*2.3. Coding*

The video recordings from each session were tagged and coded using ELAN version 6.2 (ELAN 2021). Spanish data were tagged by a fluent Spanish speaker trained by the third author; LSN data were tagged by the first and third author, both fluent signers with several years' experience coding Nicaraguan signing. Each target question was transcribed, and question words were tagged and categorized on a separate tier. Targeted question words are listed in Table 1, though all question words were tagged and coded, even if they differed from the target. Sample LSN question words are shown in Figure 1. Because we know little about the nature of the syntactic structure of questions in NSL, and even less about how this structure has changed over time, we tagged all interrogatives whether or not they included a wh-question word. Some participants asked a question more than once in response to a single prompt. Each elicited question was tagged. Once the questions were tagged, other trained coders, blind to cohort status, coded each of 6 non-manual facial expressions and body movements that had been observed with questions. Only non-manuals that overlapped with the question utterance were tagged and coded. In some cases, non-manuals began before and/or ended after the manual utterance. In rare cases, the non-manual was held through the interlocutor's response to the question. In these cases, we computed the non-manual offset as the onset of the interlocutor's response. Each of these non-manuals was coded on a dedicated tier in ELAN, indicating its onset and offset for every item in which it appeared. This enabled us to extract, for each non-manual, the number of times it was used, its duration, the question in which it occurred, and its temporal overlap with a targeted question word or sign. Examples of the six non-manuals can be seen in Table 2.

WHAT HOW-MANY

WHEN

**Figure 1.** Examples of the LSN wh-question words that were elicited.

**Table 2.** Descriptions and examples of the six non-manuals coded.



