Processing Wh-Dependencies in a Second Language

Subject: Linguistics
Pages: 5
Words: 1583
Reading time:
6 min
Study level: PhD

The processing of wh-dependencies by non-native speakers has been defined and assessed by many scholars. The article Processing Wh-dependencies in a Second Language: a Cross-Modal Priming Study by Claudia Felser and Leah Roberts is aimed at investigating “the real-time processing of wh-dependencies by advanced Greek-speaking learners of English using a cross-modal picture priming task” which previously has never been used with non-native speakers (Felser, Roberts, 2007, p. 9). In particular, in the course of the experiment, “the participants were asked to respond to different types of picture target presented either at structurally defined gap positions or at pre-gap control positions while listening to sentences containing indirect-object relative clause” (Felser, Roberts, 2007, p. 9). Thus, the article in question presents basically a review of research that defends a theoretical framework on the topic alongside the author’s own study.

In this paper, I provide the summary of the article and comments about some aspects, identify some relative divergence of views concerning the issue of wh-dependencies, and put forward some ideas where additional research would assist in understanding the current state of affairs in approaching the phenomenon of wh-dependencies in a second language.

Claudia Felser and Leah Roberts conduct their study on the basis of early linguistic research on the topic. The authors investigate the background of the issue carefully and present a full picture of research and contributions made by the linguists on this topic thus far. The article gives the coverage of the theories of Frazier and Clifton “. Besides, the authors provide the results of the study conducted by King and Just, who treat the problem from the point of event-related brain potentials (ERPs). Alongside this, the controversial views concerning the present issue are presented. According to the Direct Association Hypothesis (DAH), “dislocated constituents are linked directly to their lexical subcategorizer” and semantically associated with its structure or thematic grid, whereas Chomsky argues that “filler integration is mediated by empty syntactic categories known as “traces” in generative-transformational grammar” (Felser, Roberts, 2007, p.11). The authors touch upon various approaches of “trace-based reactivation on the studies of verb-final languages”: Clahsen and Featherston, Miyamoto and Takahashi, Nakano and Fiebach; studies of ”investigating the processing of indirect-object dependencies”: Nicol, Roberts; “subject-relative clauses”: Swinney and Zurif, Lee; and “dependencies spanning more than one clause”: Gibson and Warren, Marinis (Felser, Roberts, 2007, p. 12). The article also provides the examination of the studies that have investigated the way second language (L2) processes filler-gap dependencies in real-time, which are few, however: Williams, Juffs and Harrington, Marinis, Gibson, and Warren.

Besides, the article gives a full description of the theories and methods that served as a basis for the study, namely, the works of Nicol and Swinney, Zurif, Love, Clahsen (antecedent-priming effects at gap locations observed in both adult native speakers), and Nicol and Swinney, Roberts (antecedent-priming effects at gap locations observed in monolingual children). Alongside this, the authors provide their own treatment of the subject, methods, and results.

As it has already been mentioned, the article treats the issue of the real-time processing of wh-dependencies by advanced Greek-speaking learners of English using a cross-modal picture priming task which provides a useful tool for examining whether dislocated constituents are mentally reactivated at particular structural positions taking into account various previous treatments of the subject, such as Frazier and Clifton’s “ native (L1) sentence processing, encountering a dislocated constituent (or “filler”) which is thought to trigger the prediction of a lexical head to license it, or of the corresponding syntactic gap” (Felser, Roberts, 2007, p. 4), Chomsky’s “generative-transformational grammar” or various approaches of trace-based reactivation on the studies of verb-final languages. The study is targeted at the exposure of differences in processing the wh-dependencies between native speakers of English and Greek users speaking English as a second language.

In the course of the study, the authors set 2 main aims:

  • “To investigate whether or not the filler integration is mediated by syntactic gas in L2 processing; and
  • To examine whether non-native processing of wh-dependencies is influenced by individual WM differences” (Felser, Roberts, 2007, p. 20).

The Shallow Structure Hypothesis for L2 processing claims that “syntactically defined gaps, or ‘traces,’ are absent from the mental representations constructed during L2 sentence processing.” The Trace Reactivation Hypothesis states that promising antecedent effects should be found in the position of the indirect object gap. Both these hypotheses have been brought up for consideration (Felser, Roberts, 2007, p. 21).

The study has been conducted according to the following mode. “Twenty-four Greek-speaking learners of English (mean age: 25.17; range: 20–31), all of whom students of Essex, UK, performed in the experiment with 54 mature native speakers of English (mean age; 22.8; range; 19–42) and 44 monolingual English-speaking children (mean age; 6.25; range; 5–7) from Roberts et al.’s (2007) study”. “The participants were not informed of the ultimate purpose of the main experiment.” (Felser, Roberts, 2007, p. 23).

Before the first experiment, when “the Greek-speaking participants had been first exposed to English aged between 6 and 11 in a classroom setting or during private lessons, the learners completed the Oxford Placement Test and English proficiency Test indicating that all of them were advanced learners of English”. (Felser, Roberts, 2007, p. 25). The participants also underwent a reading span test, which allowed selecting the group with relatively high scores.

“The materials compressed 20 experimental sentences containing indirect-object relatives, plus 60 filler sentences similar in length to the experimental ones” (Felser, Roberts, 2007, p. 23).

To prevent participants from focusing their attention on specific points during the sentences, 12 of the fillers were structurally similar to the experimental sentences but with usual targets presented at positions other than the critical test points. “The remaining 48 fillers included constructions of different types” (Felser, Roberts, 2007, p. 25).

All 80 sentences were read by a female native speaker of English, with natural intonation, and pre-recorded on a digital tape recorder (Felser, Roberts, 2007, p. 25).

It is reported that “for each experimental sentence, two visual targets were selected: an ‘identical’ picture target showing the referent of indirect object noun and a picture showing an unrelated object.” (Felser, Roberts, 2007, p. 25). The experimental sentences were ensured to be listened to only once.

It is pointed out that each participant was tested individually, seated in front of the monitor, and asked to listen carefully, to watch the screen for pictures, and decide as quickly as possible whether the animal in the picture was alive or not alive by pushing different buttons.

It is also highlighted that “the conducted study shows that participants were highly accurate in the aliveness decision task and answered 91.8% (adult 96%; children 86%) identifying 96.3% of the picture targets as either ‘alive’ or ‘non-alive.’ These results indicate that the learners had no difficulty comprehending the experimental sentences, or coping with dual-task demands of the cross-modal priming experiment” (Felser, Roberts, 2007, p. 25).

Although all learners had demonstrated a relatively high level of general proficiency in English, a parallel ANOVA with proficiency was carried out but showing no significant interactions (Felser, Roberts, 2007, p. 26).

In the group of L2 learners vs. adult NSs, the difference was in that only adults displayed a position-specific advantage for targets matched to the gap. In other words, the learners could recognize pictures of a wh-filler more easily than pictures that were unrelated to any of the sentence’s components, but this was not accounted for by the structural position at which these pictures were shown. It is also stated that L2 learners but not the low-span adults displayed shorter RTs to matched targets (Felser, Roberts, 2007, p. 25).

As for results in the group L2 learners vs. children, both “the high-span and the low-span children’s RTs were longer over all the learners’, and the learners’ RT model differed from those of both the high-span and the low-span children” (Felser, 2007, p. 27). The high-span children pattern with the high-span adult NSs differs from the L2 group in that they monitored shorter RTs to matched targets at the point of the gap only, whereas with the low-span children exact opposite model was observed – longer RTs to matched targets at both test points (Felser, Roberts, 2007, p. 27).

The authors present convincing research supporting their hypotheses by derived results. First, the hypothesis that the presentation constructed during L2 processing lacks such abstract grammatical ingredients as movement traces was proved. It was observed that the learners had no difficulty in comprehending complex sentences of the types under consideration. This may be accounted for by the fact that the learners are able to compensate for their relatively shallower grammatical analyses by exploiting lexical, pragmatic, and other cues (Felser, Roberts, 2007, p. 28).

The strongest points of this study are the vast coverage of material and the great number of ideas and viewpoints that the authors offered to support their hypothesis. The study includes research references and acknowledgments that increase the reader’s attention and confidence. Besides, the article provides examples and findings of the study in a logically consistent, readable way.

Unfortunately, the study covers only the results conducted with the Greek-speaking learners of English, which may affect the comprehension of the issue with non-Greek-speaking learners of English. In other words, the study presents a restricted specific character.

Thus, the results give room to advanced research on processing wh-dependencies in a second language. Moreover, it presents a full and extended description of the methods applied, experiments conducted, and results derived, which can serve as a basis for further research.


Felser C., Roberts L., (2007). Processing Wh-dependencies in a Second Language: a Cross-Modal Priming Study. Second Language Research, Vol. 23, № 1, 9–36.