limitations and learning biases).
The challenge in learning syntax and semantics is rather different. The
learner’s task is to discover the structural units of the target grammar, and what
semantic units they correspond to. But both of these are hard to observe and
prone to misclassification. Even if we generously assume that learners can
perfectly infer a speaker’s intent with a sentence, that is very different than
knowing which parts of that intent are mapped onto semantic units, and which
are left to pragmatic inference (for a useful discussion see Lewis, 2013).
Therefore, the main problem in learning syntax and semantics is the immense
ambiguity of the evidence, and this part of the problem is largely treated as pre-
solved in GN’s morphological learning model. Their model starts with a set of
generally plausible hypotheses, and the effect of resource limitations is to focus
attention on more likely hypotheses. But in learning syntax and semantics the
greatest danger is that learners could misanalyze the syntactic and semantic units
 The role of language processing in language acquisition
435
in the input. Resource limitations can only increase this danger, by making mis-
analyses more likely.
Elman (1993) presents a different computational demonstration of the
“importance of starting small” that is closer to the syntax/semantics learning
problem. Elman’s neural network model uses a simple recurrent network to
predict the next word in sequences of pseudo-English. To do this, it needs to
implicitly learn the grammar. The model focuses on correctly learning subject-
verb agreement relations, even when the subject and verb are separated by
intervening relative clauses. Elman argues that the model learns best when it is
initially restricted, either via limited memory or via simplified input. The
conclusions about the benefits of starting small are disputed by Rohde and Plaut
(1999), who claim that Elman’s results do not replicate across a variety of model
configurations. But perhaps more important is that the syntactic domain that
Elman focuses on is unrepresentative of the task that real learners face. Subject-
verb agreement is found in almost every clause of every sentence that learners
encounter. So learners enjoy an abundance of evidence about subject-verb
agreement, and the relation between subject properties (number/person) and
agreement properties (number/person) is relatively straightforward. Child
learners’ main challenges lie elsewhere, and this domain is not one that is
affected by critical periods. (Adult learners sometimes encounter difficulty in
retrieving the correct agreement forms, but they face no special difficulty in
identifying which items should agree.)
In sum, even beyond concerns about the original computational arguments, it
is not clear how helpful existing demonstrations are in understanding children’s
success in learning syntax and semantics. The benefits of limited resources are
not magical. They may be useful in linguistically simple domains where a
constraint like “don’t memorize too much” is sufficient to point the learner in the
right direction. But what makes learning syntax and semantics so hard is that
there are too many potentially misleading analyses of the structure and meaning
of any incoming sentence, and so the learner needs more specific constraints on
his hypothesis space. So we are skeptical of the standard version of ‘less is more’
as a contributor to grammar learning. But this does not mean that we have given
up on the idea entirely, as we shall see shortly.

5.3 The nature of the child advantage: When do children shine?

In order to understand why children outperform adults in learning sentence-level

grammar, we need to know when children outperform adult learners. Put simply,
if children already outperform adults at age 2, then we need an account that can
capture how 2-year-olds notice generalizations that adults miss, despite their
436 Colin Phillips and Lara Ehrenhofer
acute limitations. But if children do not outperform adult learners until a later
age, we need a different kind of account.
In order to figure out when and where children outperform adult learners, we
need to first know what language phenomena cause the greatest difficulty for
adult learners, and then find out when children master those phenomena. This is
easier said than done, as it is hard to find a consensus on the things that adult
learners fail to master.
There are things that adults typically fail to learn, even after many years of
immersion. Different adult learners have trouble with different linguistic
phenomena, depending on the relation between their L1 and their L2. But it is
hard to distinguish slow learning from failed learning. If we are most concerned
about the things that late learners have the most difficulty with, then we need to
look at very advanced learners, i.e., at the question of ‘ultimate attainment’
(Abrahamsson & Hyltenstam, 2009; Birdsong, 2004; Coppieters, 1987;
DeKeyser, 2005; Montrul & Slabakova, 2003).
A common claim about advanced adult learners is that they fare well on
learning obligatory syntactic rules and robust word order patterns, but that they
have the greatest difficulty with forms that are used optionally, especially if the
optionality is conditioned by discourse or pragmatics (Coppieters, 1987; Sorace,
2011; Sorace & Serratrice, 2009).
Much more needs to be learned about where the most advanced adult
learners fail, but our impression is that the phenomena that adults struggle with
the most are not things that children typically master at a very young age. 5
Children appear to know a great deal about the word order and morphology of
the target language at an early age, e.g., by age 3 or 4 (e.g., Guasti, 2002;
Phillips 1995/2010; Snyder, 2007), but there are many things that children do not
master until later, and these often involve semantically- or pragmatically-
conditioned formal choices (e.g., de Villiers, 2005; Goro, 2007; Kazanina &
Phillips, 2007; Speer & Ito, 2009; Weckerly, Wulfeck, & Reilly, 2004).
Here is a critical part of our case: if the domains where adults fall short are
ones that children master relatively late, when their cognitive and language
processing resources are sharply improving, then this undermines the notion that
children outperform adults specifically because they are so cognitively limited. If
the limitations were the cause of children’s success, then surely children would
shine at an earlier age. That might be the case in some domains of language,
such as speech articulation, but it does not seem to be the case for syntax and
semantics. So the next step in our argument is to address why the things that are
learned slowest are so difficult.
A potential source of inspiration for understanding where children
outperform adults should be controlled studies of artificial grammar learning.
5 . Lexical features such as gender and tone assignment are exceptions to this generalization,
and we have little to add to existing claims about why adult learners struggle with these.
 The role of language processing in language acquisition
437
But unfortunately these studies tend to show the opposite, namely, that adults
outperform children (e.g., Ferman & Karni, 2010). This could be because
artificial grammar learning tasks favor ‘explicit learning’ over ‘implicit learning’
mechanisms (DeKeyser, 2003), giving the edge to adults. Alternatively, it could
just be that the kinds of phenomena that can feasibly be tested in a lab-based
study are too simple to allow children to shine.

5.4 What makes the hard stuff so hard?

So what is special about the linguistic phenomena that cause the greatest
difficulty for learners? In particular, what is distinctive about the things that
children learn late, and that adults may never master? We start with the question
of why these phenomena might be so hard to learn, and then proceed to the
question of why children fare better than adults.
One tempting explanation for things that children learn late is that they are
rare phenomena that children simply do not receive much evidence for. Learning
is slow because children must wait to accumulate sufficient evidence. A concern
about this line of reasoning is that blaming sparse input data for late
development should predict enormous variability in when children master a
given phenomenon. If the relevant input data is so sparse that a typical child
must wait for 6 years to accumulate enough evidence, then perhaps some
children never encounter enough evidence and would fail to learn, and other
children might get lucky and see enough evidence to learn much earlier. In this
scenario, variation in learning schedules should be linked to learners’
experience, and not to their learning abilities, so we should expect major
variability in outcomes, unrelated to children’s general cognitive or linguistic
talents. To our knowledge, we do not find cases like this, and so we are skeptical
of accounts that blame delayed development of specific linguistic phenomena on
mere sparse input.6
A more promising alternative is that there are internal changes in learners
that allow them to recognize evidence in their surroundings that they were blind
to before. For example, independent development of social cognitive abilities
might lead children to recognize semantic/pragmatic regularities that they were
not aware of earlier. Alternatively, learners might notice new evidence in their
surroundings because their language processing abilities develop in a way that
allows them to test more complex hypotheses about contingencies in the input.
For example, if they become able to generate predictions based on complex cues,
integrating information across domains, this could allow them to confirm or
disconfirm regularities that they were previously unable to test. Under either of
6 . This is to be contrasted with the effect of overall limited input, which may have severe
consequences for language development in general (Hart & Risley, 1995).
438 Colin Phillips and Lara Ehrenhofer
these scenarios, variation in learning speed is tied more to individual learners’
abilities than to properties of their language input.
This leads to a potential link between the development of language
processing abilities and the mastery of hard-to-learn linguistic generalizations.
i. Complex contingencies, especially involving the integration of information
across multiple levels of encoding, e.g., syntax and pragmatics, may be
among the hardest things to learn.
ii. In the domain of structural ambiguity resolution, children show difficulty in
rapidly integrating different sources of information to guide their parsing
choices.
iii. Learning complex contingencies often draws on predictive mechanisms,
which allow the learner to test hypotheses. (This is especially relevant in
cases where the learner needs to retreat from overgeneralization.)
iv. In the domain of linguistic prediction, some types of linguistic constraint
generate predictions more slowly than others (see Section 2.5).
v. Generating predictions based on combined information from multiple
domains may be particularly difficult, and may be a late-developing ability.
Among these points (i) is moderately well motivated; (ii) is relatively
uncontroversial; (iii) is more speculative, but we do not know of better ways that
learners could test hypotheses and retreat from overgeneralizations; (iv) is not a
well-established finding, but we are seeing more and more evidence that it is
true; and (v) is something that we have no direct evidence for, though it seems
eminently plausible if (ii–iv) are true.
Clearly the link that we are suggesting here involves some uncertain steps, so
further work would be needed to confirm or reject it. But first we need to turn to
an even more pressing concern: how could this hope to explain why children fare
better than adults?

5.5 Why children shine, and how less could (eventually) be more

Even if it is true that late-developing language processing abilities are essential

for children’s learning success, we still need to know why adults fare less well.
Shouldn’t they be able to learn using exactly the same mechanisms that children
rely on?
We could, of course, argue that adult learners might never attain the
language processing abilities that a typical primary school child develops. For
example, if the key to learning complex linguistic phenomena involves complex
predictive mechanisms, perhaps adults never get to be as good as children. But
we are not optimistic about this approach, as it would require that even the best
adult learners should have worse language processing abilities than relatively
poor child learners, which seems unlikely.
 The role of language processing in language acquisition
439
As far as we know, adults have the basic cognitive abilities that should allow
them to process incoming sentences just as proficiently as children do.
We saw that younger children show limitations in initial analysis and
reanalysis, especially in the ability to revise their initial interpretations. A
standard view is that this is not a language-specific problem, but instead reflects
the slow maturation of cognitive control abilities. If this indeed reflects a
maturational constraint, then we should expect that at early stages of learning
adults should fare better than children. And this does appear to be the case,
specifically in the domain of reanalysis.
In contrast, we suspect that children’s limitations in predictive processes
reflect proficiency rather than maturation. Children develop better prediction
mechanisms as they learn more about their language. If predictive mechanisms
are tied to proficiency, then they should develop similarly in child and adult
learners.
So why should adults, who do not obviously lag behind children at any stage
in the basic abilities that underlie sentence processing, end up as inferior
learners?
It is possible that adult learners’ deficits at other levels of language
processing, such as sounds and words, do lag behind even young children, and
that the consequences of this are so severe that they hold back sentence
processing abilities. But we do not think that we have good evidence for such a
claim.
An alternative possibility, which is probably the most speculative piece of
our already speculative argument, is that advanced adult learners do have the
same basic sentence processing abilities as advanced child learners, but that they
are held back by what they learned at earlier stages of learning. Adult learners
should, in principle, be able to analyze input sentences in the same way as
advanced child learners, and use their predictive mechanisms to generate and test
complex contingencies involving information from different linguistic domains.
But perhaps the ‘toxic combination’ is that adult learners’ early successes — at a
stage when they are relatively good (re)analyzers but not yet good predictors —
somehow lock them into sentence processing routines that make them less
sensitive to the new information that should become available to them later, once
they become more effective predictors. This is a version of the ‘blocking’
account of adult difficulties that a number of other authors have proposed (e.g.,
Arnon & Ramscar, 2012; Ellis & Sagarra, 2010; Kamin, 1969).
But if adult learners’ problem is that their early processing successes make
them less receptive to later insights, why shouldn’t children have exactly the
same problem? We do not have a good answer to this, but can offer some
suggestions for where to look. (i) Children’s immature memory mechanisms
might make them less committed to what they learn initially, and hence more
440 Colin Phillips and Lara Ehrenhofer
receptive to later insights, once they have the improved processing abilities
needed to gain those insights. (ii) Children’s inferior (re)analysis abilities might
give them less confidence in their initial learning. (iii) It could be that (simple)
predictive abilities develop earlier in adult learners than in children, in a way that
might give rise to differential learning. But the evidence about predictive
abilities in learners is currently far too limited to draw clear conclusions about
this.
Summarizing, our speculation is that adult learners might be held back
precisely because their early learning prevents later progress (cf. Snow &
HoefnagelHöhle, 1978). So this brings us surprisingly close to a ‘less is more’-
style account. But instead of proposing that children outperform adult learners
because their information processing limitations lead directly to more accurate
insights, we instead propose that their limitations might help them to avoid
damage, making it possible for them to eventually outperform adults, once their
processing resources become more fluent. In other words, our suggestion could
be summarized as ‘less is eventually more’.
6. Conclusion

Our main goal here was to advocate for a specific type of relation between
language acquisition and language processing research, especially at the
syntax/semantics level. It is easy enough to take ideas from the adult L1
psycholinguistics literature and port them to learner populations. This can teach
us a lot about the learners, but it is less clear how it helps us in understanding
learning. And that is surely our primary aim. We described some ways in which
language processing research could contribute to accounts of how learning
succeeds, and how it fails. In discussing the key components of parsing abilities,
we emphasized components that might be essential for learners’ grammatical
parsing of the input. This included some new findings about the speed of
different types of predictive process, which could turn out to be important for
language learning, because predictive processing helps the learner test
hypotheses. We gave one extended example of how language processing
mechanisms can shed light on some old chestnuts in language learning. And
finally we made some conjectures about how specific processing abilities might
contribute to the success of child learning, and perhaps even to maturational
constraints on learners. We argued that ‘less is not more’, at least in the way that
the less is more view has generally been understood, but we suggested that a
‘less is eventually more’ approach might show promise.
Over the course of the article, we tried to pull together findings from
disparate areas that we think could be relevant to our main aim. But the reader
will have noticed that in doing so we had to make many leaps of faith, to
compensate for gaps in current knowledge. A second goal of this paper was to
highlight some topics that we need to know a lot more about, if we are to
 The role of language processing in language acquisition
441
understand successful and unsuccessful learning. We need a lot more research
that compares adult and child learners, especially the specific areas where
children outstrip adult learners. We need to know a lot more about how
predictive mechanisms can be deployed in adults: psycholinguists have
repeatedly demonstrated that these mechanisms exist, but we are only now
beginning to learn about how they might work in more detail. If we can
understand those mechanisms better, it could become feasible to link a more
detailed understanding of those mechanisms to the advanced stages of language
learning.

Acknowledgments

Many people contributed to the perspective on language processing mechanisms that we

outline here. For the specific relevance of language processing to language learning we thank
Akira Omaki, Jeff Lidz, and Andrea Zukowski for valuable discussions, spread across a
number of years. Robert DeKeyser provided more key tips than he realizes. Holger Hopp,
Theres Grüter, and Irina Sekerina provided valuable feedback on an earlier version. But none
of these people deserve any blame for the things that you don’t like here. This work was
supported in part by NSF BCS-0848554 to CP, and by a Fulbright award and a University of
Maryland Flagship Fellowship to LE.

References

Abrahamsson, N., & Hyltenstamm, K. (2009). Age of onset and nativelikeness in a second
language: Listener perception versus linguistic scrutiny. Language Learning, 59, 249–
306. DOI: 10.1111/j.1467-9922.2009.00507.x
Aldwayan, S., Fiorentino, R., & Gabriele, A. (2010). Evidence of syntactic constraints in the
processing of wh-movement: A study of Najdi Arabic learners of English. In B.
VanPatten & J. Jegerski (Eds.), Research in second language processing and parsing (pp.
65–86). Amsterdam: John Benjamins. DOI: 10.1075/lald.53.03ald
Ambridge, B., Pine, J.M., & Lieven, E.V.M. (2014). Child language acquisition: Why
Universal Grammar doesn’t help. Language: Perspectives.
Aoshima, S., Yoshida, M., & Phillips, C. (2009). Incremental processing of coreference and
binding in Japanese. Syntax, 12, 93–124. DOI: 10.1111/j.1467-9612.2009.00123.x
Arnon, I., & Ramscar, M. (2012). Granularity and the acquisition of grammatical gender:
How order-of-acquisition affects what gets learned. Cognition, 122, 292–305.
DOI: 10.1016/[Link].2011.10.009
Atkinson, E., Simeon, K., & Omaki, A. (2013). The time course of filler-gap dependency
processing in the developing parser. Poster presented at the 26th Annual CUNY Sentence
Processing Conference. Columbia, S. Carolina.
Avrutin, S., & Reuland, E. (2004). Backward anaphora and tense interpretation. In L.
Verbitskaya (Ed.), Teoreticheskie problemy jazykoznanija [Theoretical issues in
language research]. St. Petersburg, Russia: St. Petersburg University Press.
442 Colin Phillips and Lara Ehrenhofer
Baauw, S., Escobar, M.A., & Philip, W. (1997). A delay of Principle B effect in Spanish
speaking children: The role of lexical feature acquisition. In A. Sorace, C. Heycock, & R.
Shillcock (Eds.), Language acquisition: Knowledge representation and processing.
Proceedings of GALA 1997. Edinburgh: HCRC.
Badecker, W., & Straub, K. (2002). The processing role of structural constraints on the
interpretation of pronouns and anaphora. Journal of Experimental Psychology: Learning,
Memory, and Cognition, 28, 748–769. DOI: 10.1037/0278-7393.28.4.748
Berwick, R.C., Pietroski, P., Yankama, B., & Chomsky, N. (2011). Poverty of the stimulus
revisited. Cognitive Science, 35, 1207–1242. DOI: 10.1111/j.1551-6709.2011.01189.x
Bever, T.G. (1970). The cognitive basis for linguistic structure. In J.R. Hayes (Ed.), Cognition
and the development of language (pp. 279–362). New York: Wiley.
Birdsong, D. (2004). Second language acquisition and ultimate attainment. In A. Davies & C.
Elder (eds.), Handbook of Applied Linguistics (pp. 82–105). London: Blackwell. DOI:
10.1002/9780470757000.ch3
Bock, K., & Miller, C.A. (1991). Broken agreement. Cognitive Psychology, 23, 45–93.
DOI: 10.1016/0010-0285(91)90003-7
Borovsky, A., Elman, J.L., & Fernald, A. (2012). Knowing a lot for one’s age: Vocabulary
skill and not age is associated with anticipatory incremental sentence interpretation in
children and adults. Journal of experimental child psychology, 112, 417–436.
DOI: 10.1016/[Link].2012.01.005
Chang, F., Dell, G., & Bock, J.K. (2006). Becoming syntactic. Psychological Review, 113,
234–272. DOI: 10.1037/0033-295X.113.2.234
Chen, Z., Jäger, L., & Vasishth, S. (2012). How structure-sensitive is the parser? Evidence
from Mandarin Chinese. In B. Stoltefoht & S. Featherson (Eds.), Empirical approaches
to linguistic theory: Studies in meaning and structure (pp. 43–62). Berlin: Mouton de
Gruyter.
Chien, Y.C. & Wexler, K. (1990). Children’s knowledge of locality conditions in binding as
evidence for the modularity of syntax and pragmatics. Language Acquisition, 1, 225–
295. DOI: 10.1207/s15327817la0103_2
Choi, Y., & Trueswell, J.C. (2010). Children’s (in) ability to recover from garden paths in a
verb-final language: Evidence for developing control in sentence processing. Journal of
Experimental Child Psychology, 106, 41–61. DOI: 10.1016/[Link].2010.01.003 Chomsky, N.
(1981). Lectures on government and binding. Dordrecht: Foris.
Chow, W.Y., Lewis, S., & Phillips, C. (2014). Immediate sensitivity to structural constraints
in pronoun resolution. Frontiers in Psychology, 5, 630. DOI: 10.3389/fpsyg.2014.00630
Chow, W.Y., Lau, E.F., Wang, S., & Phillips, C. (submitted). Timing is everything: The
temporal dynamics of word prediction.
Chow, W.Y., Smith, C., Lau, E.F., & Phillips, C. (2015). A “bag of arguments” mechanism
for verb prediction. Language, Cognition, and Neuroscience, in press.
Christianson, K., Hollingworth, A., Halliwell, J., & Ferreira, F. (2001). Thematic Roles
Assigned along the Garden Path Linger. Cognitive Psychology, 368–407. DOI:
10.1006/cogp.2001/0752
Clackson, K., Felser, C., & Clahsen, H. (2011). Children’s processing of reflexives and
pronouns in English: Evidence from eye-movements during listening. Journal of
Memory and Language, 65, 128–144. DOI: 10.1016/[Link].2011.04.007
Clahsen, H., & Felser, C. (2006). Grammatical processing in language learners. Applied
Psycholinguistics, 27, 3–42. DOI: 10.1017/S0142716406060024
Clifton, C., Kennison, S.M., & Albrecht, J.E. (1997). Reading the words her, his, him:
Implications for parsing principles based on frequency and structure. Journal of Memory
and Language, 36, 276–292. DOI: 10.1006/jmla.1996.2499
 The role of language processing in language acquisition
443
Clifton, C., Frazier, L., & Deevy, P. (1999). Feature manipulation in sentence comprehension.
Rivista di lingüística, 11, 11–40.
Conroy, A., Takahashi, E., Lidz, J., & Phillips, C. (2009). Equal treatment for all antecedents:
How children succeed with Principle B. Linguistic Inquiry, 40, 446–486. DOI:
10.1162/ling.2009.40.3.446
Coppieters, R. (1987). Competence differences between native and near-native speakers.
Language, 544–573. DOI: 10.2307/415005
Crain, S., & McKee, C. (1985). The acquisition of structural restrictions on anaphora. In S.
Berman, J.-W. Choe, & J. McDonough (Eds.), Proceedings of NELS 15 (pp. 94–110).
Amherst, MA: GLSA Publications.
Crain, S., & Thornton, R. (1998). Investigations in universal grammar. Cambridge, MA: MIT
Press.
Cunnings, I., & Felser, C. (2013). The role of working memory in the processing of
reflexives.
Language and Cognitive Processes, 28, 188–219. DOI: 10.1080/01690965.2010.548391
Davidson, M.C., Amso, D., Anderson, L.C., & Diamond, A. (2006). Development of
cognitive control and executive functions from 4 to 13 years: Evidence from
manipulations of memory, inhibition, and task switching. Neuropsychologia, 44, 2037–
2078.
DOI: 10.1016/[Link].2006.02.006
Deacon, D., Dynowska, A., Ritter, W., & Grose-Fifer, J. (2004). Repetition and semantic
priming of nonwords: Implications for theories of N400 and word recognition.
Psychophysiology, 41, 60–74. DOI: 10.1111/1469-8986.00120
DeKeyser, R.M. (2003). Implicit and explicit learning. In C. Doughty & M. Long (Eds.),
Handbook of second language acquisition (pp. 313–348). Oxford, UK: Blackwell.
DeKeyser, R.M. (2005). What makes learning second-language grammar difficult? A review
of issues. Language Learning, 55, 1–25. DOI: 10.1111/j.0023-8333.2005.00294.x
DeLong, K.A., Urbach, T.P., & Kutas, M. (2005). Probabilistic word pre-activation during
language comprehension inferred from electrical brain activity. Nature Neuroscience, 8,
1117–
1121. DOI: 10.1038/nn1504 de Villiers, J.G. (2005). Can language acquisition give
children a point of view? In J. Astington & J. Baird (Eds.), Why language matters for theory
of mind (pp. 266–297). New York: Oxford
University Press. DOI: 10.1093/acprof:oso/9780195159912.003.0013 de Villiers, J.G. &
Roeper, T. (1995). Barriers, binding, and acquisition of the DP-NP distinction.
Language Acquisition, 4(1–2), 73–104. DOI: 10.1080/10489223.1995.9671660 de
Villiers, J.G., Roper, T., Bland-Stewart, L., & Pearson, B. (2008). Answering hard questions:
wh-movement across dialects and disorder. Applied Psycholinguistics, 29, 67–103.
DOI: 10.1017/S0142716408080041
Dikker, S., Rabagliati, H., & Pylkkänen, L. (2009). Sensitivity to syntax in visual cortex.
Cognition, 110, 293–321. DOI: 10.1016/[Link].2008.09.008
Dillon, B. (2014). Syntactic memory in the comprehension of reflexive dependencies: An
overview. Language and Linguistics Compass, 8, 171–187. DOI: 10.1111/lnc3.12075
Dillon, B., Mishler, A., Sloggett, S., & Phillips, C. (2013). Contrasting intrusion profiles for
agreement and anaphora: Experimental and modeling evidence. Journal of Memory and
Language, 69, 85–103. DOI: 10.1016/[Link].2013.04.003
Dussias, P.E., Valdés Kroff, J.R., Guzzardo Tamargo, R.E., & Gerfen, C. (2013). When
gender and looking go hand in hand: Grammatical gender processing in L2 Spanish.
444 Colin Phillips and Lara Ehrenhofer
Studies in Second Language Acquisition, 35, 353–387. DOI:
10.1017/S0272263112000915
Eberhard, K.M., Cutting, J.C., & Bock, K. (2005). Making syntax of sense: Number
agreement in sentence production. Psychological Review, 112, 531–559.
DOI: 10.1037/0033-295X.112.3.531
Elbourne, P. (2005). On the acquisition of Principle B. Linguistic Inquiry, 36(3), 333–365.
DOI: 10.1162/0024389054396908
Ellis, N.C. & Sagarra, N. (2010). The bounds of adult language acquisition: blocking and
learned attention. Studies in Second Language Acquisition, 32, 553–580. DOI:
10.1017/S0272263110000264
Elman, J.L. (1993). Learning and development in neural networks: The importance of starting
small. Cognition, 48, 71–99. DOI: 10.1016/0010-0277(93)90058-4
Engelhardt, P.E. (2014). Children’s and adolescents’ processing of temporary syntactic
ambiguity: An eye movement study. Child Development Research, 2014: Article ID
475315.
Fedele, E., & Kaiser, E. (2014). Looking back and looking forward: Anaphora and cataphora
in Italian. University of Pennsylvania Working Papers in Linguistics, 20, 81–90.
Federmeier, K.D. (2007). Thinking ahead: The role and roots of prediction in language
comprehension. Psychophysiology, 44, 491–505. DOI: 10.1111/j.1469-
8986.2007.00531.x
Federmeier, K.D., & Kutas, M. (1999). A rose by any other name: Long-term memory
structure and sentence processing. Journal of Memory and Language, 41, 469–495. DOI:
10.1006/jmla.1999.2660
Felser, C., & Cunnings, I. (2012). Processing reflexives in a second language: The timing of
structural and discourse-level constraints. Applied Psycholinguistics, 33, 571–603. DOI:
10.1017/S0142716411000488
Felser, C., Cunnings, I., Batterham, C., & Clahsen, H. (2012). The timing of island effects in
nonnative sentence processing. Studies in Second Language Acquisition, 34, 67–98.
DOI: 10.1017/S0272263111000507
Ferman, S., & Karni, A. (2010). No childhood advantage in the acquisition of skill in using an
artificial language rule. PloS ONE, 5(10), e13648. DOI: 10.1371/[Link].0013648
Ferreira, F., & Henderson, J.M. (1991). Recovery from misanalyses of garden-path sentences.
Journal of Memory and Language, 30, 725–745. DOI: 10.1016/0749-596X(91)90034-H
Ferreira, F. & Patson, N.D. (2007). The ‘good enough’ approach to language comprehension.
Language and Linguistics Compass, 1, 71–83. DOI: 10.1111/j.1749-818X.2007.00007.x
Fischler, I., Bloom, P.A., Childers, D.G., Roucos, S.E., & Perry, N.W. (1983). Brain
potentials related to stages of sentence verification. Psychophysiology, 20, 400–409.
DOI: 10.1111/j.1469-8986.1983.tb00920.x
Fodor, J.D. (1978). Parsing strategies and constraints on transformations. Linguistic Inquiry,
9, 427–473.
Foucart, A., Martin, C.D., Moreno, E.M., & Costa, A. (2014). Can bilinguals see it coming?
Word anticipation in L2 sentence reading. Journal of Experimental Psychology:
Learning, Memory, and Cognition, 40, 1461–1469. DOI: 10.1037/a0036756
Frazier, L. (1999). On sentence interpretation. Dordrecht: Kluwer.
DOI: 10.1007/978-94-011-4599-2
Frazier, L., & Flores d’Arcais, G.B. (1989). Filler-driven parsing: A study of gap filling in
Dutch. Journal of Memory and Language, 28, 331–344. DOI: 10.1016/0749-
596X(89)90037-5
Gibson, E. (1998). Linguistic complexity: Locality of syntactic dependencies. Cognition, 68,
1–76. DOI: 10.1016/S0010-0277(98)00034-1
 The role of language processing in language acquisition
445
Gibson, E., & Warren, T. (2004). Reading-Time evidence for intermediate linguistic structure
in long-distance dependencies. Syntax, 7, 55–78. DOI: 10.1111/j.1368-
0005.2004.00065.x
Goldowsky, B.N. & Newport, E.L. (1993). Modeling the effects of processing limitations on
the acquisition of morphology: the less is more hypothesis. In E. V. Clark (Ed.),
Proceedings of the 24th annual child language research forum, 124–138.
Gordon, P., & Hendrick, R. (1997). Intuitive knowledge of linguistic co-reference. Cognition,
62, 325–370. DOI: 10.1016/S0010-0277(96)00788-3
Goro, T. (2007). Language-specific constraints on scope interpretation in first language
acquisition. PhD dissertation, University of Maryland.
Goro, T., & Akiba, S. (2004) The acquisition of disjunction and positive polarity in Japanese.
In V. Chand, A. Kelleher, A.J. Rodríguez, & B. Schmeiser (Eds.), WCCFL 23:
Proceedings of the 23rd West Coast Conference on Formal Linguistics (pp. 251–264).
Somerville, MA: Cascadilla Press.
Grüter, T., Lew-Wiliams, C., & Fernald, A. (2012). Grammatical gender in L2: A production
or a real-time processing problem. Second Language Research, 28, 191–215.
DOI: 10.1177/0267658312437990
Gualmini, A. (2008). The rise and fall of Isomorphism. Lingua, 118, 1158–1176.
DOI: 10.1016/[Link].2008.02.003
Guasti, M.T. (2002). Language acquisition: The growth of grammar. Cambridge, MA: MIT
Press.
Halgren, E., Dhond, R., Christensen, N., Petten, C., Marinkovic, K., Lewine, J., & Dale, A.
(2002). N400-like Magnetoencephalography Responses Modulated by Semantic Context,
Word Frequency, and Lexical Class in Sentences. NeuroImage, 1101–1116. DOI:
10.1006/nimg.2002.1268
Harris, C.L., & Bates, E.A. (2002). Clausal backgrounding and pronominal reference: A
functionalist approach to c-command. Language and Cognitive Processes, 17, 237–269.
DOI: 10.1080/01690960143000227
Hart, B., & Risley, T.R. (1995). Meaningful differences in the everyday experience of young
American children. Paul H Brookes Publishing.
Helenius, P., Salmelin, R., Service, E., & Connolly, J. (1999). Semantic Cortical Activation in
Dyslexic Readers. Journal of Cognitive Neuroscience, 535–550. DOI:
10.1162/089892999563599
Heycock, C. (1995). Asymmetries in reconstruction. Linguistic Inquiry, 26, 547–570.
Hitzcenko, K., & Jarosz, G. (2014). Cognitive limitations impose advantageous constraints on
word segmentation. Talk at the 39th Boston University Conference on Language
Development.
Holcomb, P., Grainger, J., & O’Rourke, T. (2002). An electrophysiological study of the
effects of orthographic neighborhood size on printed word perception. Journal of
Cognitive Neuroscience, 14, 938–950. DOI: 10.1162/089892902760191153
Hopp, H. (2010). Ultimate attainment in L2 inflection: Performance similarities between
nonnative and native speakers. Lingua, 120, 901–931. DOI:
10.1016/[Link].2009.06.004
Hopp, H. (2013). Grammatical gender in adult L2 acquisition: Relations between lexical and
syntactic variability. Second Language Research, 29, 33–56. DOI:
10.1177/0267658312461803
Hopp, H. (2015). Individual differences in the L2 processing of object-subject ambiguities.
Applied Psycholinguistics. Published online 11 April 2013. DOI:
10.1017/S0142716413000180
446 Colin Phillips and Lara Ehrenhofer
Huang, C.T.J. (1993). Reconstruction and the structure of VP: Some theoretical
consequences. Linguistic Inquiry, 24, 103–138.
Huang, Y.T., Zheng, X., Meng, X., & Snedeker, J. (2013). Children’s assignment of
grammatical roles in the online processing of Mandarin passive sentences. Journal of
Memory and Language, 69, 589–606. DOI: 10.1016/[Link].2013.08.002
Hurewitz, F., Brown-Schmidt, S., Thorpe, K., Gleitman, L.R., & Trueswell, J.C. (2000). One
frog, two frog, red frog, blue frog: Factors affecting children’s syntactic choices in
production and comprehension. Journal of Psycholinguistic Research, 29, 597–626.
DOI: 10.1023/A:1026468209238
Jackson, C.N., & Dussias, P.E. (2009). Cross-linguistic differences and their impact on L2
sentence processing. Bilingualism: Language and Cognition, 12, 65–82.
DOI: 10.1017/S1366728908003908
Jacob, G., & Felser, C. (2015). Reanalysis and semantic persistence in native and non-native
garden path recovery. Quarterly Journal of Experimental Psychology. Published online
January 27th, 2015. DOI: 10.1080/17470218.2014.984231
Jiang, N. (2004). Morphological insensitivity in second language processing. Applied
Psycholinguistics, 25, 603–634. DOI: 10.1017/S0142716404001298
Kaan, E. (2014). Predictive sentence processing in L2 and L1: What is different? Linguistic
Approaches to Bilingualism, 4, 257–282. DOI: 10.1075/lab.4.2.05kaa
Kamin, L.J. (1969). Predictability, attention, surprise, and conditioning. In B.A. Campbell &
R.M. Church (Eds.), Punishment and aversive behavior (pp. 276–296). New York:
AppletonCentury-Crofts.
Kaufman, D. (1988). Grammatical and cognitive interactions in the study of children’s
knowledge of binding theory and reference relations. PhD dissertation, Temple
University, Philadelphia, PA.
Kazanina, N. (2005). The acquisition and processing of backwards anaphora. Unpublished
PhD dissertation, University of Maryland, College Park.
Kazanina, N., Lau, E., Lieberman, M., Yoshida, M., & Phillips, C. (2007). The effect of
syntactic constraints on the processing of backwards anaphora. Journal of Memory and
Language, 56, 384–409. DOI: 10.1016/[Link].2006.09.003
Kazanina, N., & Phillips, C. (2001). Coreference in child Russian: Distinguishing syntactic
and discourse constraints. In A. Do, L. Domínguez, & A. Johansen (Eds.), Proceedings
of the 25th Annual Boston University Conference on Language Development (pp. 413–
424). Somerville, MA: Cascadilla Press.
Kazanina, N., & Phillips, C. (2007). A developmental perspective on the Imperfective
Paradox. Cognition, 105, 65–102. DOI: 10.1016/[Link].2006.09.006
Kazanina, N., & Phillips, C. (2010). Differential effects of constraints in the processing of
Russian cataphora. The Quarterly Journal of Experimental Psychology, 63, 371–400.
DOI: 10.1080/17470210902974120
Kennison, S.M. (2003). Comprehending the pronouns her, him, and his: Implications for
theories of referential processing. Journal of Memory and Language, 49, 335–352. DOI:
10.1016/S0749-596X(03)00071-8
Kidd, E., Stewart, A.J., & Serratrice, L. (2011). Children do not overcome lexical biases
where adults do: the role of the referential scene in garden-path recovery. Journal of
Child Language, 38, 222–234. DOI: 10.1017/S0305000909990316
Kush, D., & Phillips, C. (2014). Local anaphor licensing in an SOV language: implications
for retrieval strategies. Frontiers in Psychology, 5, 1252. DOI:
10.3389/fpsyg.2014.01252
Kutas, M., & Federmeier, K.D. (2000). Electrophysiology reveals semantic memory use in
language comprehension. Trends in Cognitive Sciences, 4, 463–470.
 The role of language processing in language acquisition
447
DOI: 10.1016/S1364-6613(00)01560-6
Kutas, M., Van Petten, C.K., & Kluender, R. (2006). Psycholinguistics electrified Ⅱ: 1994–
2005. In M.A. Gernsbacher & M. Traxler (Eds.), Handbook of Psycholinguistics (pp.
659–724). New York: Elsevier. DOI: 10.1016/B978-012369374-7/50018-3
Kutas, M., & Hillyard, S.A. (1980). Reading senseless sentences: Brain potentials reflect
semantic incongruity. Science, 207, 203–205. DOI: 10.1126/science.7350657
Lassotta, R., Omaki, A., & Franck, J. (submitted). Developmental changes in the
misinterpretation of garden-path wh-questions in French. Ms., University of Geneva and
Johns Hopkins University.
Lau, E.F., Holcomb, P.J., & Kuperberg, G.K. (2013). Dissociating N400 effects of prediction
from association in single word contexts. Journal of Cognitive Neuroscience, 25, 484–
502. DOI: 10.1162/jocn_a_00328
Lau, E.F., Stroud, C., Plesch, S., & Phillips, C. (2006). The role of structural prediction in
rapid syntactic analysis. Brain and Language, 98, 74–88. DOI:
10.1016/[Link].2006.02.003
Lau, E.F., Phillips, C., & Poeppel, D. (2008). A cortical network for semantics:
(De)constructing the N400. Nature Reviews Neuroscience, 9, 920–933. DOI:
10.1038/nrn2532
Leddon, E. and Lidz, J. (2006). Reconstruction Effects in Child Language. In D. Bamman, T.
Magnitskaia, & C. Zaller, (Eds.), Proceedings of the 30th Boston University Conference
on Language Development. Cascadilla Press: Cambridge.
Lee, M.-W., & Williams, J.N. (2006). The role of grammatical constraints in intra-sentential
pronoun resolution. Ms., Cambridge University.
Levy, R. (2008). Expectation-based syntactic comprehension. Cognition, 106, 1126–1177.
DOI: 10.1016/[Link].2007.05.006
Lew-Williams, C., & Fernald, F. (2007). Young children learning Spanish make rapid use of
grammatical gender in spoken word recognition. Psychological Science, 18, 193–198.
DOI: 10.1111/j.1467-9280.2007.01871.x
Lew-Williams, C., & Fernald, F. (2010). Real-time processing of gender-marked articles by
native and non-native Spanish speakers. Journal of Memory and Language, 63, 447–464.
DOI: 10.1016/[Link].2010.07.003
Lewis, S. (2013). Pragmatic enrichment in language processing and development. PhD
dissertation, University of Maryland.
Lewis, S., & Phillips, C. (2015). Aligning grammatical theories and language processing
models. Journal of Psycholinguistic Research, Journal of Psycholinguistic Research, 44,
27–46. DOI: 10.3389/fpsyg.2014.00630.
Lidz, J. (2014). Quantification and scope in child language. To appear in Oxford Handbook of
Developmental Linguistics.
Lidz, J., Waxman, S., & Freedman, J. (2003). What infants know about syntax but couldn’t
have learned: experimental evidence for syntactic structure at 18 months. Cognition, 89,
295– 303. DOI: 10.1016/S0010-0277(03)00116-1
Lidz, J., & Musolino, J. (2002). Children’s command of quantification. Cognition, 84, 113–
154. DOI: 10.1016/S0010-0277(02)00013-6
Lukyanenko, C., Conroy, A., & Lidz, J. (2014). Is she patting Katie? Constraints on
pronominal reference in 30-month-olds. Language Learning and Development, 10, 328–
344. DOI: 10.1080/15475441.2013.853529
MacWhinney, B., Pleh, C., & Bates, E. (1985). The development of sentence interpretation in
Hungarian. Cognitive Psychology, 17, 178–209. DOI: 10.1016/0010-0285(85)90007-6
448 Colin Phillips and Lara Ehrenhofer
Mani, N., & Huettig, F. (2012). Prediction during language processing is a piece of cake —
But only for skilled producers. Journal of Experimental Psychology: Human Perception
and Performance, 38, 843. DOI: 10.1037/a0029284
Maratsos, M., Fox, D.E., Becker, J.A., & Chalkley, M.A. (1985). Semantic restrictions on
children’s passives. Cognition, 19, 167–191. DOI: 10.1016/0010-0277(85)90017-4
Marinis, T., Roberts, L., Felser, C., & Clahsen, H. (2005). Gaps in second language sentence
processing. Studies in Second Language Acquisition, 27, 53–78. DOI:
10.1017/S0272263105050035
Martin, C.D., Thierry, G., Kuipers, J.R., Boutonnet, B., Foucart, A., & Costa, A. (2013).
Bilinguals reading in their second language do not predict upcoming words as native
readers do. Journal of Memory and Language, 69(4), 574–588. DOI:
10.1016/[Link].2013.08.001
Mazuka, R., Jincho, N., & Oishi, H. (2009). Development of executive control and language
processing. Language and Linguistics Compass, 3, 59–89.
DOI: 10.1111/j.1749-818X.2008.00102.x
McKee, C. (1992). A comparison of pronouns and anaphors in Italian and English acquisition.
Language Acquisition, 2, 21–54. DOI: 10.1207/s15327817la0201_2
Momma, S., Sakai, H., & Phillips, C. (2015). Give me several hundred more milliseconds:
The temporal dynamics of verb prediction. Talk at the 28th annual CUNY Conference on
Human Sentence Processing. Los Angeles, CA.
Montrul, S., & Slabakova, R. (2003). Competence similarities between native and near-native
speakers. Studies in Second Language Acquisition, 25, 351–398.
DOI: 10.1017/S0272263103000159
Nation, K., Marshall, C.M., & Altmann, G. (2003). Investigating individual differences in
children’s real-time sentence comprehension using language-mediated eye movements.
Journal of Experimental Child Psychology, 86, 314–329. DOI:
10.1016/[Link].2003.09.001
Newport, E. (1990). Maturational constraints on language learning. Cognitive Science, 14,
11–28. DOI: 10.1207/s15516709cog1401_2
Newport, E.L., Gleitman, H., & Gleitman, L.R. (1977). Mother, I’d rather do it myself: Some
effects and non-effects of maternal speech style. In C.E. Snow & C.A. Ferguson (Eds.),
Talking to children: Language input and acquisition (pp. 109–149). Cambridge,
England: Cambridge University Press.
Nicol, J., & Swinney, D. (1989). The role of structure in coreference assignment during
sentence comprehension. Journal of Psycholinguistic Research, 18, 5–19. DOI:
10.1007/BF01069043
Nieuwland, M.S., & Martin, A.E. (2012). If the real world were irrelevant, so to speak: The
role of propositional truth-value in counterfactual sentence comprehension. Cognition,
122, 102–109. DOI: 10.1016/[Link].2011.09.001
Novick, J.M., Trueswell, J.C., & Thompson-Schill, S.L. (2005). Cognitive control and
parsing: Reexamining the role of Broca’s area in sentence comprehension. Cognitive,
Affective, & Behavioral Neuroscience, 5, 263–281. DOI: 10.3758/CABN.5.3.263
Omaki, A. (2010). Commitment and flexibility in the developing parser. PhD dissertation,
University of Maryland.
Omaki, A., Dyer, C., Malhotra, S., Sprouse, J., Lidz, J., & Phillips, C. (2007). The time course
of anaphoric processing and syntactic reconstruction. Talk at the 20th Annual CUNY
Conference on Human Sentence Processing. San Diego, California.
Omaki, A., Davidson White, I., Goro, T., Lidz, J., & Phillips, C. (2014). No fear of
commitment: Children’s incremental interpretation in English and Japanese wh-
 The role of language processing in language acquisition
449
questions. Language Learning and Development, 10, 206–233. DOI:
10.1080/15475441.2013.844048
Omaki, A., & Lidz, J. (2014). Linking parser development to acquisition of syntactic
knowledge. Language Acquisition. Published online July 17th 2014.
Omaki, A., & Schulz, B. (2011). Filler-gap dependencies and island constraints in second-
language sentence processing. Studies in Second Language Acquisition, 33, 563–588.
DOI: 10.1017/S0272263111000313
Pablos, L., Ruijgrok, B., Doetjes, J., & Cheng, L.L.-S. (submitted). Active search for
antecedents in cataphoric pronoun resolution. Ms. Leiden University.
Parker, D., & Phillips, C. (2014). Selective priority for structure in memory retrieval. Poster
presentation at the 27th Annual CUNY Conference on Human Sentence Processing.
Columbus, OH.
Patil, U., Vasishth, S., & Lewis, R. (2011). Early retrieval interference in syntax-guided
antecedent search. Talk at the 24th Annual CUNY Conference on Human Sentence
Processing. Palo Alto, CA: Stanford University.
Pearl, L., Goldwater, S., & Steyvers, M. (2010). How ideal are we? Incorporating human
limitations into Bayesian models of word segmentation. In K. Franich, K. Iserman, & L.
Keil (Eds), Proceedings of the 34th annual Boston University Conference on Child
Language Development (pp. 315–326). Somerville, MA: Cascadilla Press.
Pearl, L., & Sprouse, J. (2013). Syntactic islands and learning biases: Combining
experimental syntax and computational modeling to investigate the language acquisition
problem. Language Acquisition, 20, 23–68. DOI: 10.1080/10489223.2012.738742
Perfors, A. (2012). When do memory limitations lead to regularization? An experimental and
computational investigation. Journal of Memory and Language, 67, 486–506. DOI:
10.1016/[Link].2012.07.009
Perfors, A., Tenenbaum, J.B., & Regier, T. (2011). The learnability of abstract syntactic
principles. Cognition, 118, 306–338. DOI: 10.1016/[Link].2010.11.001
Phillips, C. (1996). Order and Structure.1996. PhD dissertation, MIT. Distributed by MIT
Working Papers in Linguistics.
Phillips, C. (2006). The real-time status of island phenomena. Language, 82, 795–823.
DOI: 10.1353/lan.2006.0217
Phillips, C. (2013). On the nature of island constraints II: Language learning and innateness.
In J. Sprouse & N. Hornstein (Eds.), Experimental syntax and island effects (pp. 132–
158).
Cambridge University Press. DOI: 10.1017/CBO9781139035309.007
Phillips, C., & Lewis, S. (2013). Derivational order in syntax: Evidence and architectural
consequences. Studies in Linguistics, 6, 11–47.
Phillips, C., & Wagers, M. (2007). Relating structure and time in linguistics and
psycholinguistics. In G. Gaskell (ed.), Oxford Handbook of Psycholinguistics (pp. 739–
756). Oxford University Press.
Phillips, C., Wagers, M.W., & Lau, E.F. (2011). Grammatical illusions and selective fallibility
in real-time language comprehension. In J. Runner (Ed.), Experiments at the interfaces
(Syntax and Semantics, Vol. 37) (pp. 147–180). Bingley, UK: Emerald Publications.
Pickering, M.J., & Garrod, S. (2007). Do people use language production to make predictions
during comprehension? Trends in Cognitive Sciences, 11, 105–110. DOI:
10.1016/[Link].2006.12.002
Pinker, S., Lebeaux, D.S., & Frost, L.A. (1987). Productivity and constraints in the
acquisition of the passive. Cognition, 26, 195–267. DOI: 10.1016/S0010-0277(87)80001-
X
450 Colin Phillips and Lara Ehrenhofer
Pozzan, L. & Trueswell, J.C. (2013). Online processing of English garden-path sentences by
L2 learners: A visual world study. Poster presented at the 26th annual CUNY Conference
on Human Sentence Processing. Columbia, SC.
Pullum, G.K. & Scholz, B.C. (2002). Empirical assessment of stimulus poverty arguments.
The Linguistic Review, 18, 9–50. DOI: 10.1515/tlir.19.1-2.9
Pylkkänen, L., & McElree, B. (2007). An MEG Study of Silent Meaning. Journal of
Cognitive Neuroscience, 1905–1921. DOI: 10.1162/jocn.2007.19.11.1905
Rayner, K., Carlson, M., & Frazier, L. (1983). The interaction of syntax and semantics during
sentence processing: Eye movements in the analysis of semantically biased sentences.
Journal of Verbal Learning and Verbal Behavior, 22, 358–374. DOI: 10.1016/S0022-
5371(83)90236-0
Roberts, L., & Felser, C. (2011). Plausibility and recovery from garden paths in second
language sentence processing. Applied Psycholinguistics, 32, 299–331.
DOI: 10.1017/S0142716410000421
Rodríguez, G.A. (2008). Second language sentence processing: Is it fundamentally different?
PhD dissertation, University of Pittsburgh.
Rohde, D.L., & Plaut, D.C. (1999). Language acquisition in the absence of explicit negative
evidence: How important is starting small? Cognition, 72, 67–109.
DOI: 10.1016/S0010-0277(99)00031-1
Rohde, D.L.T., & Plaut, D.C. (2003). Less is less in language acquisition. In P. Quinlan (Ed.),
Connectionist modelling of cognitive development (pp. 189–231). Hove, UK: Psychology
Press.
Runner, J.T., Sussman, R.S., & Tanenhaus, M.K. (2006). Processing reflexives and pronouns
in picture noun phrases. Cognitive Science, 30, 193–241. DOI:
10.1207/s15516709cog0000_58
Saffran, J.R. (2003). Statistical language learning mechanisms and constraints. Current
Directions in Psychological Science, 12, 110–114. DOI: 10.1111/1467-8721.01243
Snedeker, J. (2013). Children’s sentence processing. In R. van Gompel (Ed.), Sentence
Processing (pp. 189–220). New York, New York: Psychology Press.
Snedeker, J., & Trueswell, J.C. (2004). The developing constraints on parsing decisions: The
role of lexical-biases and referential scenes in child and adult sentence processing.
Cognitive Psychology, 49, 238–299. DOI: 10.1016/[Link].2004.03.001
Snow, C., & Hoefnagel-Höhle, M. (1978). The critical period for language acquisition:
Evidence from second language learning. Child Development, 49, 1114–1128. DOI:
10.2307/1128751 Snyder, W. (2007). Child language: The parametric approach. Oxford:
Oxford University Press.
Sorace, A. (2011). Pinning down the concept of “interface” in bilingualism. Linguistic
Approaches to Bilingualism, 1, 1–33. DOI: 10.1075/lab.1.1.01sor
Sorace, A., & Serratrice, L. (2009). Internal and external interfaces in bilingual language
development: Beyond structural overlap. International Journal of Bilingualism, 13, 195–
210. DOI: 10.1177/1367006909339810
Speer, S.R., & Ito, K. (2009). Prosody in first language acquisition–Acquiring intonation as a
tool to organize information in conversation. Language and Linguistics Compass, 3, 90–
110. DOI: 10.1111/j.1749-818X.2008.00103.x
Staub, A., & Clifton Jr, C. (2006). Syntactic prediction in language comprehension: Evidence
from either… or. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 32, 425. DOI: 10.1037/0278-7393.32.2.425
Stowe, L.A. (1986). Parsing WH-constructions: Evidence for on-line gap location. Language
and Cognitive Processes, 3, 227–245. DOI: 10.1080/01690968608407062
 The role of language processing in language acquisition
451
Sturt, P. (2003). The time-course of the application of binding constraints in reference
resolution. Journal of Memory and Language, 48, 542–562. DOI: 10.1016/S0749-
596X(02)00536-3
Sturt, P. (2007). Semantic re-interpretation and garden path recovery. Cognition, 105, 477–
488. DOI: 10.1016/[Link].2006.10.009
Tanenhaus, M.K., Spivey-Knowlton, M.J., Eberhard, K.M., & Sedivy, J.C. (1995). Integration
of visual and linguistic information in spoken language comprehension. Science, 268,
1632– 1634. DOI: 10.1126/science.7777863
Tanner, D., Nicol, J., Herschensohn, J., & Osterhout, L. (2012). Electrophysiological markers
of interference and structural facilitation in native and nonnative agreement processing.
In A.K. Biller, E.Y. Chung, & A.E. Kimball (Eds.), Proceedings of the 36th Boston
University Conference on Language Development (pp. 594–606). Somerville, MA:
Cascadilla Press.
Thornton, R., & Wexler, K. (1999). Principle B, VP ellipsis, and interpretation in child
grammar. Cambridge, MA: MIT Press.
Townsend, D.J., & Bever, T.G. (2001). Sentence comprehension: The integration of habits
and rules. Cambridge, MA: MIT Press.
Traxler, M.J., & Pickering, M.J. (1996). Plausibility and the processing of unbounded
dependencies: An eye-tracking study. Journal of Memory and Language, 35, 454–475.
DOI: 10.1006/jmla.1996.0025
Trueswell, J.C., Sekerina, I., Hill, N.M., & Logrip, M.L. (1999). The kindergarten-path effect:
Studying on-line sentence processing in young children. Cognition, 73, 89–134. DOI:
10.1016/S0010-0277(99)00032-3
Unsworth, S., Gualmini, A., & Helder, C. (2008). Children’s interpretation of indefinites in
sentences containing negation: A reassessment of the cross-linguistic picture. Language
Acquisition, 15, 315–328. DOI: 10.1080/10489220802338742
Urbach, T.P. & Kutas, M. (2010). Quantifiers more or less quantify on-line: ERP evidence for
partial incremental interpretation. Journal of Memory and Language, 63, 158–179. DOI:
10.1016/[Link].2010.03.008
Van Berkum, J.J. (2009). The neuropragmatics of ‘simple’ utterance comprehension: An ERP
review. In U. Sauerland, & K. Yatsushiro, (Ed.), Semantics and pragmatics: From
experiment to theory (pp. 276–316). Basingstoke, UK: Palgrave Macmillan.
Van Berkum, J.J., Brown, C.M., Zwitserlood, P., Kooijman, V., & Hagoort, P. (2005).
Anticipating upcoming words in discourse: evidence from ERPs and reading times.
Journal of Experimental Psychology: Learning, Memory, and Cognition, 31, 443.
DOI: 10.1037/0278-7393.31.3.443 van Gompel, R.P.G., & Liversedge, S.P. (2003). The
influence of morphological information on cataphoric pronoun assignment. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 29, 128–139. DOI:
10.1037/0278-7393.29.1.128
Van Petten, C., & Kutas, M. (1990). Interactions between sentence context and word
frequency in event-related brain potentials. Memory & Cognition, 18, 380–393. DOI:
10.3758/BF03197127
Varlokosta, S. (2000). Lack of clitic-pronoun distinctions in the acquisition of Principle B in
child Greek. In S.C. Howell, S.A Fish, & T. Keith-Lucas (Eds.), Proceedings of the 24th
Annual Boston University Conference on Language Development (pp. 738–748).
Somerville, MA: Cascadilla Press.
Vigliocco, G., & Nicol, J. (1998). Separating hierarchical relations and word order in
language production: Is proximity concord syntactic or linear?. Cognition, 68, B13–B29.
DOI: 10.1016/S0010-0277(98)00041-9
452 Colin Phillips and Lara Ehrenhofer
Viau, J., Lidz, J., & Musolino, J. (2010). Priming of abstract logical representations in 4-year-
olds. Language Acquisition, 17, 26–50. DOI: 10.1080/10489221003620946
Wagers, M.W., Lau, E.F., & Phillips, C. (2009). Agreement attraction in comprehension:
Representations and processes. Journal of Memory and Language, 61, 206–237. DOI:
10.1016/[Link].2009.04.002
Wang, L., Bastiaansen, M., Yang, Y., & Hagoort, P. (2012). Information structure influences
depth of syntactic processing: Event-related potential evidence for the Chomsky illusion.
PloS ONE, 7, e47917. DOI: 10.1371/[Link].0047917
Weckerly, J., Wulfeck, B., & Reilly, J. (2004). The development of morphosyntactic ability in
atypical populations: The acquisition of tag questions in children with early focal lesions
and children with specific-language impairment. Brain and Language, 88, 190–201.
DOI: 10.1016/S0093-934X(03)00098–1
Weighall, A.R. (2008). The kindergarten path effect revisited: Children’s use of context in
processing structural ambiguities. Journal of Experimental Child Psychology, 99, 75–95.
DOI: 10.1016/[Link].2007.10.004
Yang, C. (2002). Knowledge and learning in natural language. Oxford University Press.
Yoshida, M., Kazanina, N., Pablos, L., & Sturt, P. (2014). On the origin of islands. Language,
Cognition and Neuroscience, 29, 761–770. DOI: 10.1080/01690965.2013.788196
Xiang, M., Dillon, B., & Phillips, C. (2009). Illusory licensing effects across dependency
types: ERP evidence. Brain and Language, 108, 40–55. DOI:
10.1016/[Link].2008.10.002
Zukowski, A., McKeown, R., & Larsen, J. (2008). A tough test of the locality requirement for
reflexives. In H. Chan, H. Jacob, & E. Kapia (Eds.), Proceedings of the 32nd Annual
Boston University Conference on Language Development (pp. 586–597). Somerville,
MA: Cascadilla
Press.
Lara Ehrenhofer
Department of Linguistics
Author’s addresses University of Maryland 1401
Colin Phillips Marie Mount Hall
Department of Linguistics College Park, MD 20742
University of Maryland 1401 USA
Marie Mount Hall
ehrenhof@[Link]
College Park, MD 20742
USA
colin@[Link]
Copyright of Linguistic Approaches to Bilingualism is the property of John Benjamins
Publishing Co. and its content may not be copied or emailed to multiple sites or posted to a
listserv without the copyright holder's express written permission. However, users may print,
download, or email articles for individual use.