Materials and methods

7.2.1.1 Participants

A total of 30 participants (17 female, 19-51 years old) were recruited via email from the University of Tübingen. All participants were healthy, right-handed, had at least college-entry level education, and did not report any issues with hearing or vision at the time of the experiment. Basic participant

175

information is summarized in Table 7-1. L1 and L2 participants did not significantly differ in age (t= -0.593, p= 0.558). L2 participants were non-native speakers of English (German L1) with a high proficiency in English. Their LexTale scores (Lemhöfer & Broersma, 2012) averaged 77.71 out of 100 (SD= 13.04), and their average self-rated proficiency was 5.79 (SD= 0.75) on a scale from 1-7. 13 native speakers of (Predominantly American) English all grew up monolingually and had at least low proficiency in German (level A2, see e.g., “Common European Framework of Reference for Languages,” 2001). The study was approved by the Ethical Committee of the University Clinic of Tübingen, Germany, and informed consent was given following an experimental description prior to taking part in the study.

Table 7-1. Participant information by language group

Participant Group n n female Age Student participants SPQ Score LexTale Score Ave. Self-rated Proficiency (1-7) L1 (native) participants 13 6 26.5 10 50.2 93.5 7.0 L2 (non-native) participants 17 11 28.0 13 41.5 77.7 5.5 7.2.1.2 Stimuli

Forty idioms were selected from the German-English Database of Idiom Norms (Chapter 2), which is a publicly available database of 300 idioms with norming values (e.g., familiarity with the form and meaning, predictability, and literality). Idioms used in the present study were selected based on their high familiarity to L1 and (German) L2 English speakers (average rating of 6.09 and 5.14 from L1 and L2 speakers, respectively on a scale from 1-7). Idioms also had the same syntactic form (verb + article + noun), and the final word was a concrete noun. Idioms with this form were chosen since they would be comparable in properties such as predictability and literality that have been shown to impact the timing of access to figurative and literal meaning during idiom processing (e.g., Titone & Connine, 1994b). Each idiom was embedded at the end of short, neutral sentences in its original idiomatic form (e.g., I told him to get a grip.) as well as with the final word replaced by another concrete noun to create a comparable literal version of the sentence (e.g., I told him to get a drink.) The idiom-final words and their literal variations neither differed in length (mean letter length idiom: 4.97, literal: 4.79,

p= 0.574) nor in frequency (mean cob/ml idiom: 124.36, literal: 124.39, p= 0.499, Max Planck Institute

for Psycholinguistics, 2001). In addition to these 80 sentences, 20 non-idiomatic filler sentences in German were included. All sentences were divided into two parts for presentation: the beginning of the sentence and the final verb phrase containing the idiom or the literal version of the idiom (see Table 7-2). Additionally, 20 sets of non-linguistic symbols were included as controls. A complete list of stimuli is available in APPENDIX F.

176

Table 7-2.Example experimental items

Item Figurativeness First Half Second Half

1a Idiom I think that John spilled the beans. 1b Literal I think that John spilled the milk.

2a Idiom I told him to get a grip.

2b Literal I told him to get a drink.

7.2.1.3 Procedure

During the fMRI scanning procedure, participants lay supine in the scanner with their heads secured by foam to minimize movement. Stimuli were presented visually on a translucent screen viewed by the subjects via a mirror. Sentences were displayed 1500ms apart in two parts, each for 1500ms. Participants were instructed to read sentences silently for comprehension, and to press a button with their right hand on the second half of sentences which were presented in German. This procedure was exercised in a training session prior to the experiment using sentences not included in the experiment. In order to ensure reading comprehension, participants were reminded that they would be asked questions about the English sentences at the end of the experiment. Items were divided into two balanced experimental blocks so that each sentence occurred in its idiomatic version in one block and its literal version in the other, and the order of the two blocks was pseudorandomized between participants. Stimuli in each block consisted of the 40 target sentences (20 idiomatic sentences and 20 literal sentences), 10 German sentences, matched on length and grammatical complexity, and 10 low- level visual baseline “sentences” (i.e., literal sentences presented in ‘‘SPSS marker set” characters/letter strings). Stimulus sequence was unpredictable and optimised using optseq software.

In addition to the fMRI task, all participants completed the LexTale vocabulary test prior to entering the scanner. Following the fMRI scanning procedure, they also completed a comprehension check, a language background questionnaire, and the schizotypal personality questionnaire (Raine, 1991). In addition to self-reported measures of proficiency, the LexTale vocabulary test (Lemhöfer & Broersma, 2012) serve as an objective measure of participants’ vocabulary and general English proficiency. The comprehension check was used as a measure to be sure that participants were reading the items carefully, as they were instructed that they would be asked questions about the English read during the fMRI task.

7.2.1.4 Imaging methods

Imaging was performed on a 3-T Scanner (Siemens, TIM TRIO). Functional images were acquired with an echoplanar image sequence which is sensitive to blood-oxygen-level dependent contrasts

177

(BOLD responses; TR 1500 ms, 72 slices). Two runs consisting of 205 volumes each were acquired during the experiment.

7.2.2 Analysis

For image processing and all statistical analysis, SPM12 (The Wellcome Centre for Human Neuroimaging) was used. The functional images of each subject were slice time-corrected to the middle slice and were corrected for motion and realigned by using the first scan of the block as a reference. Three participants (1 L1, 2 L2) were excluded from further analysis due to movement artifacts. T1 anatomical images were co-registered to the mean of the functional scans and spatial normalized to the MNI space by the combined segmentation bias correction and spatial normalization tool in SPM12. The calculated nonlinear transformation was applied to all functional images. Finally, the functional images were smoothed with an 8-mm full-width, half-maximum (FWHM) Gaussian filter.

7.2.2.1 Contrasts between conditions

A general linear model (GLM) was constructed for each participant to analyze the hemodynamic response function. In each GLM, regressors were generated by convolving the hemodynamic function with a box car function. Separate regressors were used to model the hemodynamic responses during presentation of the sentence-beginning, idiomatic sentence-ending, literal sentence-ending, German sentences, and the visual baseline condition. Moreover, a high-pass filter (1/128 Hz) was applied to remove low-frequency drifts. For each subject, T-Test contrasts for the idiomatic sentence-ending (idiom) and literal sentence ending (literal) conditions versus low level baseline were calculated. To directly compare brain activation between these conditions, subtraction analysis of idiom versus literal and vice versa were calculated. Random effects analyses on group level were calculated for each of these contrasts. Then, between-group comparisons were calculated for each of these contrasts using two sample T-Tests Activation for all the differential contrasts is reported if it exceeds a significance of p < 0.001 (uncorrected) and an extent threshold of 10 voxels.

7.2.2.2 Language proficiency correlations

To investigate the influence of language proficiency on brain activation, a simple regression analysis of SPM data was used. In this type of analysis, each single voxel in the brain is individually examined with respect to whether the size of the BOLD response is correlated with a variable over subjects. The result of this analysis is a brain map, which depicts the voxels with which there is a significant correlation between the variables. Because of the exploratory character of this study, we chose a liberal threshold of p < 0.001 (uncorrected) and an extent threshold of 10 voxels for analysis.

Separate tests were performed to detect positive correlation (i.e., the higher the score of the individual subject in the questionnaire, the stronger the BOLD response) and negative correlation (i.e., the higher the score of the individual subject in the LEXTALE total score, the weaker the BOLD response)

178

associated with language proficiency. Moreover, separate tests were performed for all participants as well as for the subgroups of native speakers (L1) and non-native speakers (L2).

7.3 R

Main effects for reading idiomatic (idiom), literal (literal) and German sentences against low level baseline (baseline) showed robust activations in a predominantly left lateralized network including visual cortices, temporal lobe and prefrontal cortex. The results concerning each question (concerning only the conditions idiom, literal, and baseline) are displayed below and placed into context in the discussion section.