SAUCE

Steps used to develop the SAUCE set. For future reference, Steps 7 and 8 are quite time-consuming.

Step 1. Obtained picture naming data from preschool-aged monolingual English-speaking children. Pictures are available by request from Creel.

Step 2. Created candidate sensical sentences (like The monkey ate the banana) for which the picture names (like banana) were the intended targets.

Step 3. Presented sentence stems (The monkey ate the ___________) to English-speaking adults to obtain sentence completion norms.

Step 4. Selected the 139 more-predictable sentences from the adult set, and presented to English-speaking children.

Step 5. Chose the top 100 sentences (without target-word repeats) for the final set. Average child cloze probability was .57 (SD = .23; range: .19-.96).

Step 6. Created 100 nonsensical sentences by rearranging the target words (e.g. The monkey ate the shark). From these, 100 nonword sentences were created by reassigning word onsets across sentences for the words in the carrier phrases.

Step 7. Recorded 20 speakers reading all sentences with related sentence triplets (those with the same stems) recorded in sequence. Order within the triplet was random (sensical, nonsensical, nonword) so that any order effects were equivalent across the three sentence types. Additionally, speakers were advised not to use contrastive intonation (i.e., not to emphasize the changing final word) and to read naturally as though speaking to a small child. There were 10 speakers of California-accented English (5 female, 5 male), and 10 speakers of Mexican-accented English (5 female, 5 male). Sentences where speakers made errors were rerecorded later in the session.

Step 8. Created individual audio files of each sentence for each speaker, as well as a final-word-only audio file of each sentence. Files with speech errors were discarded and all files were normalized to 70 dB SPL.