**5. Discussion**

Results show that YMSL signers, even from first generations, are able to use single and, crucially, double agreemen<sup>t</sup> verbs, challenging the prediction of the evolutionary path hypothesis. How can we explain such early use of space to express grammatical categories such as verb inflection in YMSL? I will review previous arguments put forward to explain

the evolutionary path towards the emergence of agreeing verbs, and I will propose that, in the case of YMSL, the use of the preferred frame of reference in the surrounding culture is what seems to motivate a specific use of space among signers.

#### *5.1. Limitations of the Studies*

While results indicate clear tendencies, it is important to consider some of the limitations of this study. First, the small number of participants is problematic in order to make overarching generalizations. However, this is not an obstacle that can be overcome as the number of deaf signers is limited in each community and in some cases (such as Trascorral for instance) there was a noted unwillingness of some participants to take part in the experiment.

A second limitation is that this task is not an everyday activity for signers. However, it is noteworthy that clear patterns arose from the signers' responses and that very fact is a confirmation that the task was not completely misunderstood, hence validating the results.

A third limitation is the low number of bilingual-bimodal signers. The inclusion of just a few of them was due to the exploratory character of the task, but a more systematic study should be run in the future to explore this population in more detail.

Finally, the task conducted was not a comprehension task as in Senghas (2003), so the results only show the tendency for production and not comprehension. This will also be the object of a follow-up study.

#### *5.2. An Evolutionary Path*

Studies conducted on emerging sign languages have been particularly helpful to determine an evolutionary path that seems to begin without the existence of verb agreemen<sup>t</sup> expressed in space and how it appears late in the life of the sign language (de Vos and Pfau 2015; Meir et al. 2010). Among the emerging sign languages documented to date, the most relevant studies for our issue at hand were conducted on the following languages: Nicaraguan Sign Language, Al-Sayyid Bedouin Sign Language, Kata Kolok and Inuit sign language. Seminal work on ASL and Auslan as well as Danish and Israeli sign languages describe the use of space for agreement, which motivated later studies in village communities. Table 4 provides some background information that will be used for comparison in the next subsections.

**Table 4.** Characteristics of various sign languages for cross-cultural comparison; based on de Vos (2012, p. 424), (Meir et al. 2007; Sandler et al. 2005; Aronoff et al. 2005), (Senghas 2003, 2005; Coppola 2020), (Schuit 2012, 2014; Schuit et al. 2011), Engberg-Pedersen (1993), de Beuzeville et al. (2009) and Liddell (2003).


The use of space to express grammatic categories, specifically agreement, has been explored in these languages but, more importantly, very similar tasks have been conducted with signers, making a cross-cultural comparison possible. Such an attempt based on published materials is proposed in Table 5.

**Table 5.** Cross-linguistic comparison of verb agreement, based on de Vos (2012, p. 424), (Meir et al. 2007; Sandler et al. 2005; Aronoff et al. 2005), (Senghas 2003; Senghas et al. 2004), (Schuit 2012, 2014; Schuit et al. 2011), Engberg-Pedersen (1993), de Beuzeville et al. (2009) and Liddell (2003).


Table 5 indicates that recently created sign languages in small communities tend not to display agreemen<sup>t</sup> and even less double agreemen<sup>t</sup> marking, as is the case for ABSL (Meir et al. 2007; Padden et al. 2009, 2010), Kata Kolok (de Vos 2012, p. 129), the first two cohorts of NSL (Senghas 2003; Senghas et al. 2004) and the first two cohorts of ISL (Padden et al. 2010). Interestingly, the development of Danish sign language also follows a similar pattern (Engberg-Pedersen 1993). On the other end of the spectrum, later generations of signers and older sign languages, such as ASL or Auslan, for instance, make use of all three verb types. However, de Beuzeville et al. (2009) remind us that the use of double agreemen<sup>t</sup> verbs, although available, remains quite rare in everyday use (around 20% of all verb occurrences; see also Bauer (2014) for similar findings in Yolngu sign language in Australia).

A review of the literature indicates that the evolutionary path seems to be a robust tendency cross-linguistically. YMSL already has signers from the first generation who are able to use all three verb types, albeit not systematically, in the task. At first glance, YMSL seems to be an exception, but Schuit (2014) documented the use of verb agreemen<sup>t</sup> in Inuit sign language (IUR) also from first generation signers. What factors might contribute or restrain the early use of the signing space for agreemen<sup>t</sup> in emerging sign languages? Several explanations have been put forward: the size of the community, the age of the sign language, the early exposure to language.

#### 5.2.1. Size of the Community and the Critical Mass of Deaf Signers

The appearance of agreeing verbs has been suggested to be connected with the size of the community of signers, especially the deaf individuals: the more (deaf) signers a community has, the more complex and refined linguistic messages ought to be. Senghas, in her review of the cases of ABSL and NSL, considers that "a language would develop more slowly within a small community" (Senghas et al. 2005, p. 464).

Table 4 shows that languages with up to 125 deaf signers still might not have developed a double agreemen<sup>t</sup> verb type. Israeli, Danish or American sign languages with signers in the thousands are more likely candidates. However, Chicán has 16 deaf signers (the maximum number of signers was 19) and Nohkop only 4 and yet, both communities display some use of double agreemen<sup>t</sup> verbs. Size of the community and the number of deaf signers alone cannot explain the emergence of agreemen<sup>t</sup> in space, at least not in the case of YMSL.

#### 5.2.2. The Age of the Sign Language and the Number of Generations

When considering the age of the language, the number of years of existence of the language per se is not what counts but instead the number of generations. Senghas et al. (2005) point out that in a stage of NSL where signers already have regular contact with each other, perspective-taking (single agreemen<sup>t</sup> construction) tends to become more differentiated and is extended to the patient in addition to the agen<sup>t</sup> role. The authors conclude that the kind of input deaf children receive has an influence on the complexity of expressing grammatical relations and marking perspective. One reason proposed is that children or new cohorts of signers build on the linguistic structures constructed by the previous generations. The authors are careful to point out that "no single age cohort can progress through the developmental stages in the order necessary to create a language in a single pass. Consequently, language genesis requires at least two cohorts of the community in sequence, the first providing the shared symbolic environment that the second can exploit" (Senghas et al. 2005, p. 304).

This proposal has more explanatory power and seems to support the data from YMSL. Although signers of first generations do use single and sometimes double agreemen<sup>t</sup> constructions, it is notable that Chicán deaf signers from the second generation used double agreemen<sup>t</sup> constructions systematically in the task. Still, the generational hypothesis fails to explain why YMSL signers were able to develop double agreemen<sup>t</sup> verbs since the first generation. It appears that just having successive generations of signers might not be enough either, as demonstrated by ABSL with three generations or Kata Kolok with six.

#### 5.2.3. Early Exposure to the Language

Early exposure implies that certain periods of language exposure are crucial in order to develop a more complex grammatical system. Various studies have shown that native learners display superior performance to early and late learners in their knowledge and use of agreemen<sup>t</sup> (Emmorey and Reilly 1995; Newport 1990). However, Meier suggests that some linguistic structure might be harder to acquire: for instance, Turkish children master verb agreemen<sup>t</sup> around the age of 2.0 while ASL deaf children only around 3.0 (Meier 2002, p. 126).

Early exposure definitely contributes to the successful development of new sign languages and the refinement of their grammatical structures, but cannot be the main explanatory criteria in the case of YMSL as signers did not learn from previous sign languages. However, exposure to gestural habits from the speakers of the surrounding community might have an effect on how to use space, as I will argue below.

#### *5.3. The Impact of the Geocentric Frame of Reference and the Semiotic Jump towards an Abstract Use of Space*

I will argue that the preference for the geocentric frame of reference, especially through its use in gestures among speakers of Yucatec Maya in the surrounding culture, facilitates a cognitive and semiotic jump for signers towards a more abstract use of the signing space under certain conditions. Such analysis also put emphasis on the idea that some linguistic habits of the surrounding community can be a seed for the emergence of the sign language (e.g., Kocab et al. 2022), namely the gestural habits of the speakers of Yucatec Maya.

#### 5.3.1. Looking at Signing Space in Terms of Frames of Reference (FoR)

A study conducted by Senghas (2003) with various cohorts of Nicaraguan Sign Language signers compared the production and comprehension data of spatially modulated verbs of first and second cohort signers. The author concentrated her analysis on whether signers used a rotated or an unrotated representation of events (see also Gagne 2017). That

is, if presented with a video stimulus where a woman is giving a cup to a man to her right, will the signer produce the movement of the verb to his/her own right or to the left side of his/her body? A movement to the left would correspond to an unrotated representation and a movement to the right to a rotated one (Senghas 2003, p. 518). Senghas finds that while first cohort signers accept both a rotated or an unrotated layout for expressing spatially modulated verbs, second cohort signers all consistently apply a rotated representation which also limited their acceptance of the event depictions accordingly. Such innovation restrains the way signs can be produced and, in doing so, also limits what the signs can mean, and hence makes grammatical references more specific.

However, a different reading is possible taking into account spatial Frames of Reference (FoRs). In this interpretation, the rotated perspective would correspond to the egocentric FoR while the unrotated perspective to the geocentric. A schematic representation is proposed in Figure 9.

**Figure 9.** A new reading of Senghas (2003, p. 518)'s proposal in terms of spatial Frames of Reference. 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).

In the case of the rotated perspective strategy, the point of reference to locate entities is the signer's point of view in considering the stimulus in video. Basically, this means that the signer takes the perspective of one of the characters in this video and reproduces the movement from his/her perspective. Such a strategy implies a mental rotation from the stimulus and means that if the person in the video clip gives an object to the person on her right, the signer will produce a sign oriented to his/her own right. This use of space relies on the use of an egocentric Frame of Reference. For the unrotated perspective, one explanation is that the signer reads the same spatial arrangemen<sup>t</sup> but based on a geocentric frame of reference, meaning that the movement is not taken as being to the "right to the person in the stimulus" but, say, "north of the person" or "towards the door (in the real world)". Although it might be highly counterintuitive for speakers of languages that heavily rely on egocentric FoRs (such as English or Spanish), this is actually the most common way to use space in locating objects in space in rural and non-Western communities around the world (de Vos 2012; Bauer 2014; Levinson 2003).

#### 5.3.2. The Condition of Truth in the Use of Space

Schegloff (1984) notes that sometimes, in daily conversation, U.S. English speakers from the West Coast point to places in the real world but indicate these entities in a somewhat random manner, since the orientation of their pointing does not correspond to the actual location of the places they are mentioning. McNeill (2003) looking at the same issue among English speakers in Chicago, considers that what is at stake is the morality of pointing and how much intersubjectivity is involved as well as the speakers' willingness to negotiate meaning. If English speakers from urban settings are able to point randomly to real places in the world without consequences, this is not the case in other cultures. Among speakers of Yucatec Maya (but also in many other settings, see Bauer (2014) de Vos (2012) or Haviland (1993), inter alia), one is entitled to point towards real entities and places in retaining their actual orientation (Le Guen 2011a). Doing otherwise is considered lying, hence, the significance of the "condition of truth".

Among geocentric coders (i.e., language users that prefer to use geocentric FoRs in locating objects in space), the cultural tendency to maintain real orientation is prevalent and has even been described in spontaneous narratives by Haviland (1993) among Tsotsil speakers from Mexico and in stories from deaf signers of Kata Kolok by De Vos (de Vos 2012, p. 268-ff). In these cases, it is noteworthy that direction and orientation are important and foregrounded. For instance, in the story of a shipwreck reported by Haviland, the orientation towards the sea is a crucial element and, in a narrative reported in De Vos' work, the path taken inside a tunnel cannot be modified as it implies going one way and not the other. In both cases, when the speaker changes location and orientation, (s)he will always locate the elements of the story according to their actual orientation and not based on his/her point of view at the place and time of the narration.

While culturally prevalent, speakers and signers do not always stick to these principles, specifically when they judge that the spatial location or orientation of the entities is not that crucial (or that it can be backgrounded) and the story can be told in a more abstract spatial setting, usually relying on an intrinsic frame of refence (in which both cardinal directions and the signer's point of view are irrelevant). In such narrative contexts, both entities are located with respect to one another and not according to their real-world orientation (a fact also acknowledged in De Vos' data). Such differentiation in the use of space is crucial and understanding how truth conditions can be cancelled by the signers provides one key to explain why they are able to make a semiotic jump towards placing entities in the signing space in an abstract manner.

In the task used in his study, the setting, the people involved and the actions performed are not familiar, habitual and, crucially, not localizable in real space. As a result, it is not surprising to see signers treating it as the kind of narrative that can be freed from the spatial condition of truth. I argue that this condition is one significant argumen<sup>t</sup> to understand why the use of the signing space in an abstract way is possible and can give rise to an agreeing verb class. However, this criterion is not the only thing that can trigger an abstract use of space, since signers have built upon the use of allocentric localizations habitually used by Yucatec Maya speakers when talking and gesturing about spatial arrangements.

#### 5.3.3. A Cultural Habit of Allocentric Localization

Most readers might be familiar with locating two entities in the real world using their projected point of view on a scene, locating them to the left and right (e.g., "Leaving the faculty building, I parked my car at the right end of the parking lot"). Such a conception of space relies on an egocentric FoR from which relations between external entities are made based on the speaker/signer's transposed point of view. However, in many cultures, speakers or signers prefer to use a geocentric FoR and explain that "the well is to the south-side of the palm tree". In more extreme cases, such as among the Yucatec Mayas, an indication can be reduced to a verbal cue, such as a manner deictic, and it is a gesture oriented according to the real-world orientation that provides the relevant spatial information, as in the following statement *le áasulero máa bey yanika'?* (accompanied by a waving gesture towards the north) "isn't the Azulero (a shop) located like this (=on the north side)?".<sup>3</sup>

Yucatec Maya speakers have a habit of using an allocentric perspective when locating two entities with respect to one another. To do so, they place both entities in the gesturing/signing space but, instead of using their point of view, they rely on external anchors, such as cardinal directions or salient elements in the world. In such types of locative

constructions, the body is not considered, not even the projected point of view, so the use of the left and right becomes irrelevant. What is central however, is the positioning of the entities according to their real-world orientation. Consequently, speakers make use of the symbolic space all around their body, locating the entities either in front of them, on each side of the body or even behind them (see Le Guen (2011b) for examples).

In order to illustrate this strategy, consider the following example in Figure 10 where a deaf signer from the second generation in Chicán was asked to perform a task similar to the one used in Le Le Guen (2011b). In Figure 10, the signer is facing East and the array she has to describe is located on the other side of the village. The signer was asked to locate the water tank with respect to the *comisaria* (the local administrative office). From where she is sitting, the arrangemen<sup>t</sup> stands to her back. She first mentions both entities using NPs and then places them in the signing space. However, her placement is not random, nor does it rely on her projected point of view (using an egocentric FoR), and she correctly locates the water tank south of the soccer field (where the *comisaria* is situated). Such a strategy uses the signing space symbolically but also respects the requirement of the truth condition exposed above and is performed with a geocentric FoR.

**Figure 10.** Spatial localization using the geocentric FoR, in placing the entities in the signing space according to their real-world orientation (on the left are diagrams that represent the array of objects and the signers and the signing area as viewed from above). "The soccer field (=*comisaria*) (**<sup>a</sup>**,**b**) is here (**c**), and the water tank (**d**,**<sup>e</sup>**) is there (=south of the *comisaria*) (**f**)" (Deaf signer form second generation of Chicán).

This same task was conducted with all the deaf signers in Chicán and results show that signers use the space around their body quite like Yucatec Maya speakers. When asked to locate two entities distant from their current location (but within their village), the grea<sup>t</sup> majority of signers placed them in the signing space according to a geocentric FoR, that is, according to their real-world orientation (see Tuz Baas forthcoming for more detail).

5.3.4. Towards an Arbitrary Use of the Signing Space

The explanation I propose is that signers behave like Yucatec Maya speakers in their use of space for spatial localizations. Signers also locate spatial arrangements based on their real-world orientation, using the signing space symbolically. However, when it

comes to agreement, speakers have an oral language to express these relations. Because YMSL is a visual medium, signers can recruit the spatial localization strategy and, putting on hold the truth conditions, they can now assign arbitrary points in the signing space to unknown entities (i.e., creating pronouns) and establish relations between them (i.e., marking agreemen<sup>t</sup> with the verb's movement). Such recruitment of local norms for the use of space in the surrounding culture seems to be a better explanatory factor as for why signers were able to use the signing space symbolically even since the first generation of deaf signers.
