2.5 Experiments
2.5.3 Experiment 1: Processing of compounds
2.5.3.3 Stimuli and procedure
A target noun or verb was paired with primes of different types. The relationship between primes and targets varies across conditions. Experiment 1.1 consisted of the morphologically- and-semantically related condition (MS), the morphologically-related condition (M), and the semantically-related condition (S). Experiment 1.2 comprised the MS condition, the M condition, the phonologically-related condition (Ph), where the first element of the prime
rhymes, i.e., sharing vowels, codas, and tones5, with the target, and the control condition (C), where primes and targets were not related in any way. Example test items are provided in Table 2.6. The differences between Experiment 1.1 and 1.2 are summarized in Table 2.7. The primes are compound words, with the exception of eleven monomorphemic nominal primes in Condition S. The prime-target pairs were distributed across lists, according to the Latin Square design, such that each target appeared only once in each list. Each list contained the same number of stimuli from every priming condition, which were distributed across eight blocks. Each participant participated in only one experimental list.
Table 2.6: Example test items.
Target MS M S Ph C
lˆaw lˆaw khˇa:n lˆaw ri:an b`O:k kl`a:w khˆaw caj d `W:m d`am
‘tell’ ‘tell (a story)’ < ‘tell’ + ‘cry out’
‘study’ < ‘tell’ + ‘study’ ‘notify’ < ‘tell’ + ‘say’ ‘understand’ < ‘enter’ + ‘heart’ ‘indulge’ < ‘drink’ + NA
kl`an kl`an krO:N kl`an klE:N h `W:at ra-hˇ7:j p`an p`u:an khwˆı:aN th´ıN
‘sift, distill’ ‘screen’ < ‘distill’ + ‘filter’ ‘mistreat’, ‘defame’ < ‘distill’ + ‘tease’ ‘evaporate’< ‘dry up’ + ‘evaporate’
‘be frantic/chaotic’ < ‘shake’ + ‘cause turmoil’
‘throw away’ < ‘throw’ + ‘dump’
phr´aP phr´aP-sˇoN phr´aP-P`e:k n´ak b`ua:t f´ak fE:N l´ın ch´ak
‘monk’ ‘monk’ < ‘monk’ + ‘monk’ ‘leading actor’ < ‘monk’ + ‘one’ ‘priest’ < ‘expert’ + ‘ordain’ ‘gourd’ <
‘gourd’ +‘cucurbita pepo’
‘drawer’ < ‘tongue’ + ‘pull’
n´a:m n´a:m ta: n´a:m ta:n khˇO:N lˇe:w klˆa:m nW:a kho:m faj
‘water’ ‘tear’ < ‘water’ + ‘eye’ ‘sugar’, ‘brown’ < ‘water’ + ‘palm’ ‘liquid’ < ‘thing’ + ‘liquidy’ ‘muscle’ < ‘muscle’ + ‘meat’ ‘lamp’ < ‘lamp’ + ‘fire’
Note: Experiment 1.1 includes Conditions MS, M, and S, while Experiment 1.2 includes Conditions MS, M, Ph, and C. In each condition, the IPA transcription is provided on the first line. The meaning of the entire compound is provided on the second line, while the meaning of each component is on the third line.
Table 2.7: Design differences between Experiment 1.1 and 1.2.
Experiment 1.1 Experiment 1.2
ISI 800-900 ms 400-600 ms
Conditions: MS, M, and S MS, M, Ph, and C
Test items:
144 pairs over 3 lists (48 pairs per list; 16 pairs per list per condition)
- 26 target nouns - 22 target verbs
180 pairs over 4 lists (45 pairs per list; 11-12 pairs per condition)
- 24 target nouns - 21 target verbs Fillers: 30% Test items 40% Non-word fillers 30% Word fillers 20% Test items 50% Non-word fillers 30% Word fillers
The experiment began with 10 practice trials. During the session, a fixation cross
5Five words in Condition Ph do not have a complete rhyme, but they differ in no more than one additional
appeared at the center of the screen. Each auditory stimulus was presented through child- friendly/adult headphones with an interstimulus interval (ISI) of 800-900 ms in Experiment 1.1 and 400-600 ms in Experiment 1.2. The instruction was given both verbally and on the screen. Participants were asked to press the button with a check mark with their dominant hand, when they heard a sound they understood or had heard being used in the language, or press the other button when they heard a sound which was nonsense or had not been heard being used in the language. They were also instructed to do so as accurately and as fast as possible. No feedback was given after each trial. The experiment lasted approximately 25 minutes with three self-administered breaks.
2.5.3.4 Modeling
For the accuracy models, run only on the child data, the data were subset based on the performance group (Group 1 or Group 2). Using the same factors, the accuracy models for each performance group were run separately. Both models contained 11 fixed effects factors, including prime condition (MS, M, or S for Exp1.1 and MS, M, Ph, or C for Exp1.2), word type (noun or verb; average across word types on the intercepts), z-scored duration of target, z-scored trial number, z-scored interstimulus interval (ISI; measured in ms), prime accuracy (average on the intercept), log-transformed RTs to primes, participant group (ASD or TD), z-scored Ravens nonverbal IQ, z-scored age, and gender (average across genders on the intercepts). Moreover, the interactions between prime condition and participant group, along with two random effects factors (random intercepts for participants and items) were also included in the models. For the effect of prime condition and participant group, dummy coding was used to relevel to the desired baseline.
The reaction time models contained 10 fixed effects factors for the child data and 8 factors for the adult data, using the same contrast coding schemes. The factors include prime condition, word type, z-scored duration of target, z-scored trial number, z-scored ISI, log-transformed RTs to primes, participant group (only for the child data), z-scored NVIQ (only for the child data), z-scored age, and gender. Interactional effects were coded between
prime condition and participant group in the child data. Two random effects factors of participants and items were also included in the models.
2.5.3.5 Results: Experiment 1.1 and 1.2