Content uploaded by Holger Hopp
Author content
All content in this area was uploaded by Holger Hopp on Oct 23, 2023
Content may be subject to copyright.
Annual Review of Linguistics
Second Language Sentence
Processing
Holger Hopp
Department of English and American Studies, University of Braunschweig, Braunschweig,
Germany; email: h.hopp@tu-braunschweig.de
Annu. Rev. Linguist. 2022. 8:235–56
The Annual Review of Linguistics is online at
linguistics.annualreviews.org
https://doi.org/10.1146/annurev-linguistics-030821-
054113
Copyright © 2022 by Annual Reviews.
All rights reserved
Keywords
second language, bilingualism, sentence comprehension, parsing, grammar
Abstract
Second language (L2) sentence processing research studies how adult L2
learners understand sentences in real time. I review how L2 sentence pro-
cessing differs from monolingual rst-language (L1) processing and outline
major ndings and approaches. Three interacting factors appear to man-
date L1–L2 differences: (a) capacity restrictions in the ability to integrate
information in an L2; (b) L1–L2 differences in the weighting of cues, the
timing of their application, and the efciency of their retrieval; and (c) varia-
tion in the utility functions of predictive processing. Against this backdrop, I
outline a novel paradigm of interlanguage processing, which examines bilin-
gual features of L2 processing, such as bilingual language systems, nonse-
lective access to all grammars, and processing to learn an L2. Interlanguage
processing goes beyond the traditional framing of L2 sentence processing
as an incomplete form of monolingual processing and reconnects the eld
with current approaches to grammar acquisition and the bilingual mental
lexicon.
Annu. Rev. Linguist. 2022.8:235-256. Downloaded from www.annualreviews.org
Access provided by WIB6277 - TU Braunschweig on 05/31/23. For personal use only.
1. INTRODUCTION
When comprehending sentences in real time, language users rapidly integrate incoming input to
project syntactic relations and structure and map these to interpretation. Research in second lan-
guage (L2) sentence processing investigates how non-native L2 users, typically adult L2 learners,
who have learned an L2 later in life, differ in real-time comprehension from monolingual rst-
language (L1) speakers who acquired a language from birth onward. On top of providing insights
into how the language processing system handles an additional, typically less-used language, L2
processing research explores the extent to which differences in language processing, rather than
in linguistic knowledge, between L1 and L2 speakers can account for characteristic features of late
(second) language acquisition, such as slow and variable development as well as non-native-like ul-
timate attainment (Bley-Vroman 1988). Broadly speaking, L2 sentence processing can differ from
L1 processing in four ways.
First, effects of the L1 could affect L2 processing in that L2 learners transfer process-
ing strategies based on the L1 or recruit L1-based grammatical knowledge when parsing
the L2, which could lead to partially different processing proles compared to those of
monolinguals.
Second, adult L2 learners process the L2 in a noisier cognitive architecture, which may lead
to slower, more effortful, and more error-prone processing. In particular, parsing a less-
used L2 places larger demands on a limited capacity parser so that restrictions in working
memory, slower processing speed, and a reduced ability to integrate different information
types in real time may compromise the ability of L2 learners to effect complete parses.
Third, L2 processing could be qualitatively different from L1 processing in that L2 learners
weight processing constraints differently or rely on less-detailed processing heuristics.
Fourth, L2 processing may be fundamentally distinct from L1 parsing in that maturational
constraints render the grammatical processing system used by native speakers inaccessible
and force late L2 learners to recruit explicit knowledge in sentence comprehension.
Although obviously not mutually exclusive, these options have given rise to a number of ap-
proaches to L2 sentence processing that build on models familiar from research on monolingual
sentence processing.
Capitalizing on the rst two potential differences, identity-oriented approaches posit that the
linguistic representations and the relevant neurocognitive systems are fully engaged by L1 and
L2 users alike, and that L1–L2 differences reduce to performance differences or effects from the
L1. Specically, L2 users may not fully compute and complete all processing steps because of
the generally higher cognitive demands of processing an L2 (e.g., Hopp 2010, McDonald 2006),
the more costly processing for particular types of information at linguistic interfaces (the Interface
Hypothesis; Sorace 2011), a reduced capacity to recruit predictive mechanisms in processing [the
Reduced Ability to Generate Expectations (RAGE) Hypothesis; Grüter et al. 2017, Kaan 2014],
and/or cross-linguistic interference and slower retrieval from the bilingual mental lexicon (the
Lexical Bottleneck Hypothesis; Hopp 2018). In appealing to the additional demands of processing
a non-native language, this set of identity-oriented models refers to capacity models of language
processing (e.g., Just & Carpenter 1992).
In contrast, difference-oriented accounts focus on the nal two options and claim that L2 pro-
cessing differs from L1 processing with respect to the detail of linguistic representations used or
the neurocognitive (memory) systems recruited during real-time comprehension. For instance,
the Shallow Structure Hypothesis (Clahsen & Felser 2006, 2018) refers to dual-pathway models
of monolingual sentence processing, such as Good-Enough processing (Ferreira & Patson 2007),
Hopp
Annu. Rev. Linguist. 2022.8:235-256. Downloaded from www.annualreviews.org
Access provided by WIB6277 - TU Braunschweig on 05/31/23. For personal use only.
and argues that L2 users predominantly rely on nongrammatical processing heuristics instead
of engaging in structure-based syntactic processing. Other approaches hold that L2 learners are
more susceptible to interfering cues than monolinguals are (Cunnings 2017), as captured by cue-
based retrieval models of sentence processing (e.g., McElree 2000). An even stronger position is
espoused by the Declarative/Procedural Model, which claims that adult L2 learners process syn-
tactic dependencies and nonlocal agreement in different neurocognitive systems using declarative
rather than procedural memory (Ullman 2016).
To explore the scope of L1–L2 differences, in this review I survey the main strands of research
in L2 sentence processing, focusing on grammatical phenomena for which adult L2 processing has
been found to differ from L1 processing. Across these areas, L2 sentence processing displays all
the hallmarks of L1 processing. Against this backdrop, I argue that the predominant concern with
explaining differences between L1 and L2 processing should be abandoned—not least because it
invariably conceptualizes L2 processing as a decient or deviant form of monolingual processing.
In doing so, it glosses over “bilingual” features of L2 processing, such as the constant coactivation
of two languages. Reorienting L2 processing to focus on these bilingual features will allow the eld
to address fundamental architectural questions about relations between grammar and the parser
and will provide insights into the role of language processing for learning. With these questions
in mind, I suggest widening the perspective toward interlanguage processing, and I sketch some
promising lines of future inquiry in broad brushstrokes.
2. TOPICS AND MODELS OF L2 SENTENCE PROCESSING
In real-time sentence comprehension, monolingual speakers rapidly integrate bottom-up informa-
tion extracted from linguistic items encountered in the speech stream or text input with top-down
contextual information or world knowledge. This integration has been modeled as the interaction
of (a) universal and language-particular linguistic constraints, (b) processing constraints reecting
the architecture of the parser including restrictions in cognitive capacities (e.g., working memory),
and (c) experience-based constraints indexing language use (for review, see MacDonald & Hsiao
2018, van Gompel 2013).
Linguistic research on L2 sentence processing predominantly investigates how adult L2 learn-
ers may differ from monolingual natives in their use of (a) linguistic and (b) processing constraints,
while processing differences owing to (c) experiential factors have mainly been explored within
usage-based frameworks (for review,see Sagarra 2022).With an emphasis on behavioral research,
I review three core areas in research on L2 sentence processing: the processing of temporary syn-
tactic ambiguities, the recruitment of structure-based syntactic constraints, and the use of mor-
phosyntactic information in agreement relations, in particular for predictive processing (for a re-
view of neurophysiological L2 research on sentence processing, see van Hell & Tokowicz 2010).
2.1. Temporary Syntactic Ambiguities, Reanalysis, and the Integration
of Information
Like L1 adults, L2 learners incrementally integrate each incoming element into a partial and un-
folding interpretation when comprehending sentences in their L2. As a consequence, L2 sentence
comprehension is also subject to misanalyses and reanalyses in the processing of temporarily
ambiguous sentences, such as object–subject ambiguities in sentences like While the band played
the song pleased all the customers. L2 readers are led down the garden path in initially interpreting
the postverbal noun phrase (i.e., the song) as an object. Subsequently, they show slowdowns on
the main clause verb, indicating that they engage in syntactic reanalysis of the noun phrase as
being the subject of the main clause ( Juffs & Harrington 1996). These effects hold irrespective
www.annualreviews.org •Second Language Sentence Processing
Annu. Rev. Linguist. 2022.8:235-256. Downloaded from www.annualreviews.org
Access provided by WIB6277 - TU Braunschweig on 05/31/23. For personal use only.
of L1 differences in word order ( Juffs 2004), suggesting that L2 learners apply grammatical L2
properties to guide parsing. Moreover, L2 speakers use other types of L2 information for the
resolution of syntactic ambiguities, such as verb transitivity (Hopp 2015a), frequency information
of verbal complements, so-called verb biases (e.g., Dussias & Cramer Scaltz 2008, Lee et al. 2013),
plausibility (e.g., Hopp 2015a, Roberts & Felser 2011, Williams et al. 2001), discourse context
(e.g., Pan & Felser 2011), and prosody (e.g.,Dekydtspotter et al. 2008, Nickels et al. 2013). In these
respects, the L1 and L2 processing of temporary ambiguities appear to be broadly comparable.
At the same time, L2 readers struggle with reanalysis more than native speakers do (Roberts
& Felser 2011). For instance, Jacob & Felser (2016) tested L1 German learners of English on
sentences like example 1 that either marked clause boundaries using commas or lacked commas,
allowing for an initial object reading of the postverbal noun:
(1) While the gentleman was eating(,) the burgers were still being reheated in the microwave.
In example 1, the auxiliary in the main clause (were) forces reanalysis of the noun phrase (the
burgers) to a subject reading. However, both native controls and the L2 group continued to read
the following regions more slowly in the garden-path conditions compared to the comma con-
ditions; this result suggests that the syntactic disambiguation did not erase the initial analysis of
the burgers as an object, creating semantic persistence. In this and other studies, L2 learners often
demonstrated larger and longer-lasting semantic persistence than did monolinguals (Pozzan &
Trueswell 2016, yet see Fujita & Cunnings 2020). To pinpoint the nature of reanalysis difculty,
¸Safak & Hopp (2021) manipulated the strength of the initial garden path in terms of the verb
bias toward an object reading of the postverbal noun. Both English monolinguals and L1 Turkish
and L1 German advanced L2 learners showed different strengths of garden-pathing according to
verb bias. Unlike in English monolinguals, though, garden-path strength did not modulate later
semantic persistence in advanced L2 readers, which suggests that the L2 learners could not in-
tegrate earlier verb-bias information when recovering from garden paths. Such a reduced facility
to integrate different types of information for reanalysis in real-time processing is characteristic
even of highly procient L2 learners.
For instance, L2 learners do not reliably use case marking for word order revisions in process-
ing exible word order languages (e.g., Hopp 2006, Jackson 2008), and they recruit plausibility
information more slowly for reanalysis than natives do (Dussias & Piñar 2010, Jessen & Felser
2019). Importantly, individual differences among L2 learners affect the extent of how reliably L2
learners integrate different information types in L2 sentence comprehension. For one thing, L1–
L2 similarity in the type of information that needs to be integrated facilitates reanalysis in the L2
in that, for instance, L2 learners whose L1 has case marking outperform L2 learners whose L1s do
not mark case (Frenck-Mestre et al. 2019, Hopp 2010). For another, learners are more native-like
in the processing of temporary ambiguities at higher L2 prociency (Hopp 2010, Hoshino et al.
2010), as well as if they have larger cognitive resources in working memory (Dussias & Piñar 2010),
faster processing speeds (Kaan et al. 2015), and/or greater automaticity in lower-level linguistic
processing, such as lexical access (Hopp 2014). Conversely, L1 speakers sometimes show integra-
tion difculties similar to those of L2 learners when processing the L1 under adverse conditions,
for instance, at higher speeds or in noisy conditions (e.g., Hopp 2010, López Prego & Gabriele
2014, McDonald 2006). Further, sentence processing in L2 learners often resembles processing
in other language learners, such as child L1 learners (Trueswell et al. 1999), or in users with re-
stricted resources, such as people with aphasia (Haarman et al. 1997) or L1 adults with lower
working-memory capacities (e.g., Havik et al. 2009).
These parallels, including studies demonstrating native-like processing for some L2 learners,
suggest a large degree of identity or continuity between L1 and L2 processing. However, the
Hopp
Annu. Rev. Linguist. 2022.8:235-256. Downloaded from www.annualreviews.org
Access provided by WIB6277 - TU Braunschweig on 05/31/23. For personal use only.
inherently greater load in processing a less-used L2 compared to an L1 may lead L2 learners to
adopt economy principles in parsing that minimize dependency length (e.g., Gibson 1998) and
curtail the use of multiple information types (e.g., McRae & Matsuki 2013), so that L2 learners
do not always effect full parses.
In all, then, capacity limitations when processing a late-learned L2 can explain delayed or at-
tenuated reanalysis or failures to integrate different information types during real-time compre-
hension of the L2. At the same time, not all information types are equally hard to integrate in
L2 sentence processing: L2 learners easily accommodate lexical-semantic or discourse informa-
tion in resolving temporary ambiguities, sometimes even to larger degrees than native speakers
do ( Jackson & Roberts 2010, Pan & Felser 2011). In contrast, the integration of syntactic and
morphosyntactic agreement information is challenging in L2 processing, even for advanced L2
learners. I turn to these aspects now.
2.2. Syntactic Constraints in the L2 Processing of Nonlocal Dependencies
Research on nonlocal dependencies centers on the use of syntactic constraints in ller-gap de-
pendencies (e.g., wh-questions, relative clauses) and in anaphoric dependencies (e.g., reexives,
pronouns), in L2 processing. I discuss each in turn.
2.2.1. Filler-gap dependencies and shallow structures. When interpreting ller-gap depen-
dencies like wh-questions, comprehenders need to relate llers to their lexical subcategorizer for
interpretation. In monolingual processing, the parser immediately initiates a search for a potential
gap site for a ller, such as a wh-word, postulating a gap as soon as possible (Active Filler Strategy;
Clifton & Frazier 1989). In consequence, processing slows down when the parser encounters a
gap that is lled by an overt element (lled-gap effect; Stowe 1986). L2 learners also show lled-
gap effects, suggesting that they similarly engage in “active” gap-lling and relate llers to the
earliest available lexical licensors, such as subcategorizing verbs or prepositions (Dallas & Kaan
2008, Jessen & Felser 2019, Williams et al. 2001). To test syntactic constraints on gap lling, sev-
eral studies probed whether native and non-native speakers allow for wh-extraction from syntactic
islands, which grammatically prohibit extraction (for review, see Chaves & Putnam 2021). Across
studies using different methods (self-paced reading: Omaki & Schulz 2011; eye-tracking: Felser
et al. 2012) or diagnostics (Aldwayan et al. 2010,Kim et al. 2015a),L2 learners consistently obey
syntactic locality constraints on ller-gap dependency formation, even if they experience delays
(Boxell & Felser 2017) or L1 differences in the timing of their application (Kim et al. 2015a). In
principle, then, L2 learners apply syntactic locality constraints incrementally in processing the L2.
A related line of research investigates whether L2 learners posit structurally required gaps in
processing. Marinis et al. (2005) tested whether advanced L2 learners with various L1s would show
facilitation in ller integration in sentences like example 2a, where the clausal complement projects
an intermediate gap preceding the complementizer that. They contrasted self-paced reading times
at the nal gap position in example 2a with matched sentences that do not have intermediate gaps
(example 2b):
(2a) The nurse who the doctor argued [ that the rude patient had angered ] is refusing
to work late. (extraction across CP)
(2b) The nurse who the doctor’s [argument about the rude patient had angered ] is refusing
to work late. (extraction across NP)
For L1 readers, ller integration at the nal gap was facilitated when they could previously reacti-
vate it at the intermediate gap site in example 2a. In contrast, none of the L2 groups demonstrated
www.annualreviews.org •Second Language Sentence Processing
Annu. Rev. Linguist. 2022.8:235-256. Downloaded from www.annualreviews.org
Access provided by WIB6277 - TU Braunschweig on 05/31/23. For personal use only.
reading differences between examples 2a and b. This nding suggests that learners do not posit
hierarchical syntactic structure during sentence processing (see also Felser & Roberts 2007).
Such L1–L2 differences in the processing of ller-gap dependencies point to differences be-
tween L1 and L2 users in how much they recruit syntactic information in real time. As a con-
sequence, the Shallow Structure Hypothesis (Clahsen & Felser 2006, 2018) holds that adult L2
learners rely less than native speakers do on syntactic parses and predominantly build sentence
representations on lexical-thematic, semantic, and discourse information and/or surface heuris-
tics, such as high-frequency word orders. Initially, the hypothesis tied the underreliance on gram-
matical parses to age constraints on acquiring abstract syntactic representations (Clahsen & Felser
2006). However, highly procient learners with long L2 immersion experience display native-like
processing of intermediate gaps (Pliatsikas & Marinis 2013), and even nonimmersed L2 learn-
ers have been found to posit gaps in studies that used more sensitive eye-tracking measures like
pupillometry (Fernandez et al. 2018). Such instances of native-like L2 processing of abstract syn-
tax underscore that shallow processing among L2 learners is not absolute but only a matter of
degree, and may reect the delayed application or a reduced weighting of grammatical versus
nongrammatical information in L2 sentence processing (Clahsen & Felser 2018).
2.2.2. Referential processing and cue interference. Referential dependencies involving the
interpretation of reexives and pronouns provide insights into the different types of constraints L2
learners use in sentence comprehension, specically, whether they prioritize discourse-based re-
lations or adhere to syntactic constraints on coreference, such as the principles of Binding Theory
(Chomsky 1981). Although the degree to which monolingual adults immediately apply binding
constraints in referential processing is contested (for review, see Dillon 2014), several studies have
reported that syntactic constraints on coreference restrict coreference in the initial processing of
anaphoric dependencies (Sturt 2003). For L2 learners, Felser & Cunnings (2012) explored this
issue by manipulating coreference options between a syntactic binder and a discourse-prominent
coreferent in terms of stereotypical gender as in example 3. In example 3, the reexive pronoun
either only had one likely antecedent (soldier) or had two possible antecedents ( James/He,soldier),
with the syntactically accessible binder (soldier) being either the local (example 3a) or the nonlocal
(example 3b) antecedent:
(3a) James/Helen has worked at the army hospital for years. He/She noticed that the soldier had
wounded himself/herself while on duty in the Far East.
(3b) James/Helen has worked at the army hospital for years. The soldier that he/she treated on the
ward wounded himself/herself while on duty in the Far East.
For both sentence types, L1 German learners of English showed slowdowns in early eye-tracking
measures if the syntactically inaccessible discourse-prominent coreferent (he versus she) mis-
matched in gender with the reexive, while native English readers were sensitive to gender mis-
matches between the binder with the reexive (himself versus herself ). The L2 speakers only be-
came sensitive to the local binding mismatch in later rereading times on the reexive, which
led Felser & Cunnings (2012) to conclude that L2 learners initially construe coreference with
discourse-prominent antecedents. Research on restrictions on the local binding of personal pro-
nouns is compatible with this assumption, as L2 readers initially only consider salient nonlocal
antecedents, even when a local binder for the pronoun is available (Kim et al. 2015b, Patterson
et al. 2014).
At rst glance, such partially non-native-like L2 processing of referential dependencies
seems difcult to reconcile with the sensitivity L2 learners show for syntactic restrictions on
the formation of ller-gap dependencies (Section 2.2.1). Yet, Felser (2015) noted that ller-gap
Hopp
Annu. Rev. Linguist. 2022.8:235-256. Downloaded from www.annualreviews.org
Access provided by WIB6277 - TU Braunschweig on 05/31/23. For personal use only.
dependencies differ from anaphoric dependencies in that ller-gap relations constitute forward-
looking dependencies, while the processing of anaphors involves backward-looking dependencies.
Building on this distinction, Cunnings (2017) proposed to account for L1–L2 differences in the
context of retrieval-based models of comprehension (e.g., Lewis et al. 2006). Establishing in-
trasentential anaphoric dependencies requires access to memory representations of potential
antecedents and of partial sentence representations built previously. In cue-based parsing, all
representations encoded in memory compete for selection, and the item that best matches the
cues needed is selected. Critically, cue-based retrieval is subject to similarity-based interference.
For L2 processing, Cunnings (2017) stipulated that retrieval from memory is subject to greater
cue interference, especially from discourse cues. Beyond accounting for greater L2 susceptibility
to discourse antecedents in referential processing, Cunnings’s (2017) approach captures stronger
semantic persistence effects among L2 learners in the reanalysis of garden-path sentences (Section
2.1). Under a cue-based account, semantic persistence effects reect interference from the initial
misparse that lingers in memory and is erroneously retrieved. However, to date there is little
direct evidence that L2 learners suffer greater cue interference than do natives, for instance, in
the processing of agreement (Tanner et al. 2012).
In sum, research on syntactic constraints in L2 sentence processing suggests that L2 users
struggle in constructing nonlocal grammatical dependencies in real time and initially tend to map
a sentence to interpretation using nongrammatical cues, such as lexical-semantic or discourse in-
formation. Recent work indicates that the relative underreliance on grammar is a matter of timing
and degree rather than a qualitative difference that sets L2 processing apart from L1 sentence
comprehension. In this sense, current theorizing within difference-oriented approaches points to
some convergence with identity-oriented approaches by relating difculties with grammatical in-
formation to the less robust encoding and delayed retrieval of morphosyntactic information in
real time.
2.3. Agreement and Lower Degrees of Predictive Processing
A more recent line of research shifts attention from investigating the types of information that are
particularly challenging for L2 users to the processes that may be subject to L1–L2 differences. On
top of integrating information into an ongoing parse, monolinguals use bottom-up and top-down
information to predict how an incoming sentence will continue to unfold. The clearest evidence
for predictive processing comes from event-related potential (ERP) studies and eye-tracking stud-
ies employing the visual-world paradigm (Huettig et al. 2011), which can measure predictive pro-
cessing before readers or listeners encounter the predicted element (for review, see Pickering &
Gambi 2018). In these studies, linguistic information restricts the set of eligible referents that can
follow. For instance, in a sentence like The boy will eat..., the combination of an agentive subject
and the verb semantics restricts the possible complement of the verb to edible objects (Altmann
& Kamide 1999). In other languages, grammatical information on articles can restrict the set of
upcoming referents to members of a grammatical gender class [e.g., Spanish: Encuentra la...(Find
theFEM →NFEM)] or, for case, specify the syntactic function of upcoming nouns [e.g., German:
Den Jungen sieht... (TheACC boy sees →NNOM)]. L2 learners appear to differ from L1 speak-
ers in the degree to which they actively predict upcoming information (Kaan 2014). Moreover,
predictive processing differs according to the linguistic domain among L2 learners.
On the one hand, even intermediate-level L2 learners robustly use lexical semantic information
for prediction, for instance, by restricting their looks to edible objects after they have encountered
the verb eat in visual-world studies (e.g., Chambers & Cooke 2009, Dijkgraaf et al. 2017, Hopp
2015b, Ito et al. 2018). When there are differences between monolinguals and L2 learners in
semantic prediction, they predominantly reect experiential variation between groups in the type
www.annualreviews.org •Second Language Sentence Processing
Annu. Rev. Linguist. 2022.8:235-256. Downloaded from www.annualreviews.org
Access provided by WIB6277 - TU Braunschweig on 05/31/23. For personal use only.
and amount of input. For instance, L1 effects emerge for ne-grained semantic predictions,such as
in the domain of caused motion involving placement events (van Bergen & Flecken 2017). Unlike
English with its cover-all verb put, Dutch and German use different verbs for placing objects
into upright (zetten/stellen) or horizontal (leggen/legen) positions. When encountering placement
verbs, native speakers of Dutch immediately orient their gazes to objects that match the verb’s
spatial semantics. Among L2 speakers, only L1 German learners of Dutch, but not L1 French or
L1 English learners, demonstrated predictive use of the verb’s specic selectional requirements.
This nding suggests that L2 learners rely on L1-based distinctions. In addition, semantic and
associative predictions tend to be more local among L2 learners than among monolingual adults.
For instance, when listening to a sentence like The pirate chases the ship, L2 learners briey look at
the display of a locally coherent referent (chase →cat) even though this referent does not match
the global expectation (Peters et al. 2018). Yet, more highly procient L2 learners who have more
L2 experience make fewer locally coherent predictions, and even monolingual children and adults
show similar patterns of local or attenuated prediction, especially when they process the sentences
under increased cognitive load (Ito et al. 2018).
On the other hand, even advanced L2 learners continue to have difculties in using syntactic
structure (Kaan et al. 2016) and agreement information for prediction, such as grammatical gender
marking (Grüter et al. 2012, Lew-Williams & Fernald 2010) and case marking (Hopp 2015b,
Mitsugi & MacWhinney 2016). For instance, near-native L1 English learners of Spanish do not
employ gender marking on articles to predict upcoming familiar nouns even though they are
highly accurate in production and in ofine tasks on gender agreement (Grüter et al. 2012). As
with semantic prediction, successful morphosyntactic prediction in the L2 partially depends on
the L1 in that adult L2 learners are more target-like when their L1s have similar gender systems
(e.g., Morales et al. 2016) or employ case distinctions (Frenck-Mestre et al. 2019). However, unlike
for semantic predictions, having an analogous morphosyntactic feature in the L1 does not always
entail native-like processing of gender in the L2 (Dussias et al. 2013).
Some accounts, such as the RAGE Hypothesis (Grüter et al. 2017), argue that predictive pro-
cessing is attenuated in an L2 (see also Kaan 2014). Reduced predictive processing is by no means
limited to L2 users, though, as the degree of predictive processing differs between individuals or
situations. For instance, elderly monolinguals engage less in predictive processing than middle-
aged adults do (Federmeier 2007), as do adults with low literacy skills (Mishra et al. 2012) or
dyslexia (Huettig & Brouwer 2015). In addition, L1 adults adapt their predictions according to
the usefulness of cues available for prediction in a particular context (Henry et al. 2017). In this
vein, Kuperberg & Jaeger (2016) construe prediction in terms of its utility function, claiming that
comprehenders maximize the utility of predictive processing by preactivating information at those
levels of representation only to the extent that it serves their goal of facilitating comprehension.
While the utility of prediction may be different relative to the task and situation of the comprehen-
der,the utility of prediction in an L2 context can also vary depending on the linguistic knowledge
of the learner.
For instance, predictive processing in the L2 may be weighted differently in utility when var-
ious cues compete. Grüter et al. (2020) examined competition between grammatical class mem-
bership and semantic prototypicality in the predictive use of Mandarin classiers. Unlike native
Mandarin speakers, advanced L2 learners showed anticipatory looks to semantically matching
nonclass competitors, which suggests that the learners prioritize semantic over grammatical cues
(see also Hopp 2015b). Critically, differences between L1 and L2 processing manifested not in
less or slower prediction among the L2 learners but rather in qualitatively different predictions
in that L2 learners assigned the high reliability of semantic cues a greater utility than less trans-
parent grammatical cues. In a similar vein, Hopp (2013,2016) found that the degree of predictive
Hopp
Annu. Rev. Linguist. 2022.8:235-256. Downloaded from www.annualreviews.org
Access provided by WIB6277 - TU Braunschweig on 05/31/23. For personal use only.
processing of gender agreement was related to variability in the lexical knowledge of grammatical
gender. Even after L2 learners have acquired the gender of most nouns, their lexicon contains
nouns with unknown, variable, or incorrect gender. Since many predictions based on variable or
incorrect lexical gender representations would lead to prediction error and require revision, us-
ing gender for prediction may be too costly. In Hopp (2013, 2016), L2 learners with less than
perfect lexical knowledge of gender did not use gender predictively even though they produced
the gender agreement correctly. Only learners with consistent gender knowledge demonstrated
predictive gender processing. Critically, native speakers also suspended prediction by gender once
gender became an unreliable cue in the course of an experiment that was designed to emulate
the variability typical of the L2 lexicon by including nouns with nontarget gender marking in the
input to participants (Hopp 2016).
What these ndings suggest is that predictive processing in both L2 and L1 processing varies
as a function of its utility compared to the costs of making unreliable or erroneous predictions (see
also Grüter & Rohde 2021). When predictions keep going amiss, the parser may wait and integrate
the input rather than engage in making a prediction. In consequence, L1–L2 differences in the
degree of prediction may follow from differences in the likelihood of making false predictions.
As research on prediction in L2s accumulates, it appears that L1–L2 differences in predictive
agreement processing may not necessarily index a reduced availability of predictive mechanisms,
their limited scope, or their slower deployment. Instead,they may point to differences in the utility
function of prediction relative to the linguistic knowledge of L2 learners.
2.4. L1–L2 Differences in Sentence Processing: Conclusions and Challenges
In all fundamental respects, processing a sentence in an L2 is identical to monolingual sentence
processing. Late L2 learners process L2 sentences incrementally, and they integrate and predict
linguistic and nonlinguistic information. Across the three areas surveyed above, both the integra-
tion and prediction of nongrammatical information are robust in L2 learners; yet, L2 learners
tend to make attenuated use of grammatical information and prediction compared to monolin-
guals. From this review, it becomes clear that the reasons for these specic L1–L2 differences are
unlikely to be monolithic. Instead, the major lines of research reviewed above suggest multiple
causes, including (a) computational restrictions in the ability to integrate information or to effect
full parses including reanalysis in an L2; (b) L1–L2 differences in the weighting of information
or cues, the timing of their application, or the efciency of their retrieval; and (c) variation in
the utility functions of particular processing mechanisms in L2 versus L1 processing. Currently,
research in L2 sentence processing aims to disentangle the relative contributions of these causes
and, in particular, the extent to which they interact with individual differences in working memory
(Linck et al. 2014), executive function (e.g., Teubner-Rhodes et al. 2016), and L2 experience and
prociency (Sagarra 2022). As the eld increasingly adopts methods with higher temporal reso-
lution, like eye-tracking and electroencephalography, studies can pose more nuanced questions
about individual differences as well as between-group L1–L2 differences in terms of the timing
and weighting of different constraints (Felser 2019).
At the same time, such shifts to timing and weighting issues foreground differences across stud-
ies in how ndings are interpreted when L2 learners are delayed or less robust in their sensitivity
to some constraint compared to monolinguals. On the one hand, when L2 learners demonstrate
some sensitivity to a constraint during real-time processing at all, one may conclude that L2 sen-
tence processing is qualitatively identical to native processing, and that delays or minor differences
in performance are essentially noise, owing to factors exogenous to the parser (e.g., individual and
between-group differences in cognitive capacity, lower-level processing skills, inexperience with
www.annualreviews.org •Second Language Sentence Processing
Annu. Rev. Linguist. 2022.8:235-256. Downloaded from www.annualreviews.org
Access provided by WIB6277 - TU Braunschweig on 05/31/23. For personal use only.
the target language). On the other hand, timing differences may be taken as a signal that indexes
qualitative differences in the parser, its knowledge bases, or its mechanisms (for discussion, see
Clahsen & Felser 2018, Roberts 2013). Now that the eld is moving from posing questions in
terms of whether L2 users can process Xlike native speakers to how they process X, it is high time
to start a discussion about how to distinguish signal from noise in data interpretation, especially
since, by denition, L2 performance is more variable than monolingual performance because of
larger differences in language experience across bilingual populations.
Beyond raising these methodological issues in data interpretation, variation in bilingual perfor-
mance also points to a systematic limitation of the current research paradigm that remains wedded
to the comparison of L2 processing to monolingual parsing. In using the yardstick of monolingual
performance as the sole reference point in the study of L2 processing, much research has turned
a blind eye to features of sentence processing that are unique to bilingualism. In the following
section, I outline why these features should move center stage.
3. NEW DIRECTIONS: TOWARD INTERLANGUAGE PROCESSING
Unlike adult monolinguals, whose processing is subserved by one stable linguistic system, the
knowledge and processing of L2 learners are characterized by the copresence of L1 and L2 prop-
erties, ongoing L2 development, and sometimes fossilization in that a learner’s L2 stops developing
short of the target language. In a seminal paper, Selinker (1972) captured these features of L2 ac-
quisition in the notion of interlanguage, which he described as a system comprising L1 and target-
language features, as well as learner-specic generalizations, with learners proceeding from one
interlanguage stage to the next in the course of L2 development. Formal studies of L2 acquisition
have attempted to provide a grammatical analysis of interlanguage grammars at different stages of
development (e.g., Lardiere 2009, Schwartz & Sprouse 1994). By contrast, L2 processing studies
usually abstract away from features of the interlanguage system and make a direct comparison to
monolingual performance. Such a perspective misses out on examining whether L2 processing
is grammatically constrained by the interlanguage system, even in cases when the latter differs
from monolingual grammars by virtue of representing more than one linguistic system or when
it exhibits nontarget features stemming from the L1. Finally, this perspective cannot account for
interlanguage development, that is, how learners move from one stage to the next in interlan-
guage development. In the following sections, I discuss these three core features of interlanguage
processing, namely, (a) the role of bilingual grammars in processing, (b) L1 effects and parser–
grammar relations, and (c) relations between (interlanguage) processing and development. I build
on the scarce empirical evidence we have so far on these topics in order to sketch new directions
for research on L2 sentence processing that cross-cut and transcend the traditional framing of re-
search on L2 sentence processing within difference-based and identity-based approaches relative
to L1 processing.
3.1. Bilingual Systems and L2 Sentence Processing
Interlanguage systems comprise features of all the languages a bilingual knows as well as idiosyn-
cratic properties. During sentence comprehension and production, bilinguals accordingly acti-
vate (sub)lexical representations and grammatical features across all of their languages (for review,
see Tokowicz 2015). For instance, bilinguals show different processing of language-ambiguous
word forms, such as cognates (e.g., lmGERMAN/ENGLISH) and homographs (giftGERMAN/ENGLISH),
compared to language-particular word forms (e.g., treeENGLISH–BaumGERMAN), even when reading
these words in sentence contexts (for review, see van Assche et al. 2012). Crucially, non-language-
selective access in the bilingual lexicon has nontrivial consequences in sentence comprehension.
Hopp
Annu. Rev. Linguist. 2022.8:235-256. Downloaded from www.annualreviews.org
Access provided by WIB6277 - TU Braunschweig on 05/31/23. For personal use only.
For one thing, the non-language-selective activation of lexical representations affects the time
course and staging of L2 sentence processing. Slower lexical access among L2 learners can de-
lay the application of syntactic structure building (Cheng et al. 2021, Hopp 2016), while lexical
coactivation of the L1 in the bilingual mental lexicon, by virtue of cognates, speeds up lexical
access and thus facilitates L2 structure building (Hopp 2017, Miller 2014) and L2 agreement
processing (Morales et al. 2016). For another, the integrated nature of lexical representations in
bilinguals engenders qualitatively different proles in L2 sentence processing. For instance, adult
high-intermediate L1 Russian learners of L2 German use gender agreement on German articles
predictively only for nouns whose Russian translation equivalents have the same grammatical gen-
der,while they do not use gender for incongruent nouns despite knowing and producing the target
L2 gender for them (Hopp & Lemmerth 2018). L2 learners thus compute syntactic agreement
according to both L1 and L2 grammatical features simultaneously (see also Meir et al. 2020). Simi-
larly,habitual Spanish–English code-switchers extend grammatical asymmetries in code-switched
noun phrases to gender agreement processing in monolingual Spanish sentences (Valdés Kroff
et al. 2017), leading to performance differences compared to monolingual speakers of Spanish.
Such ndings illustrate that bilinguals may be non-native-like in their processing performance,
while their processing is in perfect alignment with their interlanguage systems. In this way, effects
of bilingualism can lead to non-native sentence processing in an L2 because the parser operates
on representations in interlanguage systems that are by denition distinct from monolingual sys-
tems. Bilingualism effects of this type are not restricted to the lexicon and can be observed at all
levels of representation. For instance, L1 constraints on prosodic structure can interfere with the
processing of L2 inection, giving rise to divergent sentence processing (Goad & White 2019).
Instead of investigating whether L2 performance during processing can be native-like, then, a
more pertinent question that directly addresses the architecture of the language processing sys-
tem in bilinguals is to examine whether the L2 parser executes parses in accordance with the
non-native-like interlanguage system, as assessed independently, for instance, in ofine tasks.
Such an orientation also widens the scope of research on L2 sentence processing. Recognizing
that the linguistic knowledge of bilinguals is different from that of monolinguals, research on L2
sentence processing is slowly beginning to follow the “bilingual turn” in studies on L2 acquisition
(Ortega 2010) and increasingly compares sentence processing across different types of bilinguals.
Such comparisons allow researchers to hold the factor of bilingualism constant to explore differ-
ences within bilingual sentence processing in terms of variation in input quantity and quality,age,
and sequence of acquisition.1
For instance, Lemmerth & Hopp (2019) adapted the study on predictive gender processing by
adult L2 learners of German discussed above (Hopp & Lemmerth 2018) to simultaneous bilingual
Russian-German and early successive bilingual Russian-German children aged 8 years. While
simultaneous bilingual children patterned with monolingual age-matched German children in
using gender for prediction throughout, early successive bilinguals, who had only been exposed
to Russian up to age 2½ years, could use German gender predictively only when noun gender
was congruent with the Russian translation equivalent. Unlike the adult learners of German in
the study by Hopp & Lemmerth (2018)—who also had more difculty with gender on articles,
which do not exist in Russian—the child L2 learners were not affected in their use of gender by
differences between the L1 and L2 syntax. Such asymmetric performance of successive child L2
1Of course, bilingual comparisons do not obviate the need for an L1 control group in a study to show that the
experimental manipulation works. However, L1 speakers should not invariably constitute the reference group
for bilingual performance.
www.annualreviews.org •Second Language Sentence Processing
Annu. Rev. Linguist. 2022.8:235-256. Downloaded from www.annualreviews.org
Access provided by WIB6277 - TU Braunschweig on 05/31/23. For personal use only.
and adult L2 learners suggests that effects of bilingualism in the lexicon and in the grammar bear
different signatures depending on the age of onset.
Next to studies comparing child and adult L2 sentence processing, processing research is gath-
ering pace on adult L1 attriters (i.e., long-term emigrants with little continued L1 input whose
L2 has become their dominant language in adulthood; Schmid 2011) and L1 heritage speakers
(Montrul 2016). These studies investigate the directionality of bilingualism effects by contrasting
forward (L1 to L2) and reverse (L2 to L1) cross-linguistic inuence. Emerging research on sen-
tence processing in L1 attriters points to some L2 inuence in L1 parsing preferences (Dussias &
Sagarra 2007), but L1 attriters continue to show monolingual-like performance with the L1 gram-
mar in sentence processing (Bergmann et al. 2015, Grüter & Hopp 2021; though see Kasparian &
Steinhauer 2017). These initial ndings from L1 attrition tie in with research on sentence process-
ing in L1 heritage speakers who experienced a dominance shift from the L1 to L2 in childhood,
yet do not resemble late L2 learners in parsing, and largely pattern with monolingual children and
adults (Polinsky 2018; though see Jegerski & Sekerina 2020). Such emerging research suggests a
crucial role for age of acquisition within bilingual sentence processing. To assess whether differ-
ent processing proles between bilingual groups indicate that late L2 learners employ processing
strategies that deviate from their interlanguage representations, we need careful investigations
of both the linguistic knowledge and the language processing across different types of bilingual
populations.
In sum, going beyond the monolingual reference point in the study of L2 sentence processing
allows researchers to investigate the degree to which L2 sentence processing conforms to bilin-
guals’ linguistic knowledge, which is necessarily different from monolingual linguistic knowledge.
This shift in perspective implicates the comparative study of sentence processing in different bilin-
gual groups to assess how variation in age of onset, sequence of acquisition, and input may affect
bilingual sentence processing. Not least, such a bilingual perspective on L2 sentence processing
can reframe the view of L2 sentence processing away from being a decient counterpart to mono-
lingual sentence processing. Instead, interlanguage processing affords an opportunity to investi-
gate and comprehend how language processing operates relative to the knowledge of bilingual
speakers.
3.2. The L1 Transfer Paradox and Parser–Grammar Relations
By denition, the L1 is part of the linguistic knowledge of L2 learners, and transfer or cross-
linguistic inuence from the L1 surfaces in L2 sentence production and comprehension. Even L2
learners at more advanced stages of L2 acquisition show optionality between the use of L1-based
and L2-based grammatical options in production and often accept both L1 and L2 options in
comprehension (for review, see Foley & Flynn 2013).
As reviewed in previous sections, L2 learners also show some L1 differences in integration
and prediction in the processing of lexical and inectional information, for which L1–L2 over-
laps facilitate L2 processing (e.g., Dussias & Cramer Scaltz 2008, Frenck-Mestre & Pynte 1997,
Frenck-Mestre et al. 2019, van Bergen & Flecken 2017). For syntactic processing, though, the
evidence that L2 learners refer to the L1 grammar or L1 parsing strategies is rather limited. For
instance, L2 learners garden-path during L2 processing irrespective of L1 word order differences
( Juffs 2004), and their L1 parsing preferences do not consistently affect structure-based relative
clause attachment preferences in the L2 (e.g., Felser et al. 2003). To investigate the status of the
L1 grammar in L2 processing, several studies examined “cross-language syntactic conicts” (Kaan
et al. 2015), in which the surface word order in the L2 can be mapped to a different interpretation
using the L1 grammar. Building on the work of Jacob (2009), Hopp (2017) studied whether L1
Hopp
Annu. Rev. Linguist. 2022.8:235-256. Downloaded from www.annualreviews.org
Access provided by WIB6277 - TU Braunschweig on 05/31/23. For personal use only.
German learners coactivate the L1 grammar and assign an L1-based parse to L2 input. In exam-
ple 4, the verbatim translation of the preposed adjunct clause (underlined in the example) maps
to a canonical embedded clause in German because of its SOV word order, while it represents an
infrequent reduced relative clause in English:
(4) When the doctor Sarah ignored tried to leave the room the nurse came in all of a sudden.
Compared to control sentences, reading time differences for example 4 indicative of L1 activation
were observed only in experiments in which English target sentences were interspersed with Ger-
man ller items. By contrast, when reading in English-only contexts, no reading time differences
indicative of L1 effects surfaced. Yet, even in language-mixing contexts, L1 effects can be elusive.
For instance, in a visual world eye-tracking study using structural priming, object question primes
in their L1 German did not induce German learners of English to process English subject wh-
questions using their L1 grammar,even though the L1 grammar allows for an object wh-question
parse of this string (Hopp & Grüter 2021).
Under the widely held assumption in L1 and L2 processing research that the parser directly re-
cruits grammatical information (for review,see Juffs & Rodriguez 2014), both the L1 grammar and
the L2 interlanguage grammar should affect parsing in L2 users; yet, the L1 grammar seems to be
conspicuously absent in L2 processing. At the same time, L1 effects often surface in sentence-nal
interpretations. For instance, Kaan et al. (2015) examined whether L1 Dutch advanced English
learners would parse ungrammatical relative clauses such as Mark may know the instructor who the
students ∗has avoided as subject relative clauses as per Dutch OV syntax. In self-paced reading, the
L2 group did not show any reading time differences between locally mismatched singular or plural
nouns inside relative clauses that would reect use of Dutch word order online, yet they did give
more subject interpretations of the mismatched condition that was compatible with Dutch syntax
in end-of-sentence decisions. In other words, even though advanced L2 learners do not appear
to activate the L1 grammar during real-time processing, they make recourse to L1 grammars in
subsequent judgment or decision tasks (see also Hopp & Grüter 2021, Rankin et al. 2019, Roberts
et al. 2008).
These L1 effects indicate that L2 learners partially compute grammatical L1 parses alongside
L2 parses. However, L1 parses appear to have a low resting level of activation compared to the L2
grammar, especially among advanced L2 learners. Once this activation is upped by language mix-
ing (Hopp 2017, Jacob 2009), L1 effects become visible during processing. Similarly, L1 activation
can be enhanced by sentence-nal judgment or decision tasks (Kaan et al. 2015). Here, partici-
pants are presented with two possible interpretations: One corresponds to the L2, and the other is
licensed by the L1 grammar.When participants need to assess and compare both interpretations,
L2 learners can apply an L1-based parse to the nontarget interpretation and thus sometimes re-
cover an additional or different interpretation of the sentence when the parseable L1 option wins
out over the L2 option (Hopp & Grüter 2021).
Under this view,L2 parsing relates to grammatical approaches to L2 acquisition which argue
that interlanguage grammars simultaneously encode L1 and L2 representations, with previous
L1-based or developmental grammatical representations lingering and competing for selection
throughout the course of L2 development (Amaral & Roeper 2014, Westergaard 2021). Accord-
ing to these approaches, L1 grammatical options constitute inherent and persistent features of
interlanguage grammars and can be activated alongside L2 options. Interlanguage processing can
probe the conditions and limits of how the parser recruits the L1 grammar vis à vis the L2 gram-
mar to uncover the dynamics of grammatical cross-linguistic inuence. Specically, research can
weight the activation of the L1 grammatical options and study when and how L2 learners recruit
www.annualreviews.org •Second Language Sentence Processing
Annu. Rev. Linguist. 2022.8:235-256. Downloaded from www.annualreviews.org
Access provided by WIB6277 - TU Braunschweig on 05/31/23. For personal use only.
L1 grammatical properties in L2 parsing at different developmental stages, at different processing
steps, and under different task demands.
By investigating the degree to which sentence processing is non-language-selective, an inter-
language perspective on L2 sentence processing also enters into dialog with current approaches
to the bilingual lexicon (e.g., Dijkstra et al. 2019) and bilingual speech processing (Grosjean &
Byers-Heinlein 2018). As it strives to compare bilingual lexical and sentence processing (rather
than monolingual and L2 processing), it holds the promise of integrating sentence processing
within a unied architecture of the bilingual language processing system.
3.3. Processing to Learn and the Pitfalls of Less Prediction
To acquire a language, a learner needs to be able to process it, so that novel information can be
acquired (e.g., Fodor 1999, Omaki & Lidz 2015). The relevance of processing as a prerequisite
for learning has recently gained traction in the context of predictive processing (Section 2.3).
A parser that makes predictions about the unfolding input can learn whenever it experiences a
prediction error, that is, when the parser’s predictions are disconrmed by the input. By way of
predictive processing, a learner can thus actively test grammatical hypotheses about the target
language against the input since a prediction error creates feedback for the parser (e.g., Phillips &
Ehrenhofer 2015). After having encountered a prediction error, the parser can adjust processing
or its linguistic knowledge base to match the input and decrease future prediction error (Dell &
Chang 2014, Kleinschmidt & Jaeger 2015).
In adult sentence processing, structural priming and syntactic adaptation have been used as
methods to study learning via prediction error (for review, see Kaan & Chun 2018). When ex-
posed to a high number of sentences containing complex structures, such as reduced relative
clauses, monolingual adults cumulatively adapt their sentence processing in that the processing
costs of the complex structures are reduced (Fine et al. 2013, Yan & Jaeger 2020). These effects
of syntactic adaptation suggest that users continuously update their parsing preferences as a re-
sult of incoming input. The evidence on whether L2 learners also adapt their sentence processing
is mixed. On the one hand, studies on structural priming among L2 learners show that adult L2
learners demonstrate priming effects in similar magnitude to monolinguals (for review,see Jackson
2018). In structural priming, speakers reuse a structure they have just heard, or readers have less
processing difculty with a complex syntactic structure if the preceding sentence was of a similar
type. These trial-by-trial effects indicate that L2 learners adjust their processing in similar ways
as monolinguals do. In prediction-based accounts (e.g., Dell & Chang 2014),priming constitutes
implicit learning in that the language user decreases the misalignment between predicted and ob-
served structures in the input. On the other hand, sentence processing studies on L2 syntactic
ambiguities report that adult L2 learners continue to show processing difculty with dispreferred
or complex structures even after prolonged exposure to these structures in the input (Kaan et al.
2019), indicating that they do not use adaptive mechanisms or that they may need more input than
do native speakers to adapt.
Beyond testing whether L2 learners adapt their syntactic processing preferences in ambiguous
sentences, some studies explore whether massed input inducing prediction error can counteract
the underuse of grammatical information in L2 sentence processing and lead to the integration
of morphosyntax and successful reanalysis. In a reading study with an experimental group and
a control group, Hopp (2020) investigated whether massed input designed to systematically in-
duce prediction error would lead to the use of case marking for reanalysis among intermediate to
advanced L1 German learners of English. In potential garden-path sentences like When the spec-
tators were applauding they remained in the spotlight, L2 learners routinely construe the postverbal
Hopp
Annu. Rev. Linguist. 2022.8:235-256. Downloaded from www.annualreviews.org
Access provided by WIB6277 - TU Braunschweig on 05/31/23. For personal use only.
pronoun as the object of the verb (see also Hopp 2015a). Such garden paths indicate that readers
have a strong prediction for a (potentially) transitive verb to be followed by an object. Following
a pretest using eye-tracking during reading, the experimental group in the study by Hopp (2020)
read sentences for comprehension that contained consistent intransitive uses of optionally transi-
tive verbs (e.g., The boy played and he pleased the parents with the music). In these sentences, the verb
was followed by a coordinating conjunction (and) and an unambiguously nominative case-marked
pronoun (he), which together highlighted the beginning of a new clause. These sentences were
intended to create frequent prediction errors, since the verb was not followed by an object and
the pronoun was clearly a subject. The control group read matched input with transitive readings
of the verbs (The boy played the music and it pleased the parents) that would not lead to any prediction
error.In posttests, the reading patterns for garden-path sentences among the control group were
unchanged from the pretest, while the experimental group did not demonstrate garden paths any-
more, indicating that the readers had adapted to analyzing the postverbal pronoun as a subject,
arguably as a result of experiencing frequent prediction errors in the massed input. These initial
ndings suggest that adaptation and implicit learning via prediction error can lead L2 learners to
use case marking as a reanalysis cue during L2 sentence processing. However, more research is
needed to delineate the scope of adaptation among L2 learners and to identify the nature of the
mechanisms involved in syntactic adaptation. Furthermore, open questions remain as to whether
syntactic adaptation during online processing translates into longer-term learning as measured in
different or delayed tasks (see Jackson & Hopp 2020).
Critically, typical features of L2 sentence processing could affect the success of learning (for
discussion, see Hopp 2021). First, lower degrees of predictive processing in the L2 may create
obstacles to learning in that learners cannot generate sufcient feedback by virtue of prediction
error during sentence processing (Phillips & Ehrenhofer 2015). Second, L2 learners often base
their predictions on nongrammatical information (Section 2.3). However, learning from predic-
tion error can only be successful if learners can track the nature of their prediction error and then
update their predictions accordingly.If learners predominantly generate predictions on the basis
of semantic information, prediction errors will not be informative for learning the L2 grammar.
In these regards, characteristic features of L2 sentence processing may impose limits on learning
the L2.
4. CONCLUSION
Two decades of systematic research on L2 sentence processing have led to a better understanding
of how L2 learners differ in their real-time sentence comprehension from L1 speakers. There
is now widespread agreement that, rather than being monocausal, L2–L1 differences reect the
interplay between differences in the integration and prediction of different cues, their utility,
and cognitive constraints on processing in real time. As much as the eld is making strides
toward more nuanced views of L2 sentence processing, it upholds native-speaker benchmarks
of performance, which, in turn, perpetuate L2 sentence processing as a deviant or decient form
of monolingual parsing and limit the scope of inquiry. Interlanguage processing breaks with the
monolingual standard and widens the lens to capture characteristic features of L2 acquisition,
such as bilingual knowledge systems, L1 transfer, and L2 learning and its limits. Such a shift
in perspective comes with opportunities for novel insights, yet also with novel challenges. In
particular for lower-prociency L2 learners, who do not yet have a stable command of the L2,
careful experimentation is necessary to link the processing of a particular linguistic property to
its ongoing acquisition. In these respects, it will be fruitful to align research on natural L2s with
studies on articial or miniature languages (see sidebar titled Sentence Processing in Articial
www.annualreviews.org •Second Language Sentence Processing
Annu. Rev. Linguist. 2022.8:235-256. Downloaded from www.annualreviews.org
Access provided by WIB6277 - TU Braunschweig on 05/31/23. For personal use only.
SENTENCE PROCESSING IN ARTIFICIAL AND MINIATURE LANGUAGES
As a corollary to research on the L2 processing of natural languages, studies on articial language (AL) explore
sentence processing in laboratory-constructed nonce languages or miniature versions of natural languages. As “test
tube” models of natural language processing, they have full control over the type and amount of input, including
similarities between the AL and learners’ previous languages (for review, see Grey 2020, Morgan-Short 2020). Typ-
ically using ERPs, they examine the limits of AL/L2 processing and patterns of development. Learners who have
achieved high prociency in the AL after massed exposure in a short time demonstrate a range of early ERP signa-
tures in response to lexical-semantic and morphosyntactic violations comparable to those observed in monolingual
natural language processing. In the course of learning, AL learners often proceed through a series of processing
stages, moving from prioritizing lexical semantics or L1-based strategies to integrating morphosyntax (see also
Osterhout et al. 2006 for natural L2s). Finally, AL studies manipulate the type of input between more naturalistic,
implicit training and exposure to explicit rules. Both explicit and implicit learning lead to successful AL processing.
Yet, only learners who become aware of rules during implicit training engage the full set of processing mechanisms
in an AL, including prediction (Andringa 2020).
Despite continuing concerns about their ecological validity, AL studies cohere with ndings from natural L2
learning that L2 sentence processing is fully constrained by the architecture and processes that subserve monolin-
gual processing (Section 2). By way of longitudinal study of (fast-tracked) development, they offer complementary
insights into the dynamics of cross-linguistic inuence (Section 3.2) and the implicit and explicit processing mech-
anisms used in learning (Section 3.3).
and Miniature Languages). Moreover, the growing number of multilinguals, who speak more
than two languages, adds complexity to the study of non-native processing (González Alonso
& Rothman 2022). These challenges equally highlight the potentials and contributions that
the investigation of bilinguals and multilinguals can make to our understanding of sentence
processing and, conversely, how the study of L2 sentence processing can establish itself as a
central area in research on L2 acquisition and bilingualism.
DISCLOSURE STATEMENT
The author is not aware of any afliations, memberships, funding, or nancial holdings that might
be perceived as affecting the objectivity of this review.
ACKNOWLEDGMENTS
I would like to thank Theres Grüter,Carrie N. Jackson, Duygu ¸Safak, and the journal’s Co-Editor
Colin Phillips for critical comments that helped improve the quality of the manuscript.
LITERATURE CITED
Aldwayan S, Fiorentino R, Gabriele A. 2010. Evidence of syntactic constraints in the processing of wh-
movement. In Language Acquisition and Language Disorders, Vol. 53: Research in Second Language Processing
and Parsing, ed. B VanPatten, Jill Jegerski, pp. 65–86. Amsterdam: John Benjamins
Altmann GT, Kamide Y. 1999. Incremental interpretation at verbs: restricting the domain of subsequent ref-
erence. Cognition 73:247–64
Amaral L, Roeper T. 2014. Multiple grammars and second language representation. Second Lang. Res. 30:3–36
Andringa S. 2020. The emergence of awareness in uninstructed L2 learning: a visual world eye tracking study.
Second Lang. Res. 36:335–57
Hopp
Annu. Rev. Linguist. 2022.8:235-256. Downloaded from www.annualreviews.org
Access provided by WIB6277 - TU Braunschweig on 05/31/23. For personal use only.
Bergmann C, Meulman N, Stowe LA, Sprenger SA, Schmid MS. 2015. Prolonged L2 immersion engenders
little change in morphosyntactic processing of bilingual natives. Clin. Neurosci. 26:1065–70
Bley-Vroman R.1988. The fundamental character of foreign language learning. In Grammar and Second Lan-
guage Teaching, ed. W Rutherford, M Sharwood Smith, pp. 19–30. Rowley, MA: Newbury House
Boxell O, Felser C. 2017.Sensitivity to parasitic gaps inside subject islands in native and non-native sentence
processing. Biling.: Lang. Cogn. 20:494–511
Chambers CG, Cooke H. 2009. Lexical competition during second-language listening: Sentence context,
but not prociency, constrains interference from the native lexicon. J. Exp. Psychol. Learn. Mem. Cogn.
35:1029–40
Chaves RP, Putnam MT. 2021. Unbounded Dependency Constructions: Theoretical and Experimental Perspectives.
Oxford, UK: Oxford Univ. Press
Cheng Y, Rothman J, Cunnings I. 2021. Parsing preferences and individual differences in nonnative sentence
processing: evidence from eye movements. Appl. Psycholinguist. 42:129–51
Chomsky N. 1981. Lectures on Government and Binding. Dordrecht, Neth.: Foris
Clahsen H, Felser C. 2006. Grammatical processing in language learners. Appl. Psycholinguist. 27:3–46
Clahsen H, Felser C. 2018.Some notes on the Shallow Structure Hypothesis. Stud. Second Lang. Acquis. 40:693–
706
Clifton C, Frazier L. 1989. Comprehending sentences with long-distance dependencies. In Linguistic Structure
in Language Processing, ed. GN Carlson, MK Tanenhaus, pp. 273–317. Dordrecht, Neth.: Springer
Cunnings I. 2017. Parsing and working memory in bilingual sentence processing. Biling.: Lang. Cogn. 20:659–
78
Dallas A, Kaan E. 2008. Second language processing of ller-gap dependencies in late learners. Lang. Linguist.
Compass 2:372–88
Dekydtspotter L, Donaldson B, Edmonds AC, Fultz AL, Petrush RA. 2008. Syntactic and prosodic computa-
tions in the resolution of relative clause attachment ambiguity by English-French learners. Stud. Second
Lang. Acquis. 30:453–80
Dell GS, Chang F. 2014. The P-chain: relating sentence production and its disorders to comprehension and
acquisition. Philos. Trans. R. Soc. B 36:20120394
Dijkgraaf A, Hartsuiker RJ, Duyck W. 2017. Predicting upcoming information in native-language and non-
native-language auditory word recognition. Biling.: Lang. Cogn. 20:917–30
Dijkstra T, Wahl A, Buytenhuijs F, van Halem N, Al-Jibouri Z, et al. 2019. Multilink: a computational model
for bilingual word recognition and word translation. Biling.: Lang. Cogn. 22:657–79
Dillon B. 2014. Syntactic memory in the comprehension of reexive dependencies: an overview.Lang. Linguist.
Compass 8:171–87
Dussias P, Cramer Scaltz T. 2008. Spanish-English L2 speakers’ use of subcategorization bias information in
the resolution of temporary ambiguity during second language reading. Acta Psychol. 128:501–13
Dussias PE, Kroff JRV, Tamargo REG, Gerfen C. 2013. When gender and looking go hand in hand: gram-
matical gender processing in L2 Spanish. Stud. Second Lang. Acquis. 35:353–87
Dussias PE, Piñar P. 2010. Effects of reading span and plausibility in the reanalysis of wh-gaps by Chinese-
English second language speakers. Second Lang. Res. 26:443–72
Dussias PE, Sagarra N.2007. The effect of exposure on syntactic parsing in Spanish-English bilinguals. Biling.:
Lang. Cogn. 10:101–16
Federmeier KD. 2007. Thinking ahead: the role and roots of prediction in language comprehension.
Psychophysiology 44:491–505
Felser C. 2015. Native versus non-native processing of discontinuous dependencies. Second Lang. 14:5–19
Felser C. 2019. Structure-sensitive constraints in non-native sentence processing. J. Eur. Second Lang. Assoc.
3:12–22
Felser C, Cunnings I. 2012. Processing reexives in English as a second language: the role of structural and
discourse-level constraints. Appl. Psycholinguist. 33:571–603
Felser C, Cunnings I, Batterham C, Clahsen H. 2012. The timing of island effects in nonnative sentence
processing. Stud. Second Lang. Acquis. 34:67–98
Felser C, Roberts L. 2007. Processing wh-dependencies in a second language: a cross-modal priming study.
Second Lang. Res. 23:9–36
www.annualreviews.org •Second Language Sentence Processing
Annu. Rev. Linguist. 2022.8:235-256. Downloaded from www.annualreviews.org
Access provided by WIB6277 - TU Braunschweig on 05/31/23. For personal use only.
Felser C, Roberts L, Marinis T, Gross R. 2003. The processing of ambiguous sentences by rst and second
language learners of English. Appl. Psycholinguist. 24:453–89
Fernandez L, Höhle B, Brock J, Nickels L. 2018. Investigating auditory processing of syntactic gaps with L2
speakers using pupillometry. Second Lang. Res. 34:201–27
Ferreira F, Patson ND.2007. The ‘good enough’ approach to language comprehension. Lang. Linguist. Compass
1:71–83
Fine AB, Jaeger TF, Farmer TA, Qian T. 2013. Rapid expectation adaptation during syntactic comprehension.
PLOS ONE 8(10):e77661
Fodor JD. 1999. Learnability theory: triggers for parsing with. In The Development of Second Language Gram-
mars: A Generative Approach, ed. EC Klein, G Martohardjono, pp. 363–406. Amsterdam: John Benjamins
Foley C, Flynn S. 2013. The role of the native language. In The Cambridge Handbook of Second Language Acqui-
sition, ed. M Young-Scholten, J Herschensohn, pp. 97–113. Cambridge, UK: Cambridge Univ. Press
Frenck-Mestre C, Kim SK, Choo H, Ghio A, Herschensohn J, Koh S. 2019. Look and listen! The online
processing of Korean case by native and non-native speakers. Lang. Cogn. Neurosci. 34:385–404
Frenck-Mestre C, Pynte J.1997. Syntactic ambiguity resolution while reading in second and native languages.
Q. J. Exp. Psychol. A 50:119–48
Fujita H, Cunnings I. 2020. Reanalysis and lingering misinterpretation of linguistic dependencies in native
and non-native sentence comprehension. J. Mem. Lang. 115:104154
Gibson E. 1998. Linguistic complexity: locality of syntactic dependencies. Cognition 68:1–76
Goad H, White L. 2019.Prosodic effects on L2 grammars. Linguist. Approaches Biling. 9:769–808
González Alonso J, Rothman J. 2022. The psycholinguistics of L3/Ln acquisition. In The Routledge Handbook
of Second Language Acquisition and Psycholinguistics, ed. A Godfroid, H Hopp. New York: Routledge. In
press
Grey S. 2020. What can articial languages reveal about morphosyntactic processing in bilinguals? Biling.:
Lang. Cogn. 23:81–86
Grosjean F, Byers-Heinlein K. 2018. The Listening Bilingual: Speech Perception, Comprehension, and Bilingualism.
Hoboken: Wiley
Grüter T, Hopp H. 2021. How permeable are native and non-native syntactic processing to crosslinguistic
inuence? J. Mem. Lang. 121:104281
Grüter T, Lau E, Ling W. 2020. How classiers facilitate predictive processing in L1 and L2 Chinese: the
role of semantic and grammatical cues. Lang. Cogn. Neurosci. 35:221–34
Grüter T, Lew-Williams C,Fernald A. 2012. Grammatical gender in L2: a production or a real-time process-
ing problem? Second Lang. Res. 28:191–215
Grüter T, Rohde H. 2021. Limits on expectation-based processing: use of grammatical aspect for co-reference
in L2. Appl. Psycholinguist. 42:51–75
Grüter T, Rohde H, Schafer AJ. 2017. Coreference and discourse coherence in L2: the roles of grammatical
aspect and referential form. Linguist. Approaches Biling. 7:199–229
Haarmann HJ, Just MA, Carpenter PA. 1997. Aphasic sentence comprehension as a resource decit: a com-
putational approach. Brain Lang. 59:76–120
Havik E, Roberts L, van Hout R, Schreuder R, Haverkort M. 2009. Processing subject-object ambiguities in
the L2: a self-paced reading study with German L2 learners of Dutch. Lang. Learn. 59:73–112
Henry N, Hopp H, Jackson CN. 2017. Cue additivity and adaptivity in predictive processing. Lang. Cogn.
Neurosci. 32:1229–49
Hopp H. 2006. Syntactic features and reanalysis in near-native processing. Second Lang. Res. 22:369–97
Hopp H. 2010. Ultimate attainment in L2 inectional morphology: performance similarities between non-
native and native speakers. Lingua 120:901–31
Hopp H. 2013. Grammatical gender in adult L2 acquisition: relations between lexical and syntactic variability.
Second Lang. Res. 29:33–56
Hopp H. 2014. Working memory effects in the L2 processing of ambiguous relative clauses. Lang. Acquis.
21:250–78
Hopp H. 2015a. Individual differences in the second language processing of object-subject ambiguities. Appl.
Psycholinguist. 36:129–73
Hopp
Annu. Rev. Linguist. 2022.8:235-256. Downloaded from www.annualreviews.org
Access provided by WIB6277 - TU Braunschweig on 05/31/23. For personal use only.
Hopp H. 2015b. Semantics and morphosyntax in L2 predictive sentence processing. Int. Rev. Appl. Linguist.
53:277–306
Hopp H. 2016. Learning (not) to predict: grammatical gender processing in adult L2 acquisition.Second Lang.
Res. 32:277–307
Hopp H. 2017. Cross-linguistic lexical and syntactic co-activation in L2 sentence processing. Linguist.
Approaches Biling. 7:96–130
Hopp H. 2018. The bilingual mental lexicon in L2 sentence processing.Second Lang. Res. 17:5–27
Hopp H. 2020. Morphosyntactic adaptation in adult L2 processing: exposure and the processing of case and
tense violations. Appl. Psycholinguist. 41:627–56
Hopp H. 2021. Prediction and grammatical learning in second-language sentence processing. In Prediction in
L2 Processing and Learning, ed. E Kaan, T Grüter, pp. 277–307. Amsterdam: John Benjamins
Hopp H, Grüter T. 2021. The time-course of competition from the L1 grammar in L2 sentence pro-
cessing: evidence from cross-linguistic structural priming. Second Lang. Res. https://doi.org/10.1177/
02676583211009586
Hopp H, Lemmerth N. 2018. Lexical and syntactic congruency in L2 predictive gender processing. Stud.
Second Lang. Acquis. 40:171–99
Hoshino N, Dussias PE, Kroll JF. 2010. Processing subject-verb agreement in a second language depends on
prociency. Biling.: Lang. Cogn. 13:87–98
Huettig F, Brouwer S. 2015. Delayed anticipatory spoken language processing in adults with dyslexia—
evidence from eye-tracking. Dyslexia 21:97–122
Huettig F, Rommers J, Meyer AS. 2011. Using the visual world paradigm to study language processing: a
review and critical evaluation. Acta Psychol. 137:151–71
Ito A, Corley M, Pickering MJ. 2018. A cognitive load delays predictive eye movements similarly during L1
and L2 comprehension. Biling.: Lang. Cogn. 21:251–64
Jackson C. 2008. Prociency level and the interaction of lexical and morphosyntactic information during L2
sentence processing. Lang. Learn. 58:875–909
Jackson CN. 2018. Second language structural priming: a critical review and directions for future research.
Second Lang. Res. 34:539–52
Jackson CN, Hopp H. 2020. Prediction error and implicit learning in L1 and L2 syntactic priming. Int. J.
Biling. 24:895–911
Jackson CN, Roberts L. 2010. Animacy affects the processing of subject-object ambiguities in the second
language: evidence from self-paced reading with German second language learners of Dutch. Appl.
Psycholinguist. 31:671–91
Jacob G. 2009. The role of the native language in second-language syntactic processing. PhD Thesis, Dundee School
of Psychology, Dundee, UK
Jacob G, Felser C. 2016.Reanalysis and semantic persistence in native and non-native garden-path recovery.
Q. J. Exp. Psychol. 69:907–25
Jegerski J, Sekerina IA. 2020. The processing of input with differential object marking by heritage Spanish
speakers. Biling.: Lang. Cogn. 23:274–82
Jessen A, Felser C. 2019. Reanalysing object gaps during non-native sentence processing: evidence from E RPs.
Second Lang. Res. 35:285–300
Juffs A. 2004. Representation, processing and working memory in a second language. Trans. Philol. Soc.
102:199–225
Juffs A, Harrington M. 1996. Garden path sentences and error data in second language processing research.
Lang. Learn. 46:283–323
Juffs A, Rodriguez G. 2014. Second Language Sentence Processing. New York: Routledge
Just MA, Carpenter PA. 1992. A capacity theory of comprehension: individual differences in working memory.
Psychol. Rev. 99:122–49
Kaan E. 2014. Predictive sentence processing in L2 and L1: What is different? Linguist. Approaches Biling.
4:257–82
Kaan E, Ballantyne JC, Wijnen F. 2015. Effects of reading speed on second-language sentence processing.
Appl. Psycholinguist. 36:799–830
www.annualreviews.org •Second Language Sentence Processing
Annu. Rev. Linguist. 2022.8:235-256. Downloaded from www.annualreviews.org
Access provided by WIB6277 - TU Braunschweig on 05/31/23. For personal use only.
Kaan E, Chun E. 2018. Syntactic adaptation. In Psychology of Learning and Motivation, Vol. 68, ed. D Watson,
K Federmeier, pp. 85–116. New York: Academic
Kaan E, Futch C, Fuertes RF, Mujcinovic S, de la Fuente EÁ. 2019. Adaptation to syntactic structures in native
and nonnative sentence comprehension. Appl. Psycholinguist. 40:3–27
Kaan E, Kirkham J, Wijnen F. 2016. Prediction and integration in native and second-language processing of
elliptical structures. Biling.: Lang. Cogn. 19:1–18
Kasparian K, Steinhauer K. 2017. When the second language takes the lead: neurocognitive processing
changes in the rst language of adult attriters. Front. Psychol. 8:389
Kim E, Baek S, Tremblay A. 2015a. The role of island constraints in second language sentence processing.
Lang. Acquis. 22:384–416
Kim E, Montrul S, Yoon J. 2015b. The on-line processing of binding principles in second language acquisition:
evidence from eye tracking. Appl. Psycholinguist. 36:1317–74
Kleinschmidt DF, Jaeger TF. 2015. Robust speech perception: recognize the familiar, generalize to the similar,
and adapt to the novel. Psychol. Rev. 122:148–203
Kuperberg GR, Jaeger TF. 2016. What do we mean by prediction in language comprehension? Lang. Cogn.
Neurosci. 31:32–59
Lardiere D. 2009.Some thoughts on the contrastive analysis of features in second language acquisition. Second
Lang. Res. 25:173–227
Lee EK, Lu DHY, Garnsey SM. 2013. L1 word order and sensitivity to verb bias in L2 processing. Biling.:
Lang. Cogn. 16:761–75
Lemmerth N, Hopp H. 2019. Gender processing in simultaneous and successive bilingual children: effects of
lexical and syntactic cross-linguistic inuence. Lang. Acquis. 26:21–45
Lew-Williams C, Fernald A. 2010. Real-time processing of gender-marked articles by native and non-native
Spanish speakers. J. Mem. Lang. 63:447–64
Lewis RL, Vasishth S, Van D yke JA. 2006. Computational principles of working memory in sentence compre-
hension. Trends Cogn. Sci. 10:447–54
Linck JA, Osthus P, Koeth JT, Bunting MF. 2014. Working memory and second language comprehension and
production: a meta-analysis. Psychon. Bull. Rev. 21:861–83
López Prego B, Gabriele A. 2014. Examining the impact of task demands on morphological variability in
native and non-native Spanish. Linguist. Approaches Biling. 4:192–221
MacDonald MC, Hsiao Y. 2018. Sentence comprehension. In The Oxford Handbook of Psycholinguistics,Vol.2,
ed. SA Rüschemeyer, MG Gaskell, pp. 171–96. Oxford, UK: Oxford Univ. Press
Marinis T, Roberts L, Felser C, Clahsen H. 2005. Gaps in second language sentence processing. Stud. Second
Lang. Acquis. 27:53–78
McDonald J. 2006.Beyond the critical period: processing-based explanations for poor grammaticality judge-
ment performance by late second language learners. J. Mem. Lang. 55:381–401
McElree B. 2000. Sentence comprehension is mediated by content-addressable memory structures.J. Psychol.
Res. 29:111–23
McRae K, Matsuki K. 2013. Constraint-based models of sentence processing. See van Gompel 2013, pp. 51–77
Meir N, Parshina O, Sekerina IA. 2020. The interaction of morphological cues in bilingual sentence process-
ing: an eye-tracking study. In Proceedings of the 44th Boston University Conference on Language Development,
ed. MM Brown, A Kohut, pp. 376–89. Somerville, MA: Cascadilla
Miller AK. 2014. Accessing and maintaining referents in L2 processing of wh-dependencies. Linguist.
Approaches Biling. 4:167–91
Mishra RK, Singh N, Pandey A, Huettig F. 2012. Spoken language-mediated anticipatory eye movements
are modulated by reading ability: evidence from Indian low and high literates. J. Eye Mov. Res. 5(1).
https://doi.org/10.16910/jemr.5.1.3
Mitsugi S, MacWhinney B. 2016. The use of case marking for predictive processing in second language
Japanese. Biling.: Lang. Cogn. 19:19–35
Montrul S. 2016. The Acquisition of Heritage Languages. Cambridge, UK: Cambridge Univ. Press
Morales L, Paolieri D, Dussias PE,Kroff JRV, Gerfen C, Bajo MT. 2016. The gender congruency effect during
bilingual spoken-word recognition. Biling.: Lang. Cogn. 19:294–310
Hopp
Annu. Rev. Linguist. 2022.8:235-256. Downloaded from www.annualreviews.org
Access provided by WIB6277 - TU Braunschweig on 05/31/23. For personal use only.
Morgan-Short K. 2020. Insights into the neural mechanisms of becoming bilingual: a brief synthesis of second
language research with articial linguistic systems. Biling. Lang.: Cogn. 23:87–91
Nickels S, Opitz B, Steinhauer K. 2013. ERPs show that classroom-instructed late second language learners
rely on the same prosodic cues in syntactic parsing as native speakers. Neurosci. Lett. 557:107–11
Omaki A, Lidz J. 2015. Linking parser development to acquisition of syntactic knowledge. Lang. Acquis.
22:158–92
Omaki A, Schulz B. 2011. Filler-gap dependencies and island constraints in second-language sentence pro-
cessing. Stud. Second Lang. Acquis. 33:563–88
Ortega L. 2010. The bilingual turn in SLA. Plenary presented at the Annual Conference of the American
Association for Applied Linguistics, Atlanta, GA, Mar. 6–9
Osterhout L, McLaughlin J, Pitkänen I, Frenck-Mestre C, Molinaro N. 2006. Novice learners, longitudi-
nal designs, and event-related potentials: a means for exploring the neurocognition of second language
processing. Lang. Learn. 56:199–230
Pan H, Felser C. 2011.Referential context effects in L2 ambiguity resolution: evidence from self-paced read-
ing. Lingua 121:221–36
Patterson C, Trompelt H, Felser C. 2014. The online application of binding condition B in native and non-
native pronoun resolution. Front . Psychol. 5:147
Peters R, Grüter T, Borovsky A. 2018. Vocabulary size and native speaker self-identication inuence exi-
bility in linguistic prediction among adult bilinguals. Appl. Psycholinguist. 39:1439–69
Phillips C, Ehrenhofer L. 2015.The role of language processing in language acquisition. Linguist. Approaches
Biling. 5:409–53
Pickering MJ, Gambi C. 2018. Predicting while comprehending language: a theory and review. Psychol. Bull.
144:1002–44
Pliatsikas C, Marinis T. 2013. Processing empty categories in a second language: when naturalistic exposure
lls the (intermediate) gap. Biling.: Lang. Cogn. 16:167–82
Polinsky M. 2018. Heritage Languages and Their Speakers. Cambridge, UK: Cambridge Univ. Press
Pozzan L, Trueswell J. 2016. Second language processing and revision of garden-path sentences: a visual-world
study. Biling.: Lang. Cogn. 19:636–43
Rankin T, Grüter T, Hopp H. 2019. Investigating co-activation of L1 syntax during processing of wh-
questions: eye-tracking evidence from L1 German–L2 English. In Explorations in Second Language Ac-
quisition and Processing, ed. R Slabakova, J Corbet, L Domínguez, A Dudley, A Wallington, pp. 154–70.
Cambridge, UK: Cambridge Sch.
Roberts L. 2013. Sentence processing in bilinguals. See van Gompel, pp. 221–46
Roberts L, Felser C. 2011. Plausibility and recovery from garden-paths in second-language sentence process-
ing. Appl. Psycholinguist. 32:299–331
Roberts L, Gullberg M, Indefrey P. 2008. Online pronoun resolution in L2 discourse: L1 inuence and general
learner effects. Stud. Second Lang. Acquis. 30:333–57
¸Safak DF, Hopp H. 2021. Verb bias and semantic persistence effects in L2 ambiguity resolution. Second Lang.
Res. https://doi.org/10.1177/0267658321997904
Sagarra N. 2022. Sentence processing: cognitive approaches to L2 morphosyntactic and morphological pro-
cessing. In The Routledge Handbook of Second Language Acquisition and Psycholinguistics, ed. A Godfroid,
H Hopp. New York: Routledge. In press
Schmid MS. 2011. Language Attrition. Cambridge, UK: Cambridge Univ. Press
Schwartz BD, Sprouse R. 1994.Word order and nominative case in nonnative language acquisition: a longitu-
dinal study of (L1 Turkish) German interlanguage. In Language Acquisition Studies in Generative Grammar,
ed. T Hoekstra, BD Schwartz, pp. 71–89. Amsterdam: John Benjamins
Selinker L. 1972. Interlanguage. Int. Rev. Appl. Linguist. 10:209–31
Sorace A. 2011. Pinning down the concept of “interface” in bilingualism. Linguist. Approaches Biling. 1:1–33
Stowe LA. 1986. Parsing WH-constructions: evidence for on-line gap location. Lang. Cogn. Proc. 1:227–45
Sturt P. 2003. The time-course of the application of binding constraints in reference resolution. J. Mem. Lang.
48:542–62
www.annualreviews.org •Second Language Sentence Processing
Annu. Rev. Linguist. 2022.8:235-256. Downloaded from www.annualreviews.org
Access provided by WIB6277 - TU Braunschweig on 05/31/23. For personal use only.
Tanner D, Nicol J, Herschensohn J, Osterhout L, Biller AK, et al. 2012. Electrophysiological markers of
interference and structural facilitation in native and nonnative agreement processing. In Proceedings of
the 36th Boston University Conference on Language Development, pp. 594–606. Somerville, MA: Cascadilla
Teubner-Rhodes SE, Mishler A, Corbett R, Andreu L, Sanz-Torrent M, et al. 2016. The effects of bilingualism
on conict monitoring, cognitive control, and garden-path recovery. Cognition 150:213–31
Tokowicz N. 2015. Lexical Processing and Second Language Acquisition. New York: Routledge
Trueswell JC, Sekerina I, Hill NM, Logrip ML. 1999. The kindergarten-path effect: studying on-line sentence
processing in young children. Cognition 73:89–134
Ullman MT. 2016. The declarative/procedural model: a neurobiological model of language learning, knowl-
edge, and use. In Neurobiology of Language, ed. G Hickock, SL Small, pp. 953–68. New York: Academic
Valdés Kroff JR, Dussias PE, Gerfen C, Perrotti L, Bajo MT. 2017. Experience with code-switching modulates
the use of grammatical gender during sentence processing. Linguist. Approaches Biling. 7:163–98
van Assche E, Duyck W, Hartsuiker RJ. 2012. Bilingual word recognition in a sentence context. Front. Psychol.
3:174
van Bergen G, Flecken M. 2017. Putting things in new places: Linguistic experience modulates the predictive
power of placement verb semantics. J. Mem. Lang. 92:26–42
van Gompel RP, ed. 2013. Sentence Processing. Hove, UK: Psychol. Press
van Hell JG, Tokowicz N. 2010. Event-related brain potentials and second language learning: syntactic pro-
cessing in late L2 learners at different L2 prociency levels. Second Lang. Res. 26:43–74
Westergaard M.2021. Microvariation in multilingual situations: the importance of property-by-property ac-
quisition. Second Lang. Res. 37:379–407
Williams J, Möbius P, Kim C. 2001. Native and non-native processing of English wh-questions: parsing strate-
gies and plausibility constraints. Appl. Psycholinguist. 22:509–40
Yan S, Jaeger TF.2020. Expectation adaptation during natural reading. Lang. Cogn. Neurosci. 35:1394–1422
Hopp
Annu. Rev. Linguist. 2022.8:235-256. Downloaded from www.annualreviews.org
Access provided by WIB6277 - TU Braunschweig on 05/31/23. For personal use only.
LI08_TOC ARjats.cls December 9, 2021 11:9
Annual Review
of Linguistics
Volume 8, 2022
Contents
How I Got Here and Where I’m Going Next
Sarah G. Thomason pppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp1
Argument Structure in Sign Languages
Vadim Kimmelman pppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp19
Advances in Morphological Theory: Construction Morphology
and Relational Morphology
Jenny Audring ppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp39
Perspective Shift Across Modalities
Emar Maier and Markus Steinbach pppppppppppppppppppppppppppppppppppppppppppppppppppppppppp59
Learning Through Processing: Toward an Integrated Approach
to Early Word Learning
Stephan C. Meylan and Elika Bergelson ppppppppppppppppppppppppppppppppppppppppppppppppppppp77
The Probabilistic Turn in Semantics and Pragmatics
Katrin Erk ppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp101
Coherence Establishment as a Source of Explanation in Linguistic
Theory
Andrew Kehler ppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp123
When Do Children Lose the Language Instinct? A Critical Review
of the Critical Periods Literature
Joshua K. Hartshorne ppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp143
Music and Language
David Temperley ppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp153
Crosslinguistic Corpus Studies in Linguistic Typology
Stefan Schnell and Nils Norman Schiborr ppppppppppppppppppppppppppppppppppppppppppppppppp171
On the Acquisition of Attitude Verbs
Valentine Hacquard and Jeffrey Lidz pppppppppppppppppppppppppppppppppppppppppppppppppppppp193
Meaning and Alternatives
Nicole Gotzner and Jacopo Romoli pppppppppppppppppppppppppppppppppppppppppppppppppppppppppp213
Second Language Sentence Processing
Holger Hopp ppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp235
Annu. Rev. Linguist. 2022.8:235-256. Downloaded from www.annualreviews.org
Access provided by WIB6277 - TU Braunschweig on 05/31/23. For personal use only.
LI08_TOC ARjats.cls December 9, 2021 11:9
Nominalization and Natural Language Ontology
Scott Grimm and Louise McNally pppppppppppppppppppppppppppppppppppppppppppppppppppppppppp257
Perfects Across Languages
Östen Dahl ppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp279
Speech and Language Outcomes in Adults and Children
with Cochlear Implants
Terrin N. Tamati, David B. Pisoni, and Aaron C. Moberly ppppppppppppppppppppppppppppppp299
Glue Semantics
Ash Asudeh ppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp321
Intonation and Prosody in Creole Languages: An Evolving Ecology
Shelome Gooden ppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp343
Navigating Accent Variation: A Developmental Perspective
Elizabeth K. Johnson, Marieke van Heugten, and Helen Buckler pppppppppppppppppppppppp365
Reverse Engineering Language Acquisition with Child-Centered
Long-Form Recordings
Marvin Lavechin, Maureen de Seyssel, Lucas Gautheron,
Emmanuel Dupoux, and Alejandrina Cristia ppppppppppppppppppppppppppppppppppppppppppp389
Stance and Stancetaking
Scott F. Kiesling pppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp409
Neurocomputational Models of Language Processing
John T. Hale, Luca Campanelli, Jixing Li, Shohini Bhattasali,
Christophe Pallier, and Jonathan R. Brennan ppppppppppppppppppppppppppppppppppppppppppp427
Semantic Structure in Deep Learning
Ellie Pavlick pppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp447
Deriving the Wug-Shaped Curve: A Criterion for Assessing Formal
Theories of Linguistic Variation
Bruce Hayes pppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp473
Structural, Functional, and Processing Perspectives on Linguistic
Island Effects
Yingtong Liu, Elodie Winckel, Anne Abeillé, Barbara Hemforth,
and Edward Gibson ppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp495
Errata
An online log of corrections to Annual Review of Linguistics articles may be found at
http://www.annualreviews.org/errata/linguistics
Annu. Rev. Linguist. 2022.8:235-256. Downloaded from www.annualreviews.org
Access provided by WIB6277 - TU Braunschweig on 05/31/23. For personal use only.
