Review of Research in Second Language Processing and Parsing
|
|
|
|
|
Review:
|
EDITORS: VanPatten, Bill and Jegerski, Jill
TITLE: Research in Second Language Processing and Parsing
SERIES TITLE: Language Acquisition and Language Disorders 53
PUBLISHER: John Benjamins
YEAR: 2010
Christos Pliatsikas, School of Psychology, University of Birmingham, UK
SUMMARY
This volume brings together a selection of 13 papers that were presented in the
first Second Language (L2) Processing and Parsing Conference, held in Lubbock,
Texas, US, in May 2009. The book begins with an introductory chapter, outlining
the current status of the field, and the papers are distributed into six parts,
according to their topic. This review will follow the same structure.
Part I: Introduction
In Chapter, 1 the editors start by pointing out the recent boom in L2 processing
research and identify the key terms related to this body of research, namely
processing and parsing, as well as the psycholinguistic methods that are widely
used in the field. Attention is subsequently turned to the role of native
language (L1), and the authors suggest that L1 processing strategies may
selectively affect processing in L2, depending on the linguistic features under
investigation. The final section is concerned with how native-like L2 processing
can become, and focuses on the Shallow Structure Hypothesis (SSH) (Clahsen &
Felser, 2006). Various researchers in this volume discuss their findings in
light of the SSH’s predictions and limitations, while at times also offering
alternative interpretations.
Part II considers the processing of relative clause (RC) attachment and
wh-movement in L2. Chapter 2 describes Dinçtopal-Deniz’s study on monolingual
Turkish participants, monolingual English participants and highly proficient
Turkish-English L2 learners in online and offline tasks investigating RC
attachment preferences in their L1 and L2. The online tasks contained both
globally and temporarily ambiguous sentences, the latter being disambiguated
based on animacy information present downstream within the sentence (e.g. The
author of the play [RC that was killed last month] was famous). These tasks
revealed that both monolingual groups showed low-attachment preferences,
irrespective of the animacy of the antecedent. The L2 group, however, revealed a
different pattern, in that animacy information affected their attachment
preferences. More specifically, L2 learners preferred high attachment when the
RC modified an animate Noun Phrase (NP) and low attachment when it modified an
inanimate NP. This suggests that, whereas monolingual participants are guided by
syntactic information (NP syntactic position, demonstrating a recency
preference), L2 participants are only guided by lexical information (animacy).
This, according to the authors, supports the SSH in that hierarchical structures
are largely ignored during L2 processing, which is informed by lexico-semantic
information.
Different suggestions are put forward by Aldwayan, Fiorentino and Gabriele in
Chapter 3, where it is proposed that non-native speakers are constrained by
syntactic information, similar to native speakers. More specifically, Aldwayan
et al. tested Najdi Arabic learners of English and native English speakers in an
adapted version of the Stowe (1986) self-paced reading study, in order to
investigate whether L2 learners of a wh-in situ L1 background are able to
process wh-movement. Experiment 1 included sentences with wh-extractions (e.g.
My bother asked who Barbara will photograph Sam beside at the graduation), and
revealed that both groups demonstrated longer reading times (RTs) at the
post-verbal site of the embedded sentence in the wh-movement sentences, compared
to the corresponding site in the non-wh-movement sentences. This was interpreted
as an erroneous attempt to integrate a wh-filler to a structurally defined gap,
which was revised upon processing of the existing object. In Experiment 2, the
embedded sentences in both conditions included a complex NP that prohibited
wh-extraction (NP-island), which also featured a preposition (about) that could
accommodate a gap. However, due to being on an NP island, this gap is not in a
grammatically licensed position, and as a result, no filled-gap effects should
be observed at this site, if parsing is structure-based. This was what the
authors reported, and importantly for both groups, suggests that sentence
processing is incremental for both L1 and L2 speakers. Additionally, it appears
that the parser takes into account abstract structural information by positing
gaps in grammatically valid positions only.
More evidence of L2 learners' sensitivity to wh-islands is provided by Cunnings,
Batterham, Felser and Clahsen in Chapter 4. Cunnings et al. tested German and
Chinese L2 learners of English, along with native English speakers, in an
eye-tracking experiment based on Traxler and Pickering (1996). They monitored
eye-movements during the reading of sentences with wh-dependencies that either
included wh-islands or not, crossed with the plausibility of the filler as the
object of the first available verb, yielding four conditions. In the wh-island
conditions, the verb was always in the wh-island, so the prediction was that no
gap should be posited and processed at that point, at least by native speakers.
When implausible fillers were present (e.g. the city that the author wrote),
elevated RTs were expected at the site of the verbal object, as erroneous
integration should be followed by reanalysis; but this should take place only in
the non-island condition, as processing of the island (and the verb) in the
island conditions should suspend the search for a gap. This was found for all
three groups, indicating that all participants postulated a wh-gap in the
post-verbal region in the non-island conditions only. Importantly, this was not
related to the learners' L1 typology (wh-movement (German) vs. wh-in-situ
(Chinese) languages), in accordance with the suggestions by Aldwayan et al. in
the previous chapter. The only difference between native and non-native
participants was an apparent difficulty for the latter in forming a
wh-dependency when a wh-island is present, demonstrated as longer RTs at the
site of the ultimate gap.
Part III focuses on the processing of gender and number. Chapter 5 is based on
Keating’s eye-tracking experiment, where native and advanced L2 speakers of
Spanish were presented with sentence violations in adjective-noun gender
agreement. While keeping the structural distance constant, Keating manipulated
the linear distance between a control noun and the site of the adjective in
order to investigate whether linear distance, and the working memory demands it
imposes, affects L2 processing of agreement violations. The noun-adjective
distances were one, four or seven words, forming three conditions. Keating found
that the noun-adjective linear distance affected processing of agreement
violations by both groups, demonstrated as longer fixations on ungrammatical
adjectives, compared to those in grammatical (control) conditions. Natives were
sensitive to violations in one- and four-word distances, whereas non-natives
were sensitive to one-word distance only. Notably, native speakers showed their
sensitivity to the violations during first-pass reading, whereas this occurred
with non-natives during second-pass reading, a finding that Keating attributes
to working memory differences between the groups. Keating interprets these
results as evidence for shallow processing by both groups, which is
substantiated earlier for non-native speakers. Keating also attributes the
observed between-group differences to processing limitations in non-native
speakers, rather than representational ones.
In Chapter 6, Renaud focuses on early stages of acquisition of number agreement
in L2 French by administering a task combining self-paced reading (SPR) and
acceptability judgment. The segmented sentences in the SPR task were preceded by
sentences that provided context, and could either be grammatical but unrelated
to the context, or context-compatible but ungrammatical because of lack of
number agreement. The participants had to judge offline whether the segmented
sentences fit the preceding context, a task that required them to read every
segment, and, as a result, detect the number mismatch in the context-compatible
sentences. Additionally, the experimental sentences contained both an auxiliary
verb and a past participle, the number of which was manipulated in different
conditions. The results suggest that number mismatch was processed by more
advanced L2 learners of French for both auxiliaries and participles. Renaud
suggests that L2 learners are able to select the necessary features for
comprehension in L2, and this selection takes place earlier than checking for
the (mis-)match of those features. This checking is indicated by longer RTs for
number mismatches, compared to matches, which also signifies a re-assembly of
features.
Part IV is concerned with how L2 learners process word order in their non-native
language, especially in cases where word order follows different rules or is not
crucial for comprehension. For example, in Chapter 7, Mitsugi & MacWhinney focus
on the heavy reliance on case marking in encoding grammatical relationships in
Japanese, as opposed to the reliance on word order that is common in other
languages. They tested native Japanese speakers in addition to Japanese learners
with L1s that do or do not rely on case-marking (Korean and English,
respectively). Participants were presented with an SPR experiment in which
stimuli included sentences with ditransitive constructions (i.e. including verbs
with two objects) that appeared in four word orders: canonical, dative
scrambling, accusative scrambling and dative-accusative scrambling. Mitsugi &
MacWhinney did not find statistically significant differences in RTs among the
four conditions in either of the groups. The authors interpret this finding as
evidence that thematic assignment was successfully carried out based on surface
cues (case markers), which also assist in building expectations for upcoming
constituents. Importantly, a similar result was revealed for L2 learners,
independently of their linguistic background. The authors conclude that Japanese
scrambling does not pose any increased processing load to either native or
non-native Japanese speakers, pointing to a cue-based processing strategy.
In a similar vein, in Chapter 8, Hara investigates L2 processing of syntactic
gaps in Japanese sentences with scrambling. The verb-final nature of Japanese
means that gaps are posited before the verb and its argument structure become
available. Hara used sentences with double object constructions, where he
dislocated the direct object to either a short or a long distance (short vs.
long scrambling) from its canonical preverbal position, and tested Korean and
Chinese L2 learners of Japanese, as well as native speakers. Hara predicted that
the displaced element should reveal effects of gap processing in its canonical
position, and that long scrambling should pose greater processing difficulty
than short scrambling, demonstrated as longer RTs at the gap position. This was
found true for native speakers only; Korean learners only revealed a slow down
at the gap position for short scrambling, whereas Chinese learners showed no
evidence of processing the gap in either condition. Hara interpreted his
findings in light of the Simple Syntax approach (Culicover & Jackendoff, 2005),
according to which L2 processing stands somewhere in between being based on
structure and on verb agreement and pragmatics, maintaining the capability of
computing some legitimate representations. The absence of any effects for
Chinese learners, whose L1 does not permit scrambling, may indicate L1 influence.
Another SPR study, by Jackson, in Chapter 9, investigates L2 processing of
subject-object ambiguities by focusing on transfer effects from L1. Jackson
followed up from a previous study of hers (Jackson, 2008) and tested English and
Dutch L2 learners of German in sentences with temporarily ambiguous
wh-questions, which were later disambiguated via case-marking information into
subject-first or object-first sentences. German permits the lexical verb to
appear in clause-final position, and therefore, native speakers need to assign
grammatical roles prior to encountering the verb. Jackson investigates whether
the same processing is available to learners with an L1 that either does or does
not permit (Dutch and English, respectively) the verb to appear in final
position. She manipulated the position of the verb (verb-second vs. verb-final
position), i.e. before or after a disambiguating region. Jackson's results
reveal that all groups preferred a subject-first interpretation, demonstrated as
longer RTs at the disambiguating region in the object-first sentences, which
notably posed more difficulty for the English group. Jackson concluded that, in
general, L2 learners utilise morphosyntactic information in a native-like
fashion, while some L1 influence is possible.
Chapter 10 features a study by Malovrh and Lee, who investigate acquisition of
object pronouns and processing of word order in L2 Spanish. They administered
two offline tasks (comprehension and production) to university students enrolled
in four different Spanish courses at varying levels. The comprehension task
involved listening to two-sentence sequences per experimental trial, where the
second sentence always started with an object pronoun. The participants were
instructed to identify the subject of the second sentence, and were expected to
interpret the object pronoun as the object of the verb, which would indicate
successful processing of the object-verb-subject (OVS) word order. The
production task involved silent film extracts, and the participants were
instructed to narrate the film twice, the first time by describing what they
were seeing, and the second time by putting themselves into the action of the
film. The results reveal that production, placement and comprehension of OVS
sentences increases with proficiency; however, a developmental effect was found
in that learners in a fifth-semester course (the second lowest of the four
courses) demonstrated a drop in accuracy that was concurrent with the emergence
of third person pronouns. The authors propose that there is a connection among
OVS production, placement and comprehension, in terms of a ''common development
of syntax-before-morphology''.
Part V focuses on processing that precedes the sentence level, more
specifically, at the levels of phonology and the lexicon. In Chapter 11,
Shoemaker investigates processing of continuous speech by L2 learners of French,
with particular focus on the particularities of French phonology, such as
liaison (i.e. when a consonant in word-final position is realised phonologically
only when the next word starts with a vowel). These particularities may posit
identification of word boundaries as very demanding for language learners. She
created phonologically ambiguous phrases which included four vowel-initial words
always preceded by one of three consonants that are realised in liaison
environments in French, namely /n, t, z/. Ambiguity was introduced by the
artificial shortening or lengthening of the liaison consonants, while the rest
of the phrase was kept constant. Native and non-native speakers of French were
tested in an auditory discrimination task, where they had to decide whether the
two versions of the manipulated phrases were identical or not, and a
forced-choice identification task, where they heard one version of the
manipulated phrase and had to choose among two visually presented possible
interpretations of the ambiguous word. Shoemaker's results suggest a native-like
sensitivity of the non-native speakers of French to the durational variation of
the stimuli, which affected their interpretations, indicating that non-native
speakers can acquire and process phonetic detail in their L2.
In Chapter 12, Tokowicz & Degani review studies concerned with translation
ambiguity, which is when a word in one language can have several potential
translations in another language (cross-language ambiguity). The authors
identify several sources of within-language ambiguity: a) homonymy, where a word
has more than one meaning (e.g. 'calf'); b) polysemy, when several meanings of a
word are semantically related to each other (e.g. 'clown'); c) near-synonymy,
when an object or an abstract concept can be described with several different
terms (e.g. 'couch' and 'sofa'). They continue by arguing that these same
sources are potential sources of cross-language ambiguity, demonstrated as
translational ambiguity, and focus on the effects of (a) and (b) in L2
processing. After reviewing several studies on this particular issue, the
authors suggest that translational ambiguity has a deteriorating effect on
participants' processing in that it reduces their performance speed and their
accuracy in online tasks. They attribute this finding to competition among
various translational candidates, which is mediated by the strength of
associations between the lexical entry and its various interpretations.
Additionally, Tokowitz & Degani show that ambiguity effects are present in
various levels of L2 proficiency. Proficient bilinguals are less affected by
pure synonyms compared to less proficient L2 learners.
Part VI concludes the volume with papers dealing with the interplay between
syntax and prosody, as well as syntax and discourse context. In Chapter 13,
Fernández analyses recorded speech of balanced Spanish-English bilinguals and
early bilinguals (L2 age of acquisition <5 years) in order to investigate
whether different syntactic categories correspond to different aspects of
prosody, as well as whether these correspondences differ across the two
languages of a bilingual. The participants followed a protocol that elicited
both informal and formal spontaneous speech in English, formal reading in both
English and Spanish, and informal spontaneous speech in Spanish. Fernández
analysed various fluency markers (durations of words and pauses, number and
types of disfluencies), as well as data on the likelihood of speakers pausing at
different clause boundaries, which indicate prosodic phrase divisions. Her
results suggest that bilingual speech production follows predicted prosody
patterns which correspond to underlying syntactic structure. Most importantly,
this was shown for both languages of speakers. Fernández concludes by suggesting
that prosodic phonology may have an effect on syntactic processing, such as in
ambiguity resolution, and therefore, she stresses the importance of the
syntax-prosody interface in L2 acquisition.
In Chapter 14, the last paper in this volume, Reichle reports on an
Event-Related Potentials (ERP) study on the processing of contrastive focus in
L2 French. Focus in spoken French heavily relies on syntactic constructions such
as c'est (it’s), an uncommon strategy in languages like English, where focus is
marked by accent and prosody. Reichle manipulated processing of informational
vs. contrastive focus by native speakers of French and two groups of
French-English L2 learners of different proficiency levels (low vs. high). Based
on previous findings (Cowles, 2003), Reichle investigated whether the two types
of focus elicit different ERP effects for native and non-native speakers of
French, and whether L2 proficiency affects processing of focus. His results
reveal that native speakers elicit an early negativity (between 200-400 ms) for
contrastive vs. informational focus, an effect that was also revealed for L2
learners of high proficiency. Reichle interprets this negativity as evidence for
increased Working Memory (WM ) load for the contrastive condition. On the other
hand, learners of low proficiency did not reveal any processing difference
between the two conditions, suggesting a native-likeness for highly proficient
learners.
EVALUATION
This volume represents a good collection of cutting-edge research papers on the
contemporary topic of L2 processing and parsing. Being the product of one of the
few conferences that specialises on the topic, it presents the work of
well-known researchers in the field, as well as several new and promising ones.
The volume samples papers focusing on a variety of structures in L2, and how
they are processed by learners in the domains of morphology, syntax,
morphosyntax, lexicon, prosody, etc. These papers utilise a variety of offline
and online behavioural methods, most notably SPR, which is common in several
studies, while Reichle also uses ERPs. Therefore, this volume provides a good
account of how contemporary research is advancing, as well as the topics that
engage researchers, and the methods that are primarily used. Modern approaches
to L2 processing, such as the SSH are complemented (Dinçtopal-Deniz, Cunnings et
al.) and challenged (Aldwayan et al.), whereas alternative theories, such as
Simpler Syntax, are sometimes proposed (Hara) as explaining L2 processing and
parsing.
The structure of the book is particularly organised for the reader because of
the allocation of papers in thematic parts, which allow for similar, and
sometimes complementary, studies (such as the Mitsugi & MacWhinney, and Hara
studies) and methodologies to be contrasted and interpreted together. As a
result, this volume also functions as an excellent summary of ongoing research,
and also as an inspiration for future studies that will address the issues
proposed by the featured studies.
This book is highly recommended to scholars and students in the field of second
language acquisition and processing, especially psycholinguists and neurolinguists.
REFERENCES
Clahsen, H., & Felser, C. (2006). Continuity and shallow structures in language
processing. Applied Psycholinguistics, 27, 107-126.
Cowles, H. W. (2003). Processing information structure: Evidence from
comprehension and production. University of California, San Diego.
Culicover, P. W., & Jackendoff, R. (2005). Simpler syntax: Oxford University
Press, USA.
Jackson, C. N. (2008). Proficiency Level and the Interaction of Lexical and
Morphosyntactic Information During L2 Sentence Processing. Language Learning,
58(4), 875-909.
Stowe, L. A. (1986). Parsing WH-constructions: Evidence for on-line gap
location. Language and Cognitive Processes, 1(3), 227 - 245.
Traxler, M. J., & Pickering, M. J. (1996). Plausibility and the processing of
unbounded dependencies: An eye-tracking study. Journal of Memory and Language,
35, 454-475.
ABOUT THE REVIEWER
|
| |
ABOUT THE REVIEWER:
Christos Pliatsikas received his PhD from the Department of Clinical
Language Sciences, University of Reading. His research interests are in the
area of psycholinguistics and neurolinguistics. He has used behavioural and
neuroimaging (fMRI) methods for the study of online morphological and
syntactic processing, especially by late second language learners. He is
currently employed as a Research Fellow by the School of Psychology,
University of Birmingham.
|
|
|
|
|
|
