An illustration of the target detection task, for both Experiment 1 and 2.

Two sets of sample vocabulary, targets, and syllable sequences are shown. The arrows indicate where the targets are in the syllable sequence.

Reaction time (RT) data with syllable position (first, second, or third syllable in the word) on the x-axis, and word presentation (first, second, third, or fourth occurrence of the word in the stream) as different lines in the Figure.

The left panels show the raw data means, and the right panels show the regression model fit. The top panels showed the data from the conceptual-replication condition, and the bottom panel showed the data from the exact-replication condition. The style of the plot is similar to the one in Batterink (2017) for ease of comparison. Error bars represent ±1 SEM.

Reaction time (RT) data with syllable position (first, second, or third syllable in the word) on the x-axis, and word presentation (first, second, third, or fourth occurrence of the word in the stream) as different lines in the Figure.

The left panels show the raw data means, and the right panels show the regression model fit. The top panels showed the data for the disyllabic words, and the bottom panel showed the data for the trisyllabic words. Error bars represent ±1 SEM.

Regression model fit from the three-way interaction between condition (mixed/uniform, categorical), word presentation (1-4, categorical), and position (1-3 for trisyllabic words and 1-2 for disyllabic words, continuous).

Figure 4A showed results from the uniform condition from Experiment 1, and Figure 4B showed results from the mixed condition from Experiment 2. Error bars represent ±1 SEM.

Means for words and part-words, and difference scores for each subject in Experiment 1, collapsed over counterbalancing conditions.

In Figures 5A and 5C, each circle represents the mean rating of a subject for all words and part-words in the uniform and mixed word length conditions, with a solid line indicating the mean value for each item type. In Figure 5B and 5D, each circle represents the difference between mean ratings (words - part-words) for each subject in the uniform and mixed word length conditions, with the solid line showing the mean difference, and shadows showing 95% confidence intervals around the mean. The dotted line at 0 represents chance.

Regression model fit from the three-way interaction between condition (mixed/uniform, categorical), word presentation (1-4, categorical), and position (1-3 for trisyllabic words and 1-2 for disyllabic words, continuous) for the simulated data.

Figure 6A showed results from the simulation for the uniform condition, and Figure 6B showed results from the simulation for the mixed condition.

The set-up for simulation one, where four trisyllabic words occur 10 times each.

The results for Simulation 2, where the thin lines show the weight changes of words and part-words over the course of learning in individual runs, the thick lines show the mean weight of words and part-words across runs over the course of learning.

On the x-axis, each number represents the model having processed 1/10 of the sequence (i.e., 3 means the model has processed 3/10 of the sequence so far). The y-axis represents weights, which are in arbitrary units (as described in PARSER documentation).

The set-up for the mixed condition in Simulation 2.

The results for Simulation 3, where Figure 10A shows the weight changes of words and part-words in the uniform condition, and Figure 10B shows the weight changes of words and part-words in the mixed condition.

On the x-axis, each number represents the model having processed 1/10 of the sequence (i.e., 3 means the model has processed 3/10 of the sequence so far). The y-axis represents weights, which are in arbitrary units (as described in PARSER documentation).