Who’s going the distance here? Fictive motion in (German) sign language

2.2.1 Sign language constructions involved in fictive motion

2.2.1.1 Classifier constructions

Classifier constructions (CCs) are widely used across sign languages to depict the movement, location, or manipulation of an entity, as well as to provide information about the visual characteristics of an entity, e.g., its shape or size (Zwitserlood 2012).^[4] While they occur in ordinary conversation, CCs occur with higher frequency in narratives (Morford and McFarlane 2003). As Figure 2 illustrates, each form component of a CC contributes to the meaning of the whole. The handshapes of dominant and non-dominant hand characterize the entities involved, in Figure 2 an anthropomorphized cat walking upright and the top of a tram car. The locations and relative orientation of the hands with respect to each other depict where the cat is located with respect to the tram car, and the movement of the dominant right hand (index and middle finger wiggle forward on the non-dominant hand) depicts the movement of the cat.

Figure 2:

DGS: CC depicting the cat sylvester walking on top of a trolley (DGS Corpus, https://www.sign-lang.uni-hamburg.de/meinedgs/html/1419130_de.html#t00003104).

Experimental evidence suggests that not all components of CCs contribute to building meaning in the same way. Classifier handshapes are accepted as linguistic elements across linguistic theories: Their inventory is language-specific, they are acquired later than other components of CCs, and they are interpreted categorially. Locations and movements, on the other hand, form part of a more analogue system provided by human spatial cognition (for an overview of experimental work on CCs, see Zwitserlood 2021), which is why some see CCs as co-existing systems of visual and linguistic representation (e.g., Cogill-Koez 2000; Liddell 2003). More recently, the cognitive linguistics literature treats them as fully linguistic (Lepic and Occhino 2018, i.a.), a stance we build on.

Roughly, we distinguish between whole-entity classifiers, where the hands come to stand in for an entity (e.g., Figure 2), and handling classifiers, where the hands depict manipulating entities of different shapes (e.g., a mug vs. a toothbrush, Zwitserlood 2012).

2.2.1.2 Embodiment

In addition to representing referents on the hands, signers can also use their whole body to depict the actions, utterances, thoughts, and emotions of another (typically animate) referent. We use the term embodiment (Emmorey and Falgier 1999) for this phenomenon, which is also known as role shift or constructed action, among others (Aristodemo et al. 2021; Cormier et al. 2015; Hodge and Ferrara 2014; Perniss et al. 2010). With embodiment, a character’s body parts are mapped onto analogous body parts of the signer, and the event space is projected onto the signing space relative to the perspective and size of the character whose actions are portrayed (de Beuzeville et al. 2009; Quinto-Pozos 2007; Quinto-Pozos and Mehta 2010). The signer lets us see the event from the perspective of the character experiencing it, that is, from their point of view. As a discourse strategy, embodiment is considered to add vividness and informativeness to narratives, as it allows the most detailed portrayal of a referent’s actions and/or emotions. Presumably, this is because each of the signer’s (torso and up) body parts can depict an action of the referent simultaneously (Quinto-Pozos 2007).

A concept that is closely connected to embodiment is that of surrogate blends (Liddell 2003). A conceptual blend is an overlay of mental spaces in which structural components of each mental space are fused into a new, blended space (Fauconnier and Turner 1996). In a surrogate blend, the signer fuses real space, i.e. the actual space containing the signer and their immediate environment, with narrative space (Dudis 2004), which contains all the non-visible elements of the story event that are projected onto the blended space. In the example in Figure 3 below, the signer embodies a boy heaving himself onto a log. The narrative space thus contains at least a little boy and a log as potential referents to be included in the surrogate blend. The signer fuses this narrative space with the real space in front of him by creating a surrogate blend in which his head, torso, arms, and hands map onto those of the boy. The empty space that the signer’s hands are grasping is mapped onto the log in narrative space. The semantic structure of this embodiment then involves two referents, one that is made visible in the signer (the boy), and one that remains invisible (the log). However, we can infer the log’s position and potential movement via the hands’ position or movement in space.

Figure 3:

ISL: boy grabbing a log of wood.

Embodiment may be combined with CCs to show either the actions of a single referent (Figure 4a) or those of several referents (Figure 4b) (see, e.g., Dudis 2004 on body partitioning and megablends or Fischer and Kollien 2006 on parallelized constructed action). Finally, signers can use conventionalized signs simultaneously with highly depictive embodiment and classifiers (Aarons and Morgan 2003; Dudis 2004; Metzger 1995; Perniss 2007; Sallandre 2007; Sallandre and Cuxac 2002; Slobin et al. 2003).

Figure 4:

DGS: combinations of embodiment and CC. (a) The signer depicts Tweety bird swinging back and forth on a swing while whistling. He embodies the bird while at the same time showing Tweety’s legs hanging over the swing using a CC. (DGS Corpus, https://www.sign-lang.uni-hamburg.de/meinedgs/html/1413232_de.html#t00001345). (b) The signer embodies a person watching another referent (depicted on the left hand via an entity classifier) run past. (DGS Corpus, https://www.sign-lang.uni-hamburg.de/meinedgs/html/1419124_de.html#t00003533).

2.2.2 Fictive motion in the sign-linguistics literature

Though examples of fictive motion are attested in instructional materials and grammars for sign language teachers (for DGS, Papaspyrou et al. 2008: 111), descriptions of the phenomenon in the linguistics literature are rare. The earliest we know of is Lucas and Valli (1990). Through picture and video descriptions of people and objects moving past each other collected from three native ASL signers, they show that ‘perceived motion’ constructions form a productive subclass of CCs, able to depict the movement of different entities (e.g., people, vehicles, coins) in different directions, at different speeds, or in different constellations with respect to each other (e.g., passing each other on the side or one above the other). The construction is also used to show passing a plurality of entities (e.g., trees) through a repeated movement sequence of the hand representing multiple Ground objects. Perceived motion can involve either only CCs (e.g., a car driving along a road) or a combination of embodiment and a CC. In the latter case, the signer embodies the Figure involved in a motion event and their hand(s) represent some object/surface passing them at different angles.

Liddell (2003) considers different types of fictive motion in Talmy’s taxonomy and touches upon frame-relative fictive motion. Building on Lucas and Valli (1990), Liddell observes resemblances between fictive-motion constructions in sign languages and cinematographic strategies related to camera perspective. In representing a moving object as stationary, the signer creates the effect of a “viewing window” similar to a camera that films the moving object from the front while moving with it at the same speed (Liddell 2003: 297). The parallels to cinematography are elaborated in Müller (2018), who frames fictive motion in DGS in terms of tracking shots, i.e. shots in which the camera follows or moves alongside the person or entity being recorded. In cases of parallel use of embodiment and classifiers, Müller suggests this complexity allows the interlocutor to reconstruct a character’s perceptual experience during a motion event; as a narrative device, it helps us enter their viewpoint. For instance, when the hands pass at the sides of the signer’s head to show a character driving past a number of trees, the interlocutor has access to the spatial arrangement between Figure and Ground from an observer’s and from a character’s perspective simultaneously.

5.1.1 Anatomical and phonological motivations for fictive motion

Some of the instances of fictive motion we saw in Section 4 seem to be conditioned primarily by the affordances of anatomy as well as by phonotactic constraints operating across sign languages. As we showed in Section 4.2, signers sometimes embody the Figure in a motion event, i.e. the head, torso, and limbs of a character in the story are mapped onto those of the signer, and thus the signer takes on that character’s role in performing some type of movement. Now, movement of the signer’s body is limited both anatomically and phonologically in sign languages. Displacement of the signer upward along the vertical axis is limited; even standing erect on tiptoe will achieve only a few centimeters’ displacement. Downward displacement by bending the knees and/or sagittal spine flexion is anatomically feasible, as is horizontal displacement by lateral torso flexion and/or foot movement. However, the phonology of most established sign languages does not include articulation below the waist (Nyst 2007) – to put it simply, a signer embodying a character will not walk around to show that character’s movements. To circumvent the restrictions placed on a signer by their anatomy and the phonology of their language while still maintaining embodiment, they may either partition off their dominant hand to show translocatory movement with the help of an entity classifier (as illustrated in Figure 4a), or they may employ a fictive motion construction, perhaps with the help of metonymy (as illustrated in Figure 10). To show displacement of an embodied character, a signer may show its relative change of position with respect to a Ground entity by depicting the Ground as moving. The resulting construction increases articulatory efficiency by reducing the energy expenditure necessary to show the relative change in position between Figure and Ground. Instead of moving upwards either from a crouching position or onto one’s tiptoes, the signer’s hands may depict the Ground as moving down (e.g., a windowsill, a chasm), and instead of bending one’s knees one can show the Ground as moving up (e.g., a river’s surface). Fast movements in either a vertical or horizontal direction may be shown by multiple Ground objects moving past the embodied character (e.g., trees moving past a car, or different parts of a house façade racing across the signer’s field of vision).

Some movements that are within the purview of sign language phonology but physiologically difficult may be enlarged with the help of fictive motion. Embodying Sylvester peeking out from behind a corner of a wall (Figure 7b), for instance, requires the signer to displace her head laterally (and lightly sagittally) to the right to depict peeking. Displacement of the head is dispreferred from a physiological perspective: Lateral displacements are a cause for cervical spine injuries in sports (Swartz et al. 2005), and sagittal displacements cause the lower spine to go into hyperflexion and the upper cervical spine to hyperextend (Morrison 2018, cited in Tariq et al. 2022).^[8] Hence it is not surprising that this physiologically strenuous movement of the Figure should be kept small, and that it is complemented by fictive motion of the corner (Ground) to the left to enlarge the relative change in position between Figure and Ground.

While anatomy and phonology constitute one motivation for fictive motion, they are certainly not the only factors and, arguably, not even the main factors driving fictive motion constructions. In our DGS data set, only 28.8 % of all fictive motion tokens involve embodiment, which means that more than two thirds of the fictive motion constructions are formed only from classifiers and/or lexical signs, where only the signer’s hands need to move. We now turn to further, conceptual-perceptual motivations for fictive motion.

5.1.2 Cognitive-perceptual motivations for fictive motion

In fictive motion involving two classifiers, the hand depicting the Ground can either represent a single entity or a sequence of identical or at least similar entities. In both cases, fictive motion serves to show that a given translocatory movement is fast or covers a great distance. We will discuss each case in turn starting with the Ground hand representing multiple entities.

When the hand encoding the Ground depicts a series of quickly moving stationary objects, the effect is to indicate that a character is moving quickly past those objects. In our DGS dataset, nine tokens involve multiple Ground objects or multiple homogeneous parts of a single object (e.g., couplers of a drainpipe, rows of bricks). In NGT, we saw Tony Bloem depicting sections of a street passing underneath a car. This type of fictive motion is frequently found in sign languages; Papaspyrou and colleagues (2008) describe a similar scenario for DGS where a plane takes off. The signer repeatedly passes the non-dominant hand depicting stretches of runway under the dominant hand depicting the plane.^[9] Each iteration of the non-dominant hand’s path movement corresponds to one subpart of the movement event. We construe the immobile hand as representing the same entity (a car or a plane) undergoing a continuous movement, while the moving hand represents different subsections of the street/runway that are covered in each motion subevent.

Müller (2018) suggests that this strategy is grounded in physiology: it reproduces the eye saccades involved in scanning one stretch of ground when moving past it, then jumping back and scanning the next section of ground, etc. We propose that cognitive-perceptual reasons underlie the use of a fictive over a factive motion construction here. A factive-motion representation of a car racing along a street might have the B-hand representing the Figure moving forward either once or multiple times. If the hand depicting the car moves forward a single time, it is limited by the anatomy of the arm from showing that the car covers a great distance. If the car-hand moves forward multiple times, we run into a different problem. A repeatedly moving hand can be interpreted as referring to either the same entity across a series of (different) events, or to different entities engaged in the same type of event (see Kuhn and Aristodemo 2017 for pluractionality interpretations of repeated movement in LSF). In other words, repetition is interpreted as a quantification over events and repeating the movement of the car-hand should have the reading ‘the same or different car(s) repeatedly move along the same path’. This is not the reading we want to convey – we want to depict a single extended motion event. In a lexical predicate, repetition could be interpreted as marking continuous aspect and therefore at least implying that a great distance was covered by the car moving over an extended period of time (see Reagan 2007 on repetition as a continuous aspect marker). However, this marker does not apply to CCs denoting spatial events.^[10] As Wilbur (2008: 241) explains, the path movements of classifiers are “lexically specified as meaning spatial path, extent, or shape outline… Hence these paths are not available to be interpreted temporally….” Instead, movement repetition of a classifier denoting a spatial event is interpreted more iconically, as indicating multiple occurrences of an event (see Wilbur 2008: 242, fn. 28), perhaps allowing the transitional movements between each iteration to be interpreted as the entity moving back to the beginning, or, alternatively, allowing each repetition to indicate a new event with potentially different event participants. This is how we get the interpretation that several different stretches of road passed by. Hence, a fictive-motion representation is preferred over a factive-motion one in cases where a single entity passes a number of like entities.

Let us now turn to cases of fictive motion where the hand encoding the Ground object depicts a single entity. As mentioned in Section 4.1, all tokens of fictive motion with two classifiers exhibited movement on both the dominant and the non-dominant hand. The hands moved either towards one another, e.g., when Sylvester collides with a wall, or away from one another, e.g., when Sylvester climbs up a drainpipe, which itself seems to be moving downward. The effect here is to increase the perceived distance covered by the movement and/or its perceived speed, as the movement articulations of both hands are combined additively for the total effect.

At first sight, these constructions look like combinations of fictive and factive motion in one predicate. One hand encodes the Figure and the other the Ground object, and both hands articulate a path movement. Since movements in CCs are typically analyzed as meaningful (Zwitserlood 2003), each path movement should correspond to the movement of Figure or Ground, respectively. However, we argue that what we observe in these data are not, in fact, combinations of fictive and factive movement, but rather combinations of articulatory movements that create a single movement interpretation.

First, we note that combinations of fictive and factive motion within one predicate do not occur in spoken languages. Talmy (1996: 238–239) observes that English cannot represent fictive and factive motion via a single predicate (see 2b).

The jarring effect of (2b) may be comparable to that of zeugma, as in He lost his keys and his mind (Crystal 1999). Here lost is understood literally with respect to the first conjunct of the conjoined direct object, while it is understood as part of an idiom with the second conjunct. Since each sense of lose imposes different selectional restrictions on its object (a concrete vs. an abstract entity), and since a single use of the verb has to instantiate one sense, lose cannot select simultaneously a concrete and an abstract object. One can debate whether the oddity of such sentences is grounded in lexical meaning or more in discourse context and our world knowledge. Either way, the analogy holds: in fictive-motion constructions in spoken languages, unacceptability may result from the fact that a predicate cannot hold true factively of one argument (we in (2b)) and fictively of another (the scenery in (2b)) simultaneously.

Second, we observe that signers do not interpret the articulatory movement of the hand encoding the Ground object as representing movement of that object. We showed two DGS signers a re-telling of Sylvester rolling into a bowling alley, where the signer encodes the cat via a general entity classifier and the entrance to the bowling alley via the C()-classifier (see Figure 12).

Figure 12:

DGS: cat rolling into a bowling alley; 1()-hand shows factive motion of the cat and C()-hand shows fictive motion of the building’s entrance. https://www.sign-lang.uni-hamburg.de/meinedgs/html/1413234_de.html#t00004145

Both hands articulate a movement towards and past each other. To see whether the signers interpreted the construction as containing two meaningful movements, we signed each part of the construction separately (1()-hand moving towards C()-hand and vice versa) and asked whether they matched the cartoon in Figure 12. Both signers commented that only the utterance in which the 1()-hand moved towards the C()-hand matched the cartoon, but not the C()-hand representing the bowling alley’s entrance moving towards the 1()-hand representing Sylvester. Hence, signers do not interpret the movement articulated by the Ground-encoding hand to carry meaning independently of the other hand’s movement. Instead, it seems that signers perceive a single movement of the Figure, a movement composed of the two hands’ concerted articulations.

We thus propose that the encoding of an entity’s motion through space in relation to another entity, especially when this motion is fast or covers a large distance, is achieved through the combined movement of both manual articulators. Rather than assuming that each hand articulates a distinct semantic unit, we propose that the two hands together articulate one predicate, and that it predicates motion of a Figure with respect to a Ground entity by virtue of the relative change in position of the hands encoding Figure and Ground. That is, the calculation of movement distance is additive: the combination of the factive and the fictive distances.

The analysis we present raises questions for some generative theories of the morphological make-up of CCs. Supalla (1978) and subsequent publications treat movement of a single hand as a morphological unit (e.g., Emmorey 2002; Schick 1990; Zwitserlood 2003), but such a treatment is inadequate here because the movement of the hand encoding the Ground is not meaningful by itself. Alternative accounts such as DeMatteo (1977) claim that movement in CCs is best analyzed as analogous to the motion it describes rather than as a separate morpheme, but that analysis is also inadequate here, because the Ground object does not perform the motion we see the hand perform. In our examples, there is no single movement morpheme that denotes the movement of the referent associated with the moving hand; instead, this meaning arises from the interplay of articulatory gestures on both hands. A cognitive-linguistic approach like Construction Morphology (e.g., Lepic and Occhino 2018) may provide a descriptively more adequate account of frame-relative fictive motion in terms of a general ‘translocative motion’ schema that only specifies a change in position between Figure and Ground objects but leaves underspecified how this change comes about. We leave the concrete theoretical modeling of fictive motion constructions for future research.^[11]

5.2.1 Fictive motion beyond classifiers in narrative

We show here that fictive motion pervades the lexicon, as well. We can see this by comparing techniques of sign language narrative to techniques in the established lexicon of sign languages, where the data below were drawn both from our DGS dataset and from spreadthesign.com (as accessed on 1 March 2023^[12] unless otherwise noted). Where possible, we checked with dictionaries or lexical databases, including:

1. ASL: https://www.signingsavvy.com/

2. BSL: https://bslsignbank.ucl.ac.uk/

3. DGS: https://www.sign-lang.uni-hamburg.de/meinedgs/overview/start.html

4. LIS: https://pcmac.download/app/917187457/dizionario-lis

5. LSF: http://www.sematos.eu

6. NGT: https://signbank.cls.ru.nl/

The signs in question recognizably contain classifiers that have been entrenched in the lexicon of their respective sign language. In our Canary Row data from DGS, the sign enter involved movement of both hands toward each other 35 times (see Figure 13). Both the moving Figure and the enclosure that it enters are represented by entity classifiers. Whether this is fictive motion or simply a phonological process of opposite movement, we cannot say. While certainly not all movement in sign phonology encodes physical movement, the movement shown here is at least compatible with fictive motion.

Figure 13:

DGS: enter.

The fictive-motion analysis of enter finds support by looking at other lexical items where similar parallel articulations of Figure and Ground hand are attested. For example, in the DGS sign flee both hands move in eleven tokens; here the dominant B()-hand moves under the non-dominant curved-B()-hand. This curved-B()-hand represents the exit of some enclosure, from which the entity represented by the dominant hand flees. Similar movement into or away from an enclosure, where the hand representing the enclosure moves away from the hand representing the Figure, is attested in hide in the sign languages of Brazil, Bulgaria, Denmark, Estonia, and Finland; and in escape in the sign languages of China, the Czech Republic, Finland, Germany, India, Latvia, Mexico, Slovakia, Sweden, and the UK.

Further common cases of co-articulation of Figure and Ground hands are the signs down (seven tokens), and up (sixteen tokens) in our DGS data.

Another example is train in Swedish SL (STS, Figure 14). The hand representing the train tracks (Ground) moves repeatedly underneath the hand representing the train itself (Figure). This is comparable to what we saw in Figure 8d, where the hand representing the road moves while the hand representing the car also moves.

Figure 14:

STS: train.

Another common example, though complex, involves metonymy (as we saw in Figure 10 with the lip of the garbage can): drive in the sign languages of Germany, Poland, Latvia, Portugal, Sweden, and the United States. The signer’s hands move forward, seemingly gripping a steering wheel. Movement of the driver/steering wheel forward is portrayed via moving only the hands through extension of the elbow, a joint that would otherwise be stable (i.e., not articulate) during driving. Factive movement of the car is thus shown metonymically via forward movement of the steering wheel and/or the driver’s hands. Since the driver’s torso remains still, there is also a fictive component, as forward movement of the hands/steering wheel can be shown only via elbow extension, which is not a part of driving.

Likewise, grow in the sign languages of America, Austria, and the Czech Republic features a ‘growing’ entity represented by a 5()-hand (Figure) that rises out of a C()-hand forming some kind of enclosure (Ground). Meanwhile, the C()-hand moves downward, resulting in a larger relative change of location between Figure and Ground. In elevator in the sign languages of America, Brazil, and Mexico, the hand that represents the wall moves vertically repeatedly and in the opposite direction from the hand that represents the moving elevator.

Finally, a range of examples in which we have fictive motion of the Ground while the Figure is represented via embodiment also occur (as in Section 4.2). Examples echoing that in Figure 7b include peek in the sign languages of Croatia and Finland, and sneak in the sign languages of Belarus, Iceland, Spain, and the UK.

We conclude that fictive motion of the wide variety of types found in narrative sparkles across sign language lexicons.

5.2.2 Fictive motion across sign languages

The classifier examples discussed in Sections 2 through 4 come from seven different Western sign languages, which might not all be genetically related, but which are open to claims of areal similarity (see Börstell 2017 for areal phenomena in Scandinavian sign languages). The lexical examples brought up in Section 5.2.1, however, are drawn from multiple languages around the globe, where neither genetics nor areal similarities are pertinent. Such examples spur us on to expect sign languages around the globe to use classifiers that make use of fictive motion. This expectation is met. Here we present examples of fictive motion in narratives of three non-Western sign languages.

In Figure 15 from Taiwanese SL (TSL), the signer embodies a hawk but partitions his hands off to form two classifiers: the right hand encodes the hawk itself or its beak, while the left hand represents the ground underneath. As the hawk spies his prey, the right Figure hand stays in one position. The left hand flaps (wrist extension and flexion), just like we observed for B-Flap in Section 4.4, showing the fictional movement of sections of the ground. From this we understand that the hawk is factively flying fast.

Figure 15:

TSL: Hawk following his prey, fictive motion of the ground-hand (min. 0:56–0:57, https://www.youtube.com/watch?v=0vFOwgMHrSk).

In Figure 16 from Fijian SL, the signer embodies kittens running fast (Figure) in between the legs of a boy. The signer’s hands form body-part classifiers () that represent the boy’s legs (Ground) (Figure 16a). Instead of the kittens running forward, however, we see the legs of the boy moving backward, i.e. the signer’s hands articulate an arc-shaped backward movement past the signer’s head (Figure 16b). Since the boy does not, in fact, move backward, but instead stays in one place as the kittens move forward, the movement that the classifiers convey is fictive, which allows us to understand that the kittens have factively moved.

Figure 16:

Fijian SL: kittens race between a boy’s legs, fictive motion of the boy’s legs backwards (from min. 00:06 of page 10, https://deafworldaroundyou.org/View?id=168).

In Figure 17 from Turkish SL (TİD), the signer embodies a boy who ducks down behind a couch to hide from his shadow. His hands are handling classifiers, representing gripping the top edge of the back of the couch. The signer crouches forward and lowers his head a little, showing the factive motion of the boy. At the same time, his hands raise, showing the fictive upward motion of the couch through a surrogate blend.

Figure 17:

TİD: embodiment and fictive motion (from min. 0:32 https://www.youtube.com/watch?v=dyVhXCp_S1Q&list=PLz551fY8YMLgkqoQj4Gb88vpRUW7VIVzN&index=5).

These examples suggest that a wide range of types of fictive motion is available in a typologically diverse range of sign languages.