*3.2. Participants*

A total of 23 participants volunteered to join the task (see Table 2), all from three communities studied by the author since 2009: Chicán, Nohkop and Trascorral. All experiments were video recorded with the permission of the participants.

In Chicán, 14 people participated (7 women) between 11 and 61 years old at the time of the interview. The average age was 37.7 years old. Thirteen of the participants from Chicán are deaf and one is a CODA (hearing Children Of Deaf Adults). Two siblings born from deaf parents, aged between 11 and 16 (average age is 13), constitute the second generation. In Nohkop, 6 people participated, five women and one man between 2011 and 2021. The average age of the participants was 23.4 years old. Four of the 6 participants are siblings, all deaf. Of the two hearing signers, one is a Bilingual-Bimodal cousin who grew up with the deaf siblings and played an integral part in the construction of the sign language. The second is a CODA, daughter of the oldest deaf sibling and her hearing husband, also fluent in YMSL. Finally, in Trascorral, only three participants were interviewed since they were the only ones present or willing to engage in the task. Two of the participants are deaf, while the daughter of the oldest sister is a hearing CODA. All CODAs are Bilingual-Bimodal signers of YMSL, and they are also fluent in Yucatec Maya and/or Spanish.


**Table 2.** Participants data (including interactional groups).

The task was conducted with the researcher showing each video on a laptop or tablet. The computer was then closed and the participant was asked to retell the video to the camera. The video stimuli varied for movement, people involved in the actions performed, number of men and women, and the object being passed around. Each video triggered the need to provide information about agents and objects, either to the experimenter, or to another interlocutor who did not watch the clip. Indeed, the majority of signers provided such information, even when the researcher was the only interlocutor.

#### *3.3. Transcription and Coding*

All data were transcribed using the program ELAN (Brugman and Russel 2004). Coding involved transcription and translation of each complete utterance produced by signers, and special attention was paid to the coding of pronouns and the use of space or the body to express verbal inflection. In order to code how participants retold the video stimuli using the space around them, the people in the video were coded as "Persona" using a number for each one: 1 (the one on the right of the screen), 2 (the one in the middle) and 3 (the one on the left of the screen), as shown in Figure 2. Such coding allows us to understand how signers placed each Persona from the video stimulus in the signing space or if/when they used their own body instead. In total, the total possible number of items participants could indicate was 36.

**Figure 2.** Coding of the "Persona" (people on the screen).

## **4. Results**

#### *4.1. Pronoun Marking and the Use of the Signing Space*

As we saw earlier, in order to use space to express verb agreemen<sup>t</sup> in a sign language, one requisite is to be able to place referents as pronouns in the signing space. Such placement is essential for agreement, as the movement of an agreeing verb (such as GIVE or TAKE) will distinguish the agen<sup>t</sup> from the recipient (Klima and Bellugi 1979; Meier 2002).

In the data, several strategies have been used to mark pronouns in R-loci. I will not dive into too much detail as to which handshapes were used but only mention that signers used entity classifier (see Safar (2019) for a detailed description of YMSL, and Meir and Sandler (2008)) and, in the case of one signer, buoys (see Figure 5a,b and Liddell (2003, p. 223-ff)) but the most common were pointing (see Figure 6b,d,f) and placing (see Figure 6j). The coding used in the analysis contrasts the marking of *R-locus* versus *no marking*, that is, whether the signer did or did not use the space around his/her body to indicate a pronoun.<sup>1</sup> Results are presented in Figure 3.

Results indicate that many signers from Chicán and Nohkop make a productive use of the signing space to indicate pronouns in R-loci: 54% of the responses in Chicán and 76% of the responses in Nohkop. In Chicán, the three second generation participants (two deaf children born from deaf parents and a CODA) all systematically used R-loci (i.e., 36 times in the task). In Nohkop, both first and second generation signers predominantly used R-loci for pronoun reference. However, none of the signers form Trascorral indicated pronouns in R-loci.

These results predict that signers from Chicán, especially second generation ones, and most of the signers from Nohkop can potentially use agreeing verbs, while none should be expected to do so in Trascorral.

**Figure 3.** Total number of responses in percentages compared by communities for R-locus vs. no-marking.

#### *4.2. The Use of Space and the Verb Classes*

In many sign languages, verb classes have been described based on their lexical, morphological and sometimes semantic features, as in ASL for instance (Padden 1983). However, in emerging sign languages, established categories are not always applicable as signers may use different strategies and, more often than not, in-between verb classes. This is the reason why some authors have come up with ad-hoc categorizations, such as Padden et al. (2009) for ABSL.

In the task proposed here, participants had to retell, from video clips, actions involving an object being transferred from one Persona to another. If participants marked agreement, the verb should be performed between previously established R-loci (see above) and directionality was relevant. However, not all responses were uniform and results are better analyzed considering two types of verbs: uninflected and inflected verbs. In the latter case, two options are possible, either with a single argumen<sup>t</sup> or with two. Thus, three cases are observed: (a) uninflected verbs, which were only performing with an in/out center movement from the body and do not agree with the pronouns given that directionality is not relevant to indicate semantic roles; (b) single-agreement verbs, which only explicitly mark the patient or direct object argumen<sup>t</sup> while the participant enacts the default agent/subject, and; (c) double-agreement verbs, where the verb's movement explicitly indicates the relation between agent/subject and patient/direct object in the signing space (i.e., between two R-loci). A schematic representation is provided in Figure 4. The terminology is explained in the text below and in Section 5.1.

#### 4.2.1. Uninflected Verbs

Padden et al. (2009) who conducted a similar task in ABSL considered that these verbs (that the authors refer to in their paper as "plain verbs") "lack the fine locational distinctions seen in verb forms used to mark motion and location and they lack person marking as well" (Padden et al. 2009, p. 388). As pointed out by Montemurro et al. "agreement is with the subject and/or object of the verb and is dependent on the previous establishment of

the referent or referents" (Montemurro et al. 2019, p. 421). In the case of uninflected verbs (see examples in Figure 5) the path movement of the verb is not determined by any R-loci previously established in the signing space and consists simply of an in/out movement that does not indicate any specific argumen<sup>t</sup> in space. In fact, most of the participants who used uninflected verbs did not mark pronouns in R-loci prior to producing the verb and only mentioned the subject or the object lexically (Figure 5d,g, respectively).

**Figure 4.** Representation of the three verb types used by participants in the task. The diagrams represent the signers and the signing area as viewed from above. The circles with P1, P2 and P3 indicate where the signers created R-Loci (RL) and the arrows represent the movement of the sign performed in the signing space. This representation is based on the action presented in a video clip on the screen (represented by a black line in the bottom of the image).

**Figure 5.** Examples of uninflected verbs ((**<sup>a</sup>**–**f**) signers of 1st generation from Chicán; (**f**,**g**) signer of 1st generation from Trascorral) (**<sup>a</sup>**–**<sup>c</sup>**) "there is a woman, she is on this end, she takes"; (**d**,**<sup>e</sup>**) "man gives"; (**f**,**g**) "someone gives a book". (Note that here TAKE is a 'backwards verb', moving from the direct object to the subject, see Meir (1998) for a discussion).

#### 4.2.2. Single Agreement Verbs

Single agreemen<sup>t</sup> verbs involve a movement of the verb that depicts the displacement of an entity in space, and one end of the path is always associated with a single argumen<sup>t</sup> of the verb (labeled source or goal in Meir (Meir et al. 2007; Meir 2002).<sup>2</sup> In the case of GIVE and TAKE, the path always starts or ends at the body of the signer and the directionality of the verb reflects the path of the object being transferred. Crucially, and in contrast with uninflected verbs, in the case of single agreemen<sup>t</sup> verbs, the movement is usually performed in accordance with where a pronoun was placed in space in an R-Locus, i.e., either from center to the left or right or from left to right. Signers who used single agreemen<sup>t</sup> verbs always embodied the Persona from the video clips, that is, they used constructed action (also called role-shift or character perspective) to enact the action they are retelling. In constructed action, the signer takes the perspective or viewpoint of the giver or the recipient. This is the strategy that Meir et al. (2007) consider as "body as subject", in which the body is the subject agreemen<sup>t</sup> marker but is omitted while the marked argumen<sup>t</sup> is the object.

Examples of this strategy are presented in Figure 6. While the verb used by the signer is TAKE, a backwards verb, that is, the movement starts from the locus associated with the object and moves toward the locus associated with the subject (which is the goal of the transfer). In single verb agreement, the subject is only implied in the construction, making it distinct from double agreemen<sup>t</sup> verb constructions where both arguments are explicit and located in the signing space. In Figure 6a–i, the signer embodies the woman (P2) from the video stimuli. The action is that she takes a book from P3, the man to her right. When inflecting the verb, the signer acts as if he was the woman from the video clip, and takes the book from his right. The signer from Figure 6j,k first places the Persona in the signing space in R-loci, and then uses constructed action, enacting Persona (P2) taking an object from the Persona (P3) to her left. The perspective taking is particularly clear in Figure 6k, where she exaggerates on purpose and mimics the Persona from the video (based on her own interpretation).

**Figure 6.** Responses using single agreemen<sup>t</sup> verbs ((**<sup>a</sup>**–**i**) signer of 1st generation from Chicán; (**j**,**k**) signer of 1st generation from Nohkop) (**<sup>a</sup>**–**i**) "there is a man here, a man here (and) a woman where I am, from P3 I take the book"; (**j**,**k**) "P2 and P3 are thusly placed, from P3 I take the book".

#### 4.2.3. Double Agreement Verbs

Double agreemen<sup>t</sup> verbs agree with two arguments, their source argumen<sup>t</sup> and their goal argument. In sign languages, this process usually implies the creation of two R-loci in the signing space and agreemen<sup>t</sup> is performed with a movement of the verb from one R-

locus to the other (i.e., between arguments). Although double agreemen<sup>t</sup> verbs are common in most Deaf Community sign languages that are adopted by schools and interpreters, they are noticeably absent in emerging sign languages, as pointed out for instance for ABSL (Padden et al. 2009) or Kata Kolok (de Vos 2012). In the data presented here, verb agreemen<sup>t</sup> constructions with two arguments independent of the signer's body were used by the YMSL participants from first and second generations.

Figure 7 presents an example of such a construction performed by a second generation signer from Chicán. The signer first creates R-loci that arbitrarily assigns the NPs ("the man" and "the woman") to empty locations in the signing space (glossed P1, P2 and P3). She then mentions the object being transferred, in this case a book (BOOK-SASS). Finally, she moves the verb in the space between two R-loci previously assigned to P2, the woman and P3, the man, semantically assigning the role of giver to P2 and of receiver to P3. In contrast with single argumen<sup>t</sup> constructions, her body is not involved and the movement of the verb marks the agreemen<sup>t</sup> with the two arguments located in the signing space.

**Figure 7.** Second generation signer from Chicán using double agreemen<sup>t</sup> "There is a man here, a woman here, (and) a man here, the woman (P2) gives the book to the man (P3)".

#### 4.2.4. Comparison between Communities

Results of the comparison between the types of verbs used in signers from different communities are presented in percentages in Figure 8. Interestingly, we observe that YMSL signers used double agreemen<sup>t</sup> verb constructions much earlier and more extensively than what would be predicted from previous research on emerging sign languages (Padden et al. 2009).

In Trascorral, as predicted by the analysis of the pronouns (or lack thereof in this case), signers relied on an uninflected verb strategy. Their response is comparable to what the majority of ABSL signers did in a similar task (Padden et al. 2009). In the case of Nohkop, all signers used verb agreement. Most of the constructions used (almost 80%) were single agreemen<sup>t</sup> verbs, but a fourth of the responses were double agreemen<sup>t</sup> verbs. A closer analysis reveals that three of the signers from Nohkop used exclusively single agreemen<sup>t</sup> construction, while the other three alternate between single and double agreemen<sup>t</sup> constructions.

**Figure 8.** Results of the strategies used by the participants in the task separated by community (in percentages).

In the case of Chicán, conflated total responses do not show if some signers were more consistent in the verb construction types used than others. For this reason, Table 3 presents the results of Chicán signers considering consistency in individual responses. Three groups emerged based on their preference. Group 1, composed of four signers from first generation, chose to a grea<sup>t</sup> majority (95.8%) uninflected verb constructions. Group 2, composed of six adults from the first generation, chose predominantly singe agreemen<sup>t</sup> verb constructions, but did alternate during the task with other strategies. Finally, group 3 composed of 4 signers, 2 deaf signers form the second generation, a deaf adult from the first generation and his daughter, a hearing Bilingual-Bimodal signer from the second generation, exclusively used the double agreemen<sup>t</sup> strategy.


**Table 3.** Results of individuals from Chicán grouped according to the main strategy used in the task.

Comparison of the results from the three communities presents an unexpected picture. Some YMSL signers only used uninflected verbs, behaving as signers from first generations of other emerging sign languages. This is the case for signers from Trascorral and Group 1 from Chicán. However, many signers also from first generations used single verb agreemen<sup>t</sup> strategies. This is the case for many signers from Nohkop and Group 2 from Chicán. Finally, and even more surprisingly, in Nohkop and Chicán, some signers from first and second generations (although systematically in Chicán in the latter case) were able to employ double agreemen<sup>t</sup> verb constructions, making grammatical use of the signing space.
