*2.1. Participants*

The participants consisted of 19 deaf, predominantly monolingual Cena signers aged 13–59<sup>16</sup> who all live in or around Várzea Queimada, and 19 deaf adult native Libras signers based in Rio de Janeiro. Cena signers generally do not have fluency in written Portuguese, and many are not fully literate. The Libras participants in our study all have a strong grasp of written Portuguese as it forms a part of their daily lives.

#### *2.2. Materials and Task*

The current study used the Haifa Clips stimuli set, designed by Sandler et al. (2005), to elicit recounts of short events from participants. The task consisted of 30 short (1–3 s) video clips depicting a variety of intransitive, transitive, and ditransitive actions, such as a ball rolling, a woman looking at a man, and a man throwing a ball to a girl. Participants were asked to relay each event to an interlocutor, another deaf signer and native user of the language in question. Participants then later functioned as interlocutors for following participants. Although many hearing individuals can sign to varying degrees of competency within the community where Cena is used, we chose to limit participants to only deaf monolingual signers to avoid any potential effects of linguistic accommodation between deaf and non-native or non-fluent hearing signers. This also ensures the utterances most closely mirror language used in its natural form—in a communicative context with comprehension as a desired target. To maintain consistency across the two groups, all Libras signers in the study were also deaf native signers.

Once a participant had relayed the event to the interlocutor, the interlocutor chose the corresponding event from three options depicted in images. All response options were still images—no part of the form relied on written language. An example page of the interlocutor's task is shown in Figure 9. Usually, the options differed in one argumen<sup>t</sup> of the verb and/or the verb itself. For example, for the stimulus with a rolling ball, the events depicted in the three choices are a ball falling, a bottle rolling, or a ball rolling. Once the interlocutor made a choice, this response was recorded as the first attempt. If correct, the researchers showed the following clip. If the interlocutor did not understand or chose incorrectly, the participant was prompted to explain the clip again. Participants were allowed as many attempts as needed to relay the event successfully. If none of these were successful, the attempt was marked incorrect and we played the next clip. The data for both languages was glossed by two fluent hearing Libras signers who have exposure to and knowledge of Cena through fieldwork. In light of the ongoing COVID-19 pandemic and lack of internet connection in Várzea Queimada, it was unfeasible to have any native Cena signer or our bimodal bilingual consultant in the community involved in the glossing process.

**Figure 9.** Multiple choice options for the stimulus 'woman writes on refrigerator'.
