Environmental constraints shaping constituent order in emerging communication systems: Structural iconicity, interactive alignment and conventionalization

Highlights

• We report three experiments on emerging structure in gestural communication systems.

• We investigate how environmental factors shape syntax in novel communication systems.

• Constituent order reflects event structure via the principle of structural iconicity.

• Constituent order is further shaped by the principle of interactive alignment.

• Skewed distributions of event types can result in order conventionalization.

Abstract

Where does linguistic structure come from? Recent gesture elicitation studies have indicated that constituent order (corresponding to for instance subject–verb–object, or SVO in English) may be heavily influenced by human cognitive biases constraining gesture production and transmission. Here we explore the alternative hypothesis that syntactic patterns are motivated by multiple environmental and social–interactional constraints that are external to the cognitive domain. In three experiments, we systematically investigate different motivations for structure in the gestural communication of simple transitive events. The first experiment indicates that, if participants communicate about different types of events, manipulation events (e.g. someone throwing a cake) and construction events (e.g. someone baking a cake), they spontaneously and systematically produce different constituent orders, SOV and SVO respectively, thus following the principle of structural iconicity. The second experiment shows that participants’ choice of constituent order is also reliably influenced by social–interactional forces of interactive alignment, that is, the tendency to re-use an interlocutor’s previous choice of constituent order, thus potentially overriding affordances for iconicity. Lastly, the third experiment finds that the relative frequency distribution of referent event types motivates the stabilization and conventionalization of a single constituent order for the communication of different types of events. Together, our results demonstrate that constituent order in emerging gestural communication systems is shaped and stabilized in response to multiple external environmental and social factors: structural iconicity, interactive alignment and distributional frequency.

Introduction

Language structure is a highly complex phenomenon evolving in response to various potentially interacting pressures at several time scales, from online interaction to phylogenetic evolution (Beckner et al., 2009, Rączaszek-Leonardi, 2010). Consequently, it is a challenging task to reconstruct the evolutionary trajectories of existing languages and the forces that have shaped them (Tylén, Fusaroli, Bundgaard, & Østergaard, 2013). Prevalent approaches in the language sciences have pointed to a variety of motivations for linguistic structure. For instance, it has been suggested that syntactic structures are innate and genetically determined (Hauser et al., 2002, Nowak et al., 2001, Pinker and Bloom, 1990). Others have argued that linguistic structures are motivated by latent internal cognitive biases gradually amplified through iterated cultural transmissions: structures that are easier for human cognitive systems to learn and use are selected for and thus increasingly propagated through cultural history (Brighton et al., 2005, Christiansen and Chater, 2008, Deacon, 1997). Yet other approaches emphasize inherent semantic relations: for instance it is argued that subjects and objects are semantically primary and therefore tend to syntactically precede actions (Goldin-Meadow et al., 2008, Hall et al., 2014). Despite different starting points, these theories all assume that linguistic structures originate independently of the referent situations they encode and of their contextually and communicationally situated use. Rather than being an independent and closed system, we argue that language structure is intimately intertwined with its social and functional role in human life (Evans and Levinson, 2009, Tomasello, 2008). Language is first and foremost used to coordinate joint action and to communicate about the world: we share experiences, coordinate action, instruct each other, tell stories, gossip, make declarations and maintain social relations (Clark, 1996, Fusaroli et al., 2014, Tylén et al., 2010). Consequently, language is continuously shaped by external constraints, such as structure in the external world as well as social dynamics (Bergmann et al., 2013, Dale et al., 2013, Fusaroli and Tylén, 2012, Tylén et al., 2013). Adopting such functional perspectives suggests a more articulated investigation of the way in which different aspects of language use provide resources for and pressures on evolving linguistic structure (Croft, 2001). In this paper, we investigate a plurality of environmental factors shaping constituent order (word order), a key component of linguistic structure.

Many languages, including English and Danish, have fixed constituent orders encoding participant roles in verbalizations of transitive events. Some events have only one plausible interpretation regarding who is performing an action and who, or what, is affected by it. This is the case, for instance, in the English sentence “John eats cake”. The functional role of constituent order is more apparent when considering semantically ambiguous events, as in “John hits Mary”. Verb-final languages (including Turkish and Japanese) often have case marking systems, which aid speakers in attributing participant roles (Greenberg, 1963: Universal 41; Bentz & Christiansen, 2013). However, in many other languages, such as English and Danish, the participant roles are often only decipherable based on constituent order. Thus, in the above case, the unmarked and fixed subject–verb–object (SVO) constituent order facilitates the correct interpretation of the ambiguous events. Together the SOV and SVO constituent orders account for 89% of the 1188 languages included in a survey featured in the World Atlas of Language Structures (Dryer, 2011). Given a total of six possible combinations of subject, verb and object orders, a particularly interesting and persistent question is why SOV and SVO are more common than other word orders. Moreover, there is converging evidence from different strands of linguistic research suggesting that the SOV order is predominant in less conventionalized languages and emerging sign systems (Napoli & Sutton-Spence, 2014). Though only few emerging sign languages have been studied, findings seem to support a unique status for SOV order. For instance, SOV is the dominant constituent order found in the Al-Sayyid bedouin sign language, a language surrounded by older spoken languages displaying SVO order (Sandler, Meir, Padden, & Aronoff, 2005). Another line of evidence indicates that SOV order is predominant in gestural communication systems spontaneously invented in early childhood by deaf children with hearing parents prior to exposure to conventionalized signed or spoken linguistic input (Goldin-Meadow & Mylander, 1998). More recently, researchers have developed inventive experimental approaches to investigate the cognitive underpinnings of novel sign systems. These studies used nonverbal gesture elicitation tasks to demonstrate that hearing adult non-signers, when asked to represent non-ambiguous transitive events (i.e. human agents performing actions on objects) using only gesture, have a strong preference for the SOV order, regardless of their linguistic backgrounds (Goldin-Meadow et al., 2008). The authors explain their observations by reference to inherent semantic relations: Entities (such as agents and patients) are argued to be cognitively more basic and less relational than actions, which might lead participants to specify entities before the more abstract actions. The results have subsequently been replicated by Langus and Nespor (2010) with speakers of Italian (SVO) and Turkish (SOV). Despite differences in the proposed underlying mechanisms, these findings have been interpreted as evidence for a universal, internal cognitive predisposition to conceptualize transitive events according to a specific order analogous to SOV structure in natural language.

More recently, however, Schouwstra and de Swart (2014) have extended these investigations showing that intensional events – a class comprising four distinct event subtypes (Forbes, 2010) – can revert this pattern and consistently motivate SVO gesture strings. Intensional events include unobservable mental events such as “thinking of x” or “wanting x”. Again, the authors interpret this as supporting the semantic origins of constituent order: Since direct objects in intensional events (e.g. “the pirate is thinking about a guitar”) are more abstract and relational, as they do not have a real extension in space, they are preceded by actions. Other studies have shown tendencies of participants to switch to SVO when gesturing about so-called semantically reversible events (Gibson et al., 2013, Hall et al., 2013, Meir et al., 2010). Contrary to irreversible events, semantically reversible events have human agents and patients (e.g. “the baker hit the ballerina”), which potentially render their respective roles ambiguous. It has thus been speculated that SVO constituent order helps disambiguating agents and patients by maximally separating them by the verb phrase (see Hall et al., 2013 for a discussion).

Previous observations of the strong tendency for participants to produce gesture strings with SOV order seem indisputable and robust. However, we ask whether the proposed underlying mechanisms account exhaustively for the constituent order in gestural representations of transitive events. We suggest that additional factors may influence gesture order. In particular, we hypothesize that environmental and social–interactional factors not considered in previous studies might have a strong influence on gesture order. In the following, we report three experimental studies in which we systematically explore (i) whether constituent order is motivated by the structural organization of the real world referent events they denote, (ii) whether constituent order is shared through the human propensity to imitate and align linguistic behaviors, and (iii) whether, through frequency of exposure and use, these behaviors eventually conventionalize to create stable, consistent and cognitively economical patterns. Crucially, these are all environmental factors, that is, not intrinsic to language or internal cognitive systems, which may significantly contribute to the shaping of linguistic structure.

Ferdinand de Saussure famously argued that linguistic signs are by definition arbitrary, that is, related to their referents by mere convention (De Saussure, 1972), a position that is still widely represented in linguistics and psychology (Levelt et al., 1999, Nielsen and Rendall, 2011). However, many observations challenge the generality of this claim (Fischer and Nänny, 2001, Flumini et al., 2014, Perniss et al., 2010). For instance, studies on sound symbolism have shown that many linguistic sound-meaning mappings are in fact non-arbitrarily motivated. When presented with antonyms in Thai, Kanarese, and Yoruba languages, English speaking participants were found to perform significantly above chance in mapping their referents, indicating that phonetic qualities of words provide cues revelatory of their meanings (Slobin, 1968). Similar observations have been made in a number of experiments requiring participants to evolve new communication systems online in response to joint collaborative tasks (Fay et al., 2010, Galantucci, 2005, Garrod et al., 2007). Again, participants were found to reliably employ strategies based on iconic representations, that is, they developed signs bearing resemblance to their referents, arguably to facilitate their addressees’ comprehension.

While the evidence for iconic mappings between form and meaning in the lexical domain is widespread, less attention has been directed at iconic motivations in syntax (for an exception, see e.g. Haiman, 1985). Structural iconicity is a particular type of iconicity in which the structure of events or relations between referents is replicated in the syntax of a spoken or signed utterance. It can be defined as a non-arbitrary, motivated relationship between form and meaning, which is established when the arrangement of individual signs mirrors actual properties of the relations between their referents, i.e. in transitive events. Indeed, in some languages, the relative order between actual sequential events in the world is grammaticalized and reflected in speech (Itkonen, 2005). A well-known example is given by the famous quote “veni, vidi, vici” (I came, I saw, I conquered) from Julius Caesar’s letter to the Roman senate, which clearly illustrates how the sequential order of events is preserved in language, but the same kind of iconic relationship between language and the world is also evident in Mandarin Chinese (Tai, 1985). Beyond simple relations of resemblance, structural iconicity also covers more diagrammatic mapping relations (Fauconnier and Turner, 2002, Stjernfelt, 2007, Tylén et al., in press). For instance, an apparently atemporal dependency relation (x depends on y) can be represented by a temporal relation (x precedes y). This is likely to be reflected in constituent orders as well. The constituent orders found in many of the world’s languages, underlying the linguistic expression of transitive events, might thus not be coincidental or driven by language- and cognition-internal factors alone. Constituent order might also be motivated by structurally iconic relations to the referent events.

By itself, structural iconicity would lead speakers to rely exclusively on structural properties of the referent scene as motivation for representational structure in communicative exchanges posing questions for the many apparently more arbitrary relations also found in syntax. However, interactive alignment has been shown to be an important mechanism underlying online choices of structure in communicative utterances (Fusaroli and Tylén, 2012, Pickering and Garrod, 2004b). Interactive alignment is the spontaneous propensity of interlocutors to flexibly adapt to each other in the course of conversations thus displaying increasing similarity in their way of speaking and referring to communicational topics. Such adaptations have been observed in many interactional behaviors (Fusaroli, Konvalinka, & Wallot, 2014): from subtle bodily sway (Shockley, Santana, & Fowler, 2003), to speech rate, utterance length and phonetic profile (Fusaroli and Tylén, in press, Giles et al., 1991), lexical (Fusaroli et al., 2012), and conceptual alignment (Angus et al., 2012, Garrod and Anderson, 1987, Garrod and Doherty, 1994). Likewise, a number of studies show that interlocutors tend to align on their use of syntactic constructions beyond particular tokens of referent events: If a speaker uses a double object construction (“the pirate gives the chef an apple”) to refer to a ditransitive scene, there is a relatively higher probability that her interlocutor will spontaneously use the same construction to describe analogous but not identical scenes, even though the prepositional object construction (“the pirate gives an apple to the chef”) is an equally acceptable alternative (Branigan et al., 2007, Branigan et al., 2000, Hopkins et al., 2015, Reitter and Moore, 2014). Rather than purely relying on the referent event (structural iconicity), speakers widely rely on the linguistic structures offered by their interlocutor. The pressure for interactive alignment is argued to relate to the cognitive economy of communication itself: the alignment of linguistic representations is thought to facilitate the sharing of situation models (mutual understanding) and establishes parsimony between linguistic production and comprehension making dialogical interaction “easy” (Ferreira and Bock, 2006, Pickering and Garrod, 2009, Pickering and Garrod, 2013).

Albeit distinct constraints, structural iconicity and alignment can reinforce each other in shaping the structure of communication systems. This is the case when interlocutors repeatedly select and refer to the same aspect of their shared environment, reciprocally reinforcing structural iconicity through alignment. However, the two constraints may also compete: If interlocutors repeatedly encounter very different referent events, the inclination to imitate the structure of the referent and to imitate the interlocutor might be in conflict, possibly reducing the relative influence of one or both motivational pressures.

Importantly, structural iconicity and interactive alignment would potentially afford the evolution of a wide range of context-specific and highly variegated structural forms. Although interactive alignment motivates the spread and sharing of forms and representations, it cannot alone account for the gradual conventionalization of general communicative routines consistently observed in studies on emergent communication systems and for the stability observed in most well-established languages (Clark and Wilkes-Gibbs, 1986, Fay et al., 2013, Galantucci and Garrod, 2010, Mills, 2014).

Again, we would argue for the importance of communicative constraints: for a communication system to be optimally functional and efficient, the encoding of relevant meaning differences is not enough. An efficient communication system should also be easy to remember, produce and comprehend (Fay et al., 2008, Kirby et al., 2008). Having to master competing linguistic structures would require more cognitive resources. There could thus be strong motivations for stabilizing a single rather than several co-existing forms to refer to tokens of a set of related events (Kirby et al., 2008). But do all forms have an equal chance of becoming conventionalized? Other environmental factors might have an impact on which forms eventually outcompete others and stabilize. In some cases it might be a question of which form is more salient due to functional or sheer frequency factors. For instance, language learners have been shown to be sensitive to distributional frequencies of lexical and grammatical categories in their linguistic input (Reeder et al., 2013, Wonnacott et al., 2008). Furthermore, languages relying solely on an absolute spatial frame of reference (e.g. north, south, east and west) tend to be associated with speech communities living in rural and open-terrain environments offering stable landmarks such as hilltops and rivers for these cardinal directions, while relative frames of reference (right, left, next to etc.) are more consistently found in languages spoken in areas with dense forests or urban environments (Levinson, 1996, Levinson, 2003, Majid et al., 2004). It has also been found that the regional level of UV light can have implications for color categories found in the languages of the corresponding regions (Lindsey and Brown, 2002, Plewczyński et al., 2014).

Together these communicational and environmental motivations continuously shape communication systems toward parsimony and elimination of structural redundancy (Tomasello, 1999, Tomasello, 2008). For instance, most of the world’s languages have a single basic constituent order with which speakers predominantly communicate transitive events (Dryer, 2011, Greenberg, 1984). Although such generalizations seem to potentially work against the general principles of iconicity suggested in previous sections, they facilitate communication by simplifying the procedures or rules for generating and comprehending utterances: It is cognitively and communicatively more economic to rely on a limited repertoire of general principles than a larger repertoire of highly contextualized rules.

In order to systematically address the impact and interaction of multiple environmental pressures on the development of new communication systems, we conducted a series of three experiments, where pairs of participants communicated about simple transitive events using only gesture. The first study is contingent on the very general observation that the world has structural properties available as resources for human communication. The study thus investigates how structure intrinsic to referent events in the stimuli may, by itself, shape the manner in which the stimuli are communicated. In contrast to the earlier emphasis on internal semantic relations (Goldin-Meadow et al., 2008), we suggest that the choice of constituent order might reflect the structural organization of the referent events themselves, through the principle of structural iconicity. In order to test this prediction, we contrasted simple transitive events of the type originally used by Goldin-Meadow et al. (2008) with a structurally different type of transitive event. The former type can be characterized as object manipulation events (e.g. “the doctor eats the cake”). In these events, the referents assuming agent and patient roles must be physically co-present before the action can be purposefully performed. In other words, in an object manipulation event, the patient logically precedes the action being executed: obviously, one cannot manipulate or act upon an object, which is not physically or symbolically already present. By contrast, in a different type of transitive event, which we henceforth call object construction events, agents perform actions that cause objects to come into existence. This type of event can be exemplified by sentences like “the doctor bakes a cake”. In these cases, actions precede objects that, in turn, are dependent on the performed actions. We predict that the stimulus events will motivate different constituent orders following the principle of structural iconicity: Participants will produce SOV gesture strings for object manipulation events and SVO for object construction events. Notice that although our category of construction events corresponds to one of the subcategories of Schouwstra and de Swart’s intensional events (2014) the contrastive manipulation in this experiment is motivated from the quite different perspective of structural iconicity.

In experiment 2, we investigate the effect of interactive alignment on the choice of constituent order. If interlocutors are inclined to map the structure of the stimulus event, and at the same time are primed to imitate the constituent order used by their partner, what will happen when these are in conflict? We predict that in such situations, the pressures will be competing possibly resulting in weakening any prior effects of structural iconicity.

In experiment 3, we investigate one of the possible pressures leading to the stabilization and conventionalization of a single constituent order for both manipulation and construction events. We predict that if the two types of stimulus events differ in frequency so that participants encounter one type much more frequently than the other, they will, over repeated trials, be inclined to generalize the constituent order used for the more frequent event type to the less frequent one, thus pointing toward conventionalization of one general constituent order.

Lastly, we examine how these different environmental pressures might contribute to and interact with each other in the evolution of communication systems for talking about transitive events.