The eye-tracking method in interpretation studies

A pioneering study which applied the eye-tracking method to interpretation studies

was conducted by McDonald and Carpenter (1981). The study examined the processes in

sight translating ambiguous phrases and how these phrases were parsed during

comprehension and interpretation. Also, how errors were detected during sight translation

was also investigated. Two expert interpreters and two amateur German-English bilinguals

sight translated 44 texts from English into German while their eye fixations and oral

translations were recorded. The texts contained idiomatic phrases such as “hit the nail on

the head” or “break the ice”, which could be comprehended and interpreted literally or

idiomatically under two different contexts which either primed a literal interpretation or an

idiomatic interpretation. This was done by analyzing the chunking of ambiguous phrases

manifested through eye movement indices. A chunked unit was defined by the farthest word

to the right that was fixated in a first-time forward fixation scan before a regression, also

called the first-pass reading. Comprehension of an idiomatic phrase was classified as

idiomatic if the entire phrase was read as one chunk without any regressions in first-pass

reading. For example, the entire phrase of Mike hit the nail right on the head, if read in one

pass, would be regarded as idiomatic comprehension. In contrast, comprehension was

classified as literal if the phrase was chunked at any of the syntactic boundaries in the

/ right on the head, both not read in one entire chunk but parsed by regressions at syntactic

boundaries (as indicated by the slash), were both classified as literal comprehension.

Interpreters’ eye fixation patterns not only demonstrated how idiomatic phrases were

interpreted but also reflected the subprocesses of interpretation. Each phrase received at

least two scans: the initial comprehension of the phrase was marked by the initial sequence

of forward gazes, which is the first-pass reading in eye movements. The subsequent process

was marked by regressions, which constituted the second scan or the rereading of the

phrases. This second scan (also referred to as second-pass reading) was also when oral

interpretation took place. Further, if the participant detected an error in his/her

comprehension/interpretation after the second pass, the eye regressed back again to the

preceding part which contained the ambiguous idiomatic phrase while the previous oral

interpretation was corrected.

The results of this study showed that interpreters’ eye fixation patterns depended on

the preceding context as well as semantic and syntactic cues. Although this study was not

aimed at exploring the process of sight translation, it revealed that the initial process of

comprehension in sight translation was similar to the comprehension processes that occur in

normal English reading. This was evidenced by the fact that the reading speed for the initial

reading of a phrase in sight translation was similar to that for normal reading (about 200 to

300 words/minute). These results supported the vertical perspective, because readers need

to comprehend first before proceeding to interpret a meaning unit.

Figure 3 depicts a model proposed by McDonald and Carpenter (1981). The first phase

“read, integrate and parse” was manifested in the first-pass reading of eye movements. The

second phase “reread for lexical retrieval in target language” was manifested in rereading,

while error recovery was manifested by regressions to the previous fixated regions, as

shown by the two arrows pointing back to the previous regions on the right hand side.

Figure 3. The processes of the interpretation task (McDonald & Carpenter, 1981)

In sum, this research was groundbreaking for it applied the eye-tracking method to

interpretation study, yielding insightful information to the parsing of ambiguous phrases,

error detection, as well as the process of comprehension in sight translation. Despite the

findings, this study was not intended to determine the validity of the vertical and horizontal

phrase in sight translation was similar to that of normal reading, normal reading was not

included as one of the conditions in the experiment. Until the present study, there has been

a lack of follow-up research which continued the investigation into sight translation with

the eye-tracking method. Moreover, there have not been studies targeted at comparing the

process of sight translation with normal reading in Chinese nor the respective components

of sight translation (comprehension, reformulation, and production).

Other studies used the pupil diameter as a measurement in interpretation research

(Hyönä, et al., 1995; Tommola & Hyönä, 1990; Tommola & Niemi, 1986), since pupil

diameter was positively correlated with cognitive loading (Rayner, 1998). Chang (2009)

commented that Tommola and Niemi (1986) were the first to use pupil diameter as a

measurement of cognitive loading during simultaneous interpreting. In this study, one

participant conducted simultaneous interpreting (SI) of five Finnish texts into English. The

results showed that the participant’s pupil diameter was the highest during interpreting

when restructuring of the output English sentence was required because of the syntactic

differences between the source language (Finnish) and the target language (English).

Another example of the aforementioned studies was undertaken by Hyönä et al. (1995)

to understand the cognitive processing load in simultaneous interpreting (SI), speech

shadowing, and passive listening. Two experiments were conducted in this study: in

Experiment 1, nine Finnish-English T&I students conducted SI, shadowing, and passive

listening to a text of 500-600 words from English into Finnish. The findings showed that SI

produced a higher degree of pupil dilation than did speech shadowing, and speech

shadowing yielded a higher level than did listening (average pupil diameter:

listening=4.20mm, shadowing=4.72mm, SI=5.22mm). In Experiment 2, 18 T&I students

participated in a 2x3x2 within-subject design experiment. The factors were input language

(Finnish and English), task type (listening, shadowing, oral translation), and word

translatability (easy, difficult). Effect of task type showed that overall average increase in

pupil size was largest for lexical translation (0.33mm), followed by shadowing (0.24mm),

and then listening (0.16mm). Effect of language direction indicated that the pupil dilated

more for English words (0.27mm) than for Finnish words (0.22mm), demonstrating that

repeating back words in a non-native language (English) was accompanied by increased

pupil dilations, in comparison to repetition in the subject’s native language. Effect of word

translatability was also manifested: words that were determined to be more difficult to

translate induced a higher increase in pupil size (0.29mm) than did easily translatable words

(0.20mm). This study lent strong support to the use of the pupillary response as an indicator

of cognitive processing load.