Deilen, Silvana: An Example of an Eye-tracking Study in Easy Language Research (and Some Remarks on the Challenges)

In recent years, eye-tracking has been increasingly used in studies evaluating the effectiveness of the postulated rules for Easy Language (see, for example, Gutermuth 2020, Wellmann 2020). In my PhD thesis, I used eye-tracking data to determine whether segmenting complex nouns (so-called compounds, i.e., nouns that consist of more than one morpheme) facilitates or hinders cognitive processing. In this blog, I summarize some of my main findings and highlight some important requirements and challenges that researchers should be aware of when conducting eye-tracking experiments with Easy Language target groups.

In German Easy Language, four guidelines (Inclusion Europe 2009, BITV 2.0 2011, Netzwerk Leichte Sprache 2013, Bredel/Maaß 2016) postulate that compounds are to be visually segmented to facilitate lexical access to the compound’s constituents. However, there is no consensus on the best visual structuring sign to use for segmenting compounds. The first practical rulebooks of Easy Language recommend segmenting compounds with a hyphen (Hunde-Hütte [dog house]). Despite the lack of scientific proof, the rule of segmenting compounds, or more specifically “long words” (Inclusion Europe 2009: 23; BMAS 2013: 26) or “compound nouns” (BITV 2.0 2011: attachment 2, Part 2), with a hyphen has been officially enshrined in law and is consequently implemented in a wide range of German texts. However, as the above-mentioned rule was formulated without cognitive-scientific foundation and has many linguistic, social, and educational shortcomings, the German Research Centre for Easy Language has suggested structuring compounds with a hyphenation point called the mediopoint (Hunde·hütte) (for a detailed discussion on this, see Bredel/Maaß 2016; Maaß 2020). Even though the rule of segmenting compounds using a mediopoint seems to be less fraught with disadvantages and problems, we still do not know whether enhancing the perception of the compound’s constituents (either by using a hyphen or mediopoint) also reduces the cognitive processing costs for Easy Language target groups. This gap in research constituted the starting point of my study. My aim was to examine and support the postulated rule of segmenting compounds in Easy Language. Therefore, my study was not only of neurologically unimpaired readers but also deaf and hearing-impaired pupils, as they represent one of the heterogenous target groups of Easy Language.

As I recorded the participants’ eye-movements using a mobile eye tracker, they did not need to leave their familiar surroundings, and I decided to conduct my study of the target group at their school for the deaf and hearing impaired. A mobile eye tracker is particularly useful because it allows researchers to conduct experiments in different locations, enabling them to reach many of the heterogenous Easy Language target groups. I conducted four different eye-tracking experiments, in all of which I presented noun-noun compounds as either written with no visual structuring signs (Regenjacke [rain jacket]), or in one of two formats, in which I signalled constituent boundaries using visual cues, i.e., by inserting a hyphen (Regen-Jacke) or a mediopoint (Regen·jacke). For data analysis, I looked at the following eye-tracking parameter: first fixation duration, number of fixations, total reading time, and rate of regressions (regressions are eye movements in the direction opposite to that of normal reading, which means that the reader jumps backwards to the word). Below, I list some of my findings without going into too much detail. (Feel free to contact me if you have any questions)

1. The benefit of segmenting compounds depended on reading proficiency. Although the less-skilled readers of the Easy Language target group clearly benefited from the segmentation cue (i.e., they read segmented compounds faster than unsegmented compounds), it seemed that for the deaf and hard-of-hearing pupils with higher literacy skills, segmentation of the compounds was unnecessary (i.e., they processed unsegmented compounds faster than segmented compounds).

2. Segmenting compounds with a mediopoint was – in most cases – more effective than segmenting compounds with a hyphen. This was especially true for pupils with low literacy skills.

3. Processing compounds with context was easier than processing compounds without context (i.e., compounds that were presented in isolation).

4. Segmenting compounds with a hyphen led to morpheme-based, compositional processing, in which the reader processes the constituents individually rather than as a whole.

5. The first fixation duration, which is the duration of the first fixation on the compound, did not enable valid conclusions regarding the overall processing effort.

Even though using online methods such as eye-tracking offers many advantages and valuable insights into cognitive processes, I would now like to mention some aspects that should be considered when conducting eye-tracking experiments with heterogenous Easy Language target groups.

First, you need to consider that some members of Easy Language target groups might not be able to communicate their needs sufficiently or might not be able to understand all the information and instructions. Consequently, it is crucial to adjust the experimental procedure to their individual needs and requirements. This refers not only to the duration of the experiment (i.e., longer sessions should be split) but also to informing the participants of personal data protection and their right to opt out of the study at any time. Researchers should only start the experiment when they are certain that all the information is fully understood. It goes without saying that not only the participants’ consent, but also (for underage participants) parents’ consent or (for participants with intellectual disabilities) the consent of the participant’s legal guardians is required.

During data collection, it is important that the participants remain still in front of the computer. However, as limiting body and head movements for a longer time might be really difficult for some participants, it is important that the experiment is designed in a way that allows the participant to take a break whenever they want. Wherever possible, the researcher should try to eliminate distracting factors and design the experimental setting to be as natural as possible.

Furthermore, not only the instructions, but also the task itself should be kept simple when working with Easy Language target groups. For example, you need to bear in mind that many participants are probably not used to using a computer and might not be able to understand single instructions (e.g., in terms of calibration and validation). Therefore, it is important to give participants some extra time to practice.

When analysing the data, it is important to keep in mind that data cleansing cannot be as strict as usual, but needs to be adjusted to the above-mentioned facts and eventualities. This means that the researcher should allow more liberal data cleansing, in which they also look at individual trials to evaluate whether the data are useful or not. This of course also means that conducting experiments with impaired participants will probably lead to higher data loss and that the outcomes and results will not be comparable to those of participants with no impairments.

However, on a final note, I would like to point out that, looking back, the collection of eye-tracking data on deaf and hard-of-hearing pupils was (at least in my study) very successful, and provided valuable insights into the processing of compounds. Therefore, I encourage other researchers to also use eye-tracking in Easy Language research.

Silvana Deilen
Research Associate at the Department of English Linguistics and Translation Studies, Johannes Gutenberg University Mainz in Germersheim.

The article is based on my PhD thesis (written in German):
Deilen, Silvana (2021): Optische Gliederung von Komposita in Leichter Sprache: Blickbewegungsstudien zum Einfluss visueller, morphologischer und semantischer Faktoren auf die Verarbeitung deutscher Substantivkomposita. Frank & Timme: Berlin.

References
BITV 2.0 (2011): Verordnung zur Schaffung barrierefreier Informationstechnik nach dem Behinderten- gleichstellungsgesetz. (Barrierefreie-Informationstechnik-Verordnung – BITV 2.0). Accessed 7 October 2022. https://www.gesetze-im-internet.de/bitv_2_0/BJNR184300011.html
Bredel, Ursula, and Christiane Maaß. 2016. Leichte Sprache. Theoretische Grundlagen. Orientierung für die Praxis. Berlin: Dudenverlag.
Gutermuth, Silke. 2020. Leichte Sprache für alle? Eine zielgruppenorientierte Rezeptionsstudie zu Leichter und Einfacher Sprache. Berlin: Frank & Timme.
Inclusion Europe (2009): Informationen für alle! Europäische Regeln, wie man Informationen leicht les- bar und leicht verständlich macht. Ed. Inclusion Europe. Accessed 7 October 2022. https://www.lag- abt-niedersachsen.de/uploads/migrate/Download/Infofralle.pdf
Maaß, Christiane (2020): Easy Language – Plain Language – Easy Language Plus. Balancing Comprehensibility and Acceptability. Berlin: Frank & Timme.
Netzwerk Leichte Sprache. 2013. Die Regeln für Leichte Sprache. Accessed 7 October 2022. https://www.leichte-sprache.org/wp-content/uploads/2017/11/Regeln_Leichte_Sprache.pdf
Wellmann, Katharina (2020). Medio∙punkt oder Binde-Strich? Eine Eyetracking-Studie. In: Gros, Anne- Kathrin / Gutermuth, Silke / Oster, Katharina (eds.): Leichte Sprache – Empirische und multimodale Perspektiven. Berlin: Frank & Timme, p. 23 – 42.