Patterns of Conceptual Encoding in ASL Motion Descriptions Author(s): SARAH TAUB and DENNIS GALVAN Source: Sign Language Studies, Vol. 1, No. 2 (Winter 2001), pp. 175-200 Published by: Gallaudet University Press Stable URL: https://www.jstor.org/stable/26204836 Accessed: 20-03-2019 20:02 UTC REFERENCES Linked references are available on JSTOR for this article: https://www.jstor.org/stable/26204836?seq=1&cid=pdf-reference#references_tab_contents You may need to log in to JSTOR to access the linked references. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org. Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at https://about.jstor.org/terms Gallaudet University Press is collaborating with JSTOR to digitize, preserve and extend access to Sign Language Studies This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms SARAH TAUB DENNIS GALVAN Patterns of Conceptua Encoding in ASL Mot Descriptions The human mind constructs complex conceptual models of real-world situations. Yet much of this information is not explicitly represented in the linguistic messages people use to communicate about these situations. Languages divide into different typological groups based on how they encode information about motion events—in particular, which conceptual elements are encoded and by what linguistic means. The present study looks at patterns in American Sign Language, drawing data from adults' retellings of the story Frog, Where Are You? (Mayer 1969). There appear to be no absolute restrictions on what meaningful elements can co-occur in ASL classifier forms; instead, we see tendencies that are most likely based on cognitive heuristics and communicative strategies. In this sort of analysis, we look at the conceptual elements that a language might choose to express, the linguistic "surface" forms that are available to express them, and the patterns of encoding of con ceptual elements by particular types of surface forms. Talmy (1985) examines several conceptual areas (e.g., motion, causation, and tem poral distribution), but the area that has inspired the most research concerns motion events. Sarah Taub, Ph.D., is Assistant Professor of Linguistics in the Department of ASL, Linguistics, and Interpretation at Gallaudet University in Washington, D.C. Dennis Galvan, Ph.D., is Associate Professor of Psychology in the Department of Psychology at Gallaudet University in Washington, D.C. The sign illustrations are by Liz Dean. 175 This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms 176 I Sign Language Studies A motion event—that is, an event in which an entity moves from place to place or is identified as located at a particular place—may be divided into a number of basic conceptual pieces: the Figure, or moving/located entity; the Ground, or landscape against which it moves; the fact of Motion (or location) itself; and the Path along which it moves (or location where it stays). We may also identify the Manner of its movement; Manner is less conceptually unified but may include the locomotion method of an animate figure (e.g., run ning, clambering) and local details of the motion (e.g., sliding, bounc ing, or tumbling). Finally, for some events a Cause, or external event, is seen as providing the conditions for the motion event. Talmy used this analysis to identify three different patterns of ex pressing these elements (see example i). The first grouping (dubbed path-type by Slobin and Hoiting 1994) expresses Path and Motion in the verb root; Manner and Ground information, if present at all, occurs in surrounding elements, or satellites. This group includes Ro mance, Semitic, and Polynesian languages; (ia) gives an example from Spanish. The second group (manner-type), consisting of the other Indo-European languages and Chinese and exemplified by En glish in (ib), conflates Manner and Motion in the main verb root, with Path given in satellites; these languages can also combine Cause and Motion in verb roots. Finally, the figure-type languages conflate Figure and Motion in the verb root and give Cause, Path, and Ground in satellites; the Hokan group of Native American languages are of this type, as exemplified by Atsugewi in (ic). In later work, Talmy (1991) picked out Path as the "core schema" of motion events and categorized languages based on how they ex pressed Path. Languages with Path in the verb root are called verb framed, and those with Path in satellites are satellite-framed. Thus path type languages are by definition verb-framed, and all other types are satellite-framed. Talmy noted that satellite-framed languages could actually include more Path information per sentence than verb framed languages because they could concatenate in a single sentence many satellites specifying complex paths. While Talmy focused on encoding patterns at the sentence level, Slobin (1996) looked at whole texts or narratives. One might suppose that languages such as Spanish, although they encode less Path infor mation in each sentence, might "catch up" over the course of an This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms Patterns of Conceptual Encoding | 177 Example i . a. Spanish; path-type: la botella entrô a la cueva flotando the bottle moved-in to the cave floating " The bottle entered the cave floating." b. English; manner-type: The bottle floated into the cave. c. Atsugewi; figure-type: /'-w- ca third-person st'aq'for a subject fro ic't runny matter to mo / third-person —► [c'wast'aq "Icky entire texts than stuff text. include compara persist ies as T across an influ Evidence also tional Path a panying gestu example, motion move McN descri their h and other M made a thoro encoded gesture The conc is precedin languages some tak of of whic This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms 178 I Sign Language Studies signed languages have lexical verbs (e.g., run, walk, and throw in ASL), many of which undergo spatial modulation to show specific paths; classifiers (or polymorphemic verbs), which incorporate iconic or semi-iconic articulator configurations and movements (cf. Mc Donald 1982, Supalla 1982, Schick 1987, Liddell 1995); and referen tial shifts, in which the signer's face and body represent the face and body of some referent entity (cf. Emmorey and Reilly 1995). In a study of narratives by native ASL and English users, Galvan and Taub (forthcoming) found that ASL signers consistently incorpo rate much more conceptual information into their descriptions of motion events than do English speakers. In particular, significantly more Motion, Path, and Manner information was included, and there was a strong trend toward including more information about affect and spatial relations among Figures. Moreover, ASL signers re peated this information more frequently than English speakers—this contrast was significant for Motion, Path, Manner, and Affect and approached significance for Spatial Relations. We also noted a strong preference in the ASL narratives for expressing conceptual elements through classifiers and referential shifts. These findings show the deep influence of iconicity on ASL de scriptions of motion events. Even when alternate forms (nouns, lexi cal verbs, fingerspelling, or extended explanations) are available, signers overwhelmingly use iconic classifiers and referential shifts. Engberg-Pedersen (1999) notes this phenomenon in Danish Sign Language and refers to it as a signed-language narrative ideal—in a signed story, wherever possible, narrators should "show" the infor mation using iconic forms rather than "describe" it using lexical forms. The greater amount and repetition of motion infomiation in ASL stories can be attributed to iconicity as well. One might conceive of the repetition of information as linguistic anaphora or agreement, but in fact the cause might not be specific to language. When multiple iconic forms of any sort are constructed, they portray many aspects of the referent event, and each iconic form must "agree" with others in accurately portraying those aspects; this leads naturally to "redun dancy" across forms. For example, Galvan and Taub found that in narratives about a boy, nearly all the forms presenting the boy's Mo tion and Path through space also contained Figurai information; this This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms Patterns of Conceptual Encoding | 179 information consistently (and redundantly) portrayed him as a two legged animate being. Although we often think of repetition or re dundancy as boring or unnecessary, the repetition in these iconic narratives serves an important purpose: It reinforces the viewer's mental imagery and helps the viewer comprehend new information that is mixed in with the repeated information. This line of thought leads directly to the question of conflation patterns in ASL representations of motion events: Are there con straints or limits on iconicity in ASL? A priori, because the articula tors of ASL are objects (i.e., body parts) moving in space, one might expect that signers would encode all the conceptual information about a motion event in a seamless flow of iconic representation. That is, all of Talmy's pieces of a motion event might be conflated into one classifier form. Previous work, however, has made it clear that this does not happen. ASL signers do in fact separate different pieces of the event into different linguistic components. Supalla (1990) took explicit notice of this issue and pointed out a number of separation patterns in ASL. Most classifier-based verbs of motion follow one of two patterns: They may combine Figure infor mation with Manner of locomotion or with Path information. The Manner/Figure/Motion verbs use the signer's body to represent the referent's body, focusing on details of the referent's limbs; examples include the use of the signer's fingers or hands and forearms to repre sent the referent's legs and feet. The Path/Figure/Motion verbs, in contrast, use only the signer's hand or fingers to give an overall view of the referent; the signer's body is not semantically significant. Ex amples include the uses of the Flat B hand or extended index finger to show movement of a person or thing through space. The "two legged" classifier made by extended index and middle fingers (cl:v) is an exception in that it may encode both Manner and Path as well as Figure and Motion. (Although Supalla does not make this observa tion, the exception could be explained by the fact that the cl:v alone has a part-whole structure that can show limb movement as well as the relative freedom of movement of an item articulated solely by a single hand.) Supalla noted that these two types of motion verbs often combine in a series, allowing the signer to express both Manner and Motion information in the same sentence. Thus, to describe a person limping This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms i8o I Sign Language Studies up a hill, a signer would first produce a Manner motion verb the limping body motions and then a Path motion verb show person's trajectory up the hill. Even though it is physiologica ble for the signer to move the Manner verb through the uph signers do not usually do so; instead, they separate a putatively event into two conceptual chunks articulated in series. Su ferred to these patterns as arbitrary linguistic restrictions on whi of an event could appear together in which forms. One may question Supalla's interpretation, however. Nat signers do accept some Manner/Figure/Motion verbs that definite path (Samuel Hawk, personal communication). In forms, both limb movements and motion along a path ar salient. For example, picture a man shuffling sideways along window ledge to reach a trapped cat. Representing his mo with two index fingers, one for each leg, moving sideway cat's spatial locus, is acceptable to most signers, although classes the double-index-finger classifier form as incompatibl Path information. Moreover, at least some of Supalla's Path/F Motion forms do incorporate Manner information; for examp vertical index-finger classifier representing a person's motion gently up and down to show that the person is walking (rath say, running). Given this, it may be more accurate to interpr la's observations as tendencies rather than absolutes. Moreove might seek a motivation for the tendencies in general cognit perceptual functioning rather than in arbitrary restrictions. Slobin and Hoiting (1994) studied verbs of motion in Nethe Sign Language (SLN), whose classifier system is similar to nature although not in every detail. They support Supalla's vi motion events are conceptually separated into different fo disagree on the details. In particular, whereas Supalla claims t series of a Manner and a Path motion verb, the Path verb is se and "reduced," Slobin and Hoiting claim that the Path verb ally primary. To support this point, they adduced examples f ken languages that use verb series of Manner and Path verbs; cases, the Path verb generally functions as the sentence's mai Slobin and Hoiting also contend that, because (according t 1991) Path is the "core schema" and because ASL verbs of This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms Patterns of Conceptual Encoding | 181 necessarily display Path, ASL is a verb-framed, path-type language. They introduced the notion of complex verb-framed languages; this type would consist of verb-framed languages such as Turkish, ASL, and SLN that use verb series to convey motion information. As additional support for their position, Slobin and Hoiting note that verb-framed languages characteristically give special treatment to crossings of boundaries. Satellite-framed languages such as English treat boundary crossings in the same way as any other Path seg ment—through the addition of a satellite element specifying that seg ment (cf. example 2a); these satellites normally follow a verb of Manner. But verb-framed languages differ. As we have seen (cf. ia and 2b), for the most part verb-framed languages express Path in the main verb, with a satellite expressing manner. However, in certain contexts (called path focused by Slobin and Hoiting), Path may appear as a satellite with a Manner main verb (as in 2c). Yet those Paths may never include a boundary crossing. A sentence such as (2d), which attempts to combine a boundary-crossing Path with a Manner verb, is unacceptable. Example 2. The man ran out of the surf, across the field, and into the ho el hombre entrô corriendo a la casa the man moved-in running to the house el hombre corriô hasta la casa the man ran up-to the house * el hombre corriô de la playa en la casa the man ran from the beach in the house These examples show the special treatment of boundary crossings in Spanish, and Slobin and Hoiting provide analogous examples for French, Turkish, Japanese, and Korean. According to their data, the same is true for SLN: They found sentences such as (glossed in En glish) house, man run approach enter, in which the boundary crossing into the house must be marked by a separate verb. They cite this analogy between SLN and spoken verb-framed languages to support their claim that SLN is also a verb-framed language. As we shall see, ASL has this means and several other resources for encoding This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms i82 I Sign Language Studies boundary crossings; our Discussion section includes a reanalysis of the boundary-crossing data and a réévaluation of the claim for cross linguistic analogies between SLN, ASL, and spoken verb-framed lan guages. In the ongoing debate over encoding patterns in signed languages, we saw a need for corpus-based studies. We chose to look at encod ing patterns in ten signers' descriptions of a single complex event. In this event, the signers repeatedly used the cl:v animate classifier to describe a complex "fall," involving multiple simultaneous Paths and Figures. Conceptually, elements of Motion, Path, Manner, Figure, and Ground were present, and the Ground elements included a boundary crossing. The signers' representations of this event could tell us whether a constraint exists on how much information a single cl:v form can encode—either an upper limit on number of concep tual elements or restrictions on what types of information can co occur in one form. They could also tell us whether boundary cross ings are given special treatment in ASL. Methods The children's storybook. Frog, Where Are You? (Mayer 1969) was used as a prompt to elicit narratives from ten adults. This book con tains pictures only, without accompanying text. The pictures present a set of conceptual elements that narrators may choose to include, emphasize, or ignore when retelling the story. Ten deaf, native users of ASL were videotaped signing the story. Study One scene, involving a boy, a dog, and a deer, was coded for this analysis. Pictures iob-i2a of the book depict this scene: The deer has run to a cliff with the boy on its head and the dog alongside barking at the deer. In our focused event, the deer stops, and the boy and dog fall over the cliff and land in a pond. This scene was chosen because elements of Figure, Motion, Path, Manner, and Ground were all present and could all be included in the narration if desired; in addition, the characters' affect was clearly visible from the pictures. We investigated the separation and repetition patterns associated with the classifier V handshape (cl:v) in the primary event of our This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms Patterns of Conceptual Encoding | 183 scene, the boy and the dog's fall into the water. This fall is com plex, involving two Figures and simultaneous Paths from the deer's head (for the boy) and from the top of the cliff (for the dog), over the edge of the cliff down, and into the pond. Logically, this event could have been encoded in one complex form using both hands and the signer's head: One hand using a cl:v (animate classifier for the boy) could have started from the forehead (representing the deer), moved downward in an arc and stopped with a bounce, while the other hand, also using a CL:v (animate classifier for the dog) would have started lower and to the side of the neutral signing space (repre senting the top of the cliff) and moved in a parallel path also ending with a bounce. Only I of the 37 forms produced by our native sign ing subjects did this (figure 3). Most forms were far simpler. More over, every narrator produced at least three cl:v forms, each focusing on different aspects of the scene. This suggests the existence of lin guistic or perhaps cognitive constraints on simultaneous encoding of information. To give a flavor of our data, we have placed a gloss of one version of the scene in Appendix I. This signer used three c;l:v forms to describe the boy and the dog's fall into the pond, and these forms are shown in figures i, 2, and 3. Footnotes in the transcript indicate which glosses correspond to these figures. Figure I illustrates the signer's first cl:v construction. Here the signer's right hand forms the cl:v, representing the boy, and his left hand forms the B-shaped flat-object classifier (cl:b), representing the deer's head. The cl:v starts on top of the cl:b and then moves off and downward in an arc path, representing the initial location of the boy on the deer's head and his subsequent path over the cliff's edge and down. Simultaneously, the signer's referential shift and the scared expression on his face indicate the boy's distressed emotional state. Thus, this form presents Figure, Path, Ground, Manner, and Affect information about the boy's fall, plus his spatial relationship with the deer. In figure 2, the second cl:v construction, the signer uses the V handshape on both hands to represent an animate creature with four legs; contextual information lets us deduce that this classifier refers to the dog. The signer moves his hands on a path that arcs forward and This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms V* Figure i. c:l:vb "animate falls otF" Figure 2. cl:vv "four-legged animate falls' A. Figure 3 . cl:v "animate falls from head" CL'.vv "two animates intertumble" Cl:v "animate falls from cliff" This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms Patterns of Conceptual Encoding | 185 down, showing the path of the dog over the edge of the cliff. At the same time, his facial expression shows the carelessness of the dog in not noticing the cliff. This form therefore includes Figure, Path, Manner, and Affect information about the dog's fall. Finally, in figure 3, the signer puts all this information together. He forms cl:v shapes with both hands, representing the boy and the dog. At the beginning of the construction, the left-hand cl:v is held at a midlevel position in signing space, identified with the dog's posi tion on the ground. The right-hand cl:v is held at the signer's fore head. In this construction, the signer's head represents the head of the deer, and thus the cl:v at the signer's head represents the boy on the deer's head. From these initial positions, both cl:v handshapes move through signing space: First the left cl:v rises in an arc, and then the two cl:v handshapes move downward together, revolving around a common axis, so that the left cl:v ends up on top of the right one. These movements represent the simultaneous falls of the boy and the dog over the cliff and their final position with the dog on the boy's stomach. Thus, this form presents Figure, Path, Ground, and Manner information about the boy's fall; Figure, Path, and Man ner information about the dog's fall; and spatial relationships among the boy, dog, and deer. The first two forms are typical of our data in that they present over lapping partial accounts of the same scene: Although figures i and 2 each focus in on a different character, they contain similar information about the path over the cliff, and most of their information is repeated in figure 3. Figure 3, however, is not typical of our data: It is the most complex form produced by any of our subjects and contains nearly all the conceptual information given in the pictorial prompts. Yet nearly all of this information had just been presented by the signer in his two previous cl:v forms; the only new information is the spatial relationship between the boy and the dog during their falls. The overall patterns of cl:v use are summarized in table i. Our ASL signing narrators all repeated the cl:v an average of 3.7 times with a range of 3 to 5 instances. These cl:v handshapes were com bined and recombined with different additional information. Out of 37 instances of the cl:v in the "fall," only 1 could be considered an exact repetition of a preceding form (and this person did use 3 dis tinct cl:v forms). This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms 186 I Sign Language Studies Table i. Encoding of Conceptual Elements with the cl:v form Conceptual Categories Scene Elements Form Encoded Path Information ii Start Motion + Path + Animate fall Figure (boy or dog) Arc V cl:v Down End X 6 X X 2 X 1 Animate fall 4-legged animal) cl:vv Motion + Path + dual Figure Dual animate fall cl:v + cl:v X X X 2 (dog: X X 2 X X 1 X (boy and dog) 1 2 Motion + Path + Figure 4- Ground Fall from head of deer (boy) cl:v + 1 X X X X X X X forehead (initial landmark) cl:v + cl:b 5 X X X 3 X X X 2 X X + cl:b 2 (dog) X X Fall fromc:l:v cliff Motion + 2 Paths Fall from deer + 2 Figure + (boy) and fall Ground (initial landmark) Motion + Path + Figure + Ground (bounday cl:v + cl:v 1 cl:v + cl:b 5 X X X X X X from cliff (dog) animate enters water (boy or dog) crossing) *The arc in this item may have represented continuous aspect rather than arc Path. Some separation patterns emerged from the data. All signers' cl:v forms (with one possible exception; see the note to table i) included Path information. The signers divided the Path into three pieces: the initial arc (denoting a loss of balance and fall), a downward segment, and a final bounce. We saw cl:v forms that incorporated the entire Path (4), the arc and the downward segment (21), the downward segment and the bounce (2), the arc only (2), and the downward This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms X Patterns of Conceptual Encoding | 187 segment only (8). The initial arc was included in 27 forms, while the bounce was included in 6; moreover, no forms included only the bounce. This suggests a tendency to focus on the initial part of the event rather than the end; also, awareness of the lexicalized sign meaning "fall" (derived from a CL:v with an arc path) may have influenced signers to include the arc more frequently. We can divide the cl:v forms into four groups based on the con ceptual elements that combined with Path: Motion + Figure, show ing an animate entity falling (12); Motion + dual Figure, showing the boy and the dog falling together (7); Motion + Figure + Ground (initial landmark), showing the boy falling from the deer's head or the dog falling from the cliff (13); and Motion + Figure + Ground (boundary crossing), showing an animate entity entering the water (5). Additionally, there was the previously described form that included separate Paths, Figures, and Motions for the boy and dog, along with the Ground (initial landmark) of the deer's head (see fig ure 3). If there is an upper limit on the amount of conceptual infor mation in a cl:v form, we might expect that the second, third, and fourth groups and the complex form would accommodate Paths that are less complex than the first group's paths. The first group, containing the forms with the fewest additional elements (Motion + Figure), could combine with Paths of any de gree of complexity. We should note that two of the forms in this category actually used both hands to represent all four legs of the dog, as illustrated in figure 2. The second group of forms added a second cl:v to represent the second Figure. In all of these cases, the two hands moved symmetri cally, either in parallel (for the fall) or in alternation (for a few depic tions of the bounce). Because of this, we consider these forms to have a dual Figure (the boy and dog "chunked" together) tracing a single Path. The third group of forms incorporated a Ground element repre senting the initial landmark. In six cases, the signer's head was used to represent the deer's head; in five cases, the nondominant B hand shape (cl:b) represented the deer's head (see figure i); and in five cases, the cl:b represented the cliff. Table i marks the use of this initial landmark in the column labeled "start." This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms 188 I Sign Language Studies In considering the second and third groups of cl:v fo found that adding the initial landmark or the complex du did not affect the allowable complexity of the form's Path. initial-landmark and the dual-figure forms incorporated one all three Path elements. A possible exception to this generaliz the group of forms in which the signer's head was used for t landmark: None of these six forms included the final bounce. A ref erential shift is needed to define the signer's head as the deer's head, and adding the final bounce to that form may have made the form too complex for most signers. In the fourth group of forms, the nondominant cl:b represented a Ground element, the surface of the water, and the cl:v was moved downward past the cl:b. All five of these forms incorporated only the downward Path segment. No articulatory constraint limits the Path of the cl:v; presumably it could move in an initial arc down past the water to a final bounce, but this never happened. This is a striking difference from the initial-landmark group, which also incorporated a single Ground element yet could include any Path elements. It suggests that the number of conceptual elements is not the only factor in determining separation patterns and that boundary crossings may be in some way special in ASL. Finally, the single most complex form is illustrated in figure 3. It incorporates arc and downward Path elements for two separate Fig ures, plus an initial Ground for one Figure, and an ending configura tion showing a complex Spatial Relationship between the Figures. We note that before producing this form, the signer had already pro duced two simpler cl:v forms (figures 1 and 2); nearly all the infor mation in the complex form had already been presented. This is consistent with a cognitive constraint on level of complexity: This much complexity might be comprehensible only when the addressee has already had a chance to assimilate it in smaller chunks. To summarize, all signers used at least three different cl:v forms to describe the fall, showing that some degree of conceptual separation is necessary for this complex event. Also, all signers re peated Figure, some Path, and other information in their different forms, showing the importance of repetition in building up a coher ent picture of the event. Nearly all the cl:v forms consisted of This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms Patterns of Conceptual Encoding | 189 Motion, Figure, Path, and one possible additional element (Ground element or second Figure chunked with the first). We saw only one form of greater complexity, which followed (and summarized) two simpler forms. This suggests that there is a strong tendency to limit the amount of conceptual information that cl:v forms can express. But complexity may not be the only factor, as we see from the con trast in Path complexity between forms with initial landmarks and forms with boundary crossings. Study 2 We were intrigued by the apparent constraint that boundary crossings could co-occur with only straight-line paths. To further investigate this hypothesis, we examined all other boundary crossings in our sample of ten ASL narratives. Our definition of "boundary crossing" included both literal crossings (e.g., boy puts frog in jar) and metaphor ical ones (e.g., boy looks into hole in tree)', we did this because languages often use the same resources for describing literal and metaphorical motion (cf. Lakoff and Turner 1989). There were a total of 19 literal and 5 metaphorical crossing events, with altogether 89 descriptions of crossings. The data are summarized in table 2. The first column in table 2 gives the total number of descriptions for each crossing; clearly, each signer did not describe each crossing once and only once. The next column shows the number of crossings indi cated by prepositions such as in and out (5); no further Path informa tion was given in these sentences. The next two columns track lexical verbs such as disappear and enter (ii), which all inflect for straight paths, and those classifier forms that incorporated straight paths ($3). Together, these three types account for 78 percent of all crossings. The last three columns tabulate classifier forms with more complex paths: first, paths consisting of simple arcs (12); second, paths shaped like an inverted V, with a straight upward and then a straight downward seg ment (3); and last, truly complex paths involving shakes, wiggles, or bounces (5). These types account for 22 percent of all crossings. We may note that complex paths co-occurred with only six cross ings: the frog climbing out of the jar; the dog's head entering the jar; the dog falling out of the window; the boy jumping out of the win dow; the boy and dog looking over the log; and the boy and dog This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms <D O u X <D VerbsClasifersClasifersClasifersClasifers o o TotalPreposit nsLexicalStrightArcVPathComplex fO M '—1 CO M M M '—' (N ^ CN '—^ if; 1/1 M vD t—i t-h £ v „ i_i 5P w i> ^ ° cola S ^ ^ u-2 o g ^ S , en c a 22 .2 c ^ '-« > 0 23 n n Prt'CC-S.2 01 r a ^ k u S 0 -c o £ ° m ^ 3 u u Sn,"1-E ^Si!,'''^<J,JnC,S U o U O C u> 0 &_g 'g 1 V3 £ J o § <3 £ fa S-S J w O 5 '2 > S o 3 o S o » s 3 o o | a ^ 2S^||s § TP__, ca<£"-,£> S ^ og.sSs°|^^ o q. ^ ^ Z, c 0 S J oo °§rt§_ .S "rL .S ^ s s Subtoal7649 135 £ E. §j~ -^ ■£ M_c jc u «>« -" « « J —1 ooPP§oooooOoo<d2^"ooC/3-o^222 Table2.BoundaryCrosingsintheNar tives 1.boyputsfroginjar413 2.boygesintobed3 3.frogclimbsout fjar13 43 5.sunriseabovehorizn1 6.dog'shead ntersja 18 431 7.dogfalsout fwindo 21 8.boyjumpsout fwindow2 9.boy/dog o utside2 1.frog oesintohle2 12.gopher ntershole1 13.gophercomesout fhole9 14.bescome utofhive2 15.frog oesinto re hole1 16.owlfiesoutfrehole523 17.der'sheadrise pastrock1 18.boyand ogfalintowater523 19.boyand ogclimboverlog2 « jo _o tic <-ic S -a -a x> -o x <J3 scmjsUc o -a x _o n -0 -o x x x <D ^ .... 4.boyloksintohleintre 514 2.dog/boylokoutwindow2 . 4.frogleavs/disap ears2 Visual(metaphorical)Cros ings n m t m rHMmt^'Ohoooo^rjcn'tLn^h-^ON ^ ^ 10.boy/dogentrforest1 Literal Cro.■£ s ings 1.dog/boylokintojar1 3.boyloksintogrundhole3 5.boy/dglok verlog21 This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms Subtoal13 291 Total8951 32 5 Patterns of Conceptual Encoding | 191 climbing over the log. These crossings—and only these crossings— share the characteristic that the boundary is an impediment to for ward motion, and the Figure must modify its path to cross it. (The dog's head entering the jar is a seeming exception, but all but one of the four complex Path forms for that crossing actually involve a Fig ure/Ground reversal, in which the jar is depicted as moving onto the dog's head. If one imagines that scene, one will envision the jar mov ing upward onto the head, and then the rim moving downward past the head; the head forces modification of the jar's path. The final jar/ head form is highly complex and deserves further analysis.) We can immediately make the following observations. First, boundary crossings may be encoded via a variety of means, including classifiers, lexical verbs, and prepositions. Second, there is a strong tendency toward conflating boundary crossings with simple, straight line paths, but only a tendency, as nearly one-fourth of the classifier forms had more complex paths. This has more the appearance of a cognitive heuristic than of an exceptionless rule. Discussion and Conclusions Our results support Supalla's claim that ASL signers do in fact sepa rate a unitary event into different conceptual pieces. Although the linguistic resources for describing motion events are highly iconic, this is not a simple, directly depictive kind of iconicity. Instead, the event is decomposed into pieces that can be presented separately, and there is a strong tendency to limit the amount of conceptual information that appears in each iconic form. It does not seem, how ever, that there is an absolute upper bound on the amount of infor mation. This suggests that many of the separation patterns in ASL stem from limitations on cognitive processing rather than on linguis tic co-occurrence restrictions. Some evidence exists that seems at first glance to support Slobin and Hoiting's view that boundary crossings are special. In our se lected scene, complex paths never appeared in simultaneous con struction with boundary crossings, whereas they freely co-occurred with representations of other landmarks. Yet complex paths did co occur with other boundary crossings in the narratives, notably those in which the boundary impeded the Figure's motion. We propose This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms 192 I Sign Language Studies that this differential treatment of impeded and nonimpeded boundary crossings is not due to arbitrary co-occurrence rules but instead to cognitive factors. Presumably signers have mental images or models of the entire motion event they wish to describe. Let us hypothesize that for boundary crossings they focus their attention on a small piece of the image immediately surrounding the crossing (cf. Langacker's notion of scope [1987]). For nonimpeded crossings (e.g., the boy's fall into water), the immediate neighborhood of the crossing contains only a straight-line path and no effort on the Figure's part. But for impeded crossings (e.g., the frog's escape from ajar), the neighborhood con tains an arc path and potentially complex maneuvering by the Figure. If signers encode what they construe as being in the crossing's neigh borhood, they will create forms with straight paths for nonimpeded crossings and complex paths for impeded ones. On this analysis, boundary crossings are not given special treat ment in ASL. When one is to be mentioned, the signer focuses on it, and it is encoded with a complex or a simple Path depending on the Path details right at the boundary. But that is not significantly different from how any other conceptual element gets included in the narrative. Moreover, even this focus-based encoding is a ten dency rather than a rule: In a few cases, the signer includes Path elements that are not right at the crossing (e.g., the frog leaping out of the jar and bouncing as it hits the floor). The result supports the claim that spatial descriptions in ASL are constrained and structured by cognitive factors (assumed to be heuristic and flexible) rather than linguistic rules (assumed to be arbitrary and exceptionless). Where, then, does ASL fit in Talmy's typology? Is ASL in fact verb-framed, as Slobin and Hoiting claim, satellite-framed and man ner-type, as Supalla's work suggests, or something else? In searching for an answer, let us look more closely at the treatment of Path, Manner, and boundary crossings in these types of languages. What does it mean to say that verb-framed languages give special treatment to boundary crossings? This analysis draws on Aske (1989). Satellite-framed languages such as English typically have a Manner verb plus one or more Path segments in satellites. In these satellites, This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms Patterns of Conceptual Encoding | 193 boundary crossings are treated on a par with other Path segments in that they can appear as just another segment in a complex Path (cf. The man walked out of the house, across the street, through the park, and into the church.). In another pattern, a Path segment can appear in the main verb; once again, boundary crossings receive equal treatment (e.g., cross, enter, and exit vs. rise, descend, and traverse). Verb-framed languages (e.g., the Romance languages) also have two pertinent patterns. In the first, the main verb encodes a Path segment in the verb; in this case (as in the second satellite-framed pattern), boundary crossings are treated just the same as any other Path segment (e.g., French entrer 'enter,' sortir 'exit' vs. descendre 'de scend,' monter 'rise'). In the second, the sentence may have manner information in the main verb and location in a satellite. This location satellite cannot represent a boundary crossing (cf. example 2). Thus, in verb-framed languages, boundary crossings get expressed only in verbs, and each requires its own. A hypothesis that involves the conceptualization of Manner in each language will account for these facts. Let us suppose that Man ner verbs in satellite-framed languages include the notion of travel through space along with their specification of manner of locomo tion. These verbs will then be path compatible: They allow a specifica tion of a Path along with them (in their satellites). The verbs do not impose any limits on the character of that Path; it can be of any length and involve complexities such as boundary crossings (Aske [1989] referred to such Path specifications as telic path phrases). Let us suppose, on the other hand, that Manner verbs in verb framed languages do not include the notion of travel through space; they simply specify an action that the Figure is taking (i.e., some limb-movement type). These verbs will not be path compatible. In their satellites, they allow a specification only of location (not Path)— the location where the action is taking place (e.g., Spanish en la casa 'in the house'); Aske (ibid.) referred to such specifications as locative path phrases. We will have to assume that the location can include some notion of direction, as seen in example 2c (e.g., Spanish correr hasta la casa 'run up to the house'). These locations cannot, however, include a boundary crossing: Once a boundary has been crossed, the This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms 194 i Sign Language Studies space is construed as a different location, and motion from one loca tion to another necessarily constitutes a Path. Thus, the only way to express a boundary crossing in these languages is to use a Path verb that encodes the crossing. This hypothesis matches the facts described earlier. Now let us compare the treatment of Manner in ASL. As Supalla and Slobin and Hoiting observe, lexical Manner verbs (e.g., run) often appear in construction with Path classifiers or Path verbs, and these Path forms can encode boundary crossings. Also, as noted ear lier, the same classifiers that typically express Manner can also include Path (in contradiction to Supalla). Therefore, manner in ASL is path compatible and also compatible with boundary crossings. This is not the same pattern as is observed in spoken verb-framed languages. We may conclude that ASL is not a verb-framed language. Whether it is a manner-type or figure-type satellite-framed language—or something else altogether—remains to be determined. We may speculate that, contrary to Talmy (1991), the treatment of Manner rather than Path is the primary motivator for language typology. Manner that is incompatible with Path would force lan guages to be verb-framed for the following reason: Path elements could not be encoded as satellites to (path-free) Manner verbs and would necessarily be encoded in verbs; this is the definition of verb framed. If treatment of Path is not the core criterion, then the bare fact that most ASL verbs of motion contain paths would not force us to consider ASL to be verb-framed. Last of all, let us return to Supalla's observation that ASL classifiers tend to encode either Manner or Path, but not both, and his claim that this is the result of arbitrary linguistic restrictions on encoding. Some of the patterns that Supalla noted are conventional facts about ASL—for example, the fact that signers use extended index fingers rather than index and middle fingers to represent a referent's legs— and thus are arbitrary to some degree. But the overall pattern he observed of a separation between Manner of locomotion and Path may not be arbitrary at all. McNeill (1999) reported on a study of gestures that accompany spoken-language descriptions of motion events; his study involved This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms Patterns of Conceptual Encoding | 195 speakers of English, Spanish, and Mandarin, from age 3 up to age 12. His earlier work (McNeill 1992) showed that these gestures are tightly bound to the spoken message they accompany and that much conceptual information may be expressed through these gestures; in particular, gestures may carry (at least) Path, Manner, and Figure in formation. The recent study shows that starting at age 3, speakers of all three languages separate Manner and Path information in their gestures. That is, the children never produce gestures that include both Manner and Path information. Adult speakers, however, regu larly produce such combinations in their gestures. McNeill suggests that there may be a limit on children's ability to represent motion events; even though they may recognize the Manner and the Path of the motion events, they cannot yet recombine them into a single gestural portrayal. Given these results, we suggest that the locomotion/path split is a cognitive universal, not a language-specific one, and that the patterns of motion verbs in ASL derive from the same cognitive structures as spoken-language gestures. The fact that most of these patterns seem to be tendencies rather than absolute rules also supports a cognitive rather than a linguistic origin. This proposal could reconcile Supalla and Slobin and Hoiting's observations with Liddell's (1995) analysis of the use of space in ASL as "gestural." The discretization of the representation of space and movement in ASL would not then derive from separate "mor phemes" in the traditional sense but separate cognitive representa tions. The logical next step is to link these results to separate cognitive/perceptual structures in the brain that recognize and repre sent different aspects of motion events and other structures that inte grate those representations. Maintaining the locomotion/path separation in gesture and lan guage could serve a number of functions. First, it could be conceptu ally easier to produce expressions that match one's internal representation of the event. If the process of perception requires this split, then it might be simpler to match that structure rather than to recombine the pieces into a single representation. This would fit with McNeill's observation that combined Manner/Path gestures develop This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms 196 I Sign Language Studies after age 12, when a child is cognitively ready to reintegrate these elements. Second, for similar reasons, separated representations might be easier for an addressee to understand. Why create a complex, fully integrated classifier when the addressee must break it down again in the comprehension process? Presenting a separated, "predigested" form instead would allow the addressee to gradually build up a model of the event from simpler conceptual pieces. In addition, the repeti tion of old information with new information would help the ad dressee reassemble the separated pieces into a coherent whole. The serial and sequential constructions noted for ASL may in fact be bet ter suited to communication than a single mimetic portrayal of all the aspects of a complex event. This hypothesis is supported by the fact that our only example of a highly complex form follows and repeats the information in two simpler forms. Finally, separating out pieces of information allows the expresser to direct the addressee's attention to particular pieces. As Talmy (1985) noted, information conflated into one lexical item is "back grounded" and not available for focus; compare I flew to Hawaii and I went by plane to Hawaii—in the second, the mode of transport is available for emphasis. Thus, if signers and gesturers can present par ticular aspects of the scene in isolation, they can emphasize those aspects. To sum up, ASL encodes a great deal of conceptual information about motion events, significantly more than English and presumably more than most other spoken languages. This information is largely presented through referential shift and iconic classifier forms. Yet the iconic forms do not usually give a fully simultaneous portrayal of all aspects of complex motion events. Instead, many aspects (Manner of locomotion, details of Path, different Figures and Ground elements) are routinely separated into different pieces and presented sequen tially; moreover, some aspects are presented repeatedly to provide coherence to the separated presentation. There are general tenden cies to limit the amount of information in any given form and to split up certain types of information, but the signer may override them. This type of separation occurs both in signed languages and in This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms Patterns of Conceptual Encoding | 197 gestures accompanying spoken languages; it may be a consequence of our cognitive and perceptual resources for understanding motion events rather than of arbitrary rules specific to linguistic stmcture. Appendix I ASL transcription of one scene from Frog, Where Are You? Adult Male, Pictures iob, n, and 12a Right Hand: CL:v (on top of head) "animate on head" Both Hands: Left Hand: cl:y (on side of forehead) "antlers" Perspective: narrator —» Translation: The hoy is on the deer's head. Right Hand: cl:5 (to side of head) "deer" Both Hands: cl:vv "4-legged animate is running" Left Hand: Perspective: narrator —> | deer —» Translation: The deer is running. Right Hand: boy cl:v (on head) "animate on head" stuck Both Hands: cl:vv "4-legged animate is running" Left Hand: Perspective: narr. | boy (scared) —► Translation: The scared hoy is stuck on the running deer's head. Right Hand: dog says get-doWNjmper get-do WNimp„ Both Hands: bark (dir: up (dir: up and side) and to side) Left Hand: Perspective: narrator | dog | narrator | dog (worried about boy) —> Translation: The concerned dog barks up at the boy and says "get down, get down. " Right Hand: deer Both Hands: cl:vv "4-legged animate is running" Left Hand: Perspective: narrator | deer -> Translation: The deer is running This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms 198 I Sign Language Studies Right Hand: Both Hands: to land cl:bb "flat land to cliff' Left Hand: Perspective: narrator —> Translation: to the cliff. Right Hand: Both Hands: cl:vv (same location as the cliff) "4-legged animate stops at cliff" stop Left Hand: Perspective: deer (scared) —* Translation: The scared deer screeches to a halt at the edge of the cliff. Right Hand: boy Both Hands: cl:vb (same location ofthe deer's stop and cliff) "animate falls off"1 Left Hand: Perspective: narrator | boy (scared) —» Translation: The scared boy falls off the deer's head and over the cliff. Right Hand: dog Both Hands: with cl:vv (at cliff location) "4-legged animate falls" Left Hand: Perspective: narrator | dog —* Translation: With the dog, the dog falls, Right Hand: doesn't-notice Both Hands: cl:vv (at cliff space) "4-legged animate falls'" Left Hand: Perspective: | narrator | dog —> Translation: without noticing (the cliff) the dog falls. Right Hand: Cl:v (at head) "animate falls from head" —► Both Hands: cl:vv "two animates intertumble"3 Left Hand: cl:v (at cliff location) Perspective: boy (scared) —* Translation: the As two the of boy them falls from the intertumbling 1. Figure l represents this form. 2. Figure 2 represents this form. 3. Figure 3 represents this entire This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms "an head until line. Patterns of Conceptual Encoding | 199 Appendix 2 ASL Transcription Conventions The glosses of ASL in Appendix I consist of five lines. The top three lines describe what the signer's hands are doing: The first, labeled "Right Hand," shows the right hand when it acts alone; the second, labeled "Both Hands," shows what the hands do together; and the third, labeled "Left Hand," shows what the left hand does on its own. The fourth line gives the persona assumed by the signer: narrator, boy, deer, or dog. The fifth line shows the English translation of the ASL sentence. Vertical lines indicate the end of one persona and the assumption of a new persona. Personas' affect is some times described in parentheses. Lexical items are glossed by English words in small capital letters (e.g., dog). A dash joining two English words indicates a multiword gloss of a lexical sign (e.g., doesn't-notice). The subscript "imper" indicates that the verb has the imperative inflection. Classifier glosses consist of the letters cl, followed by an abbreviation for the classifier handshape, followed by a description of the classifier's meaning. For some classifiers, a parenthetical note gives additional place ment information. For example, cl:y (on side of forehead) "antlers" denotes a Y-shaped classifier placed at the side of the signer's forehead to indicate antlers. Classifier handshapes appearing in this transcript include the following: V—index and middle fingers extended from a closed fist; fingers may be curved or straight Y—thumb and pinky finger extended from a closed fist 5—all fingers and thumb extended and spread B—all fingers extended and touching; thumb may touch fingers or be extended outward References Aske, J. 1989. Path Predicates in English and Spanish: A Closer Look. In Proceedings of the Fifteenth Annual Meeting of the Berkeley Linguistics Society, i—14. Berkeley: Berkeley Linguistics Society. Engberg-Pedersen, E. 1999. Path- and Ground-Denoting Expressions in Descriptions of Motion Events in Danish Sign Language: A Typological Perspective. Paper presented at the Sixth International Cognitive Lin guistics Conference, July, Stockholm, Sweden. Emmorey, K., and J. S. Reilly. 1995. Theoretical Issues Relating Language, Gesture, and Space: An Overview. In Language, Gesture, and Space, ed. This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms 200 I Sign Language Studies K. Emmorey and J. Reilly. Hillsdale, N.J.: Lawrence Erlbaum ates. Galvan, D., and S. Taub. Forthcoming. The Encoding of Motion Informa tion in American Sign Language. In The Frog Story Revisited, ed. R. Berman, D. Slobin, S. Strömqvist, and L. Verhoeven. Hillsdale, N.J.: Lawrence Erlbaum Associates. LakofF, G., and M. Turner. 1989. More than Cool Reason. Chicago: Univer sity of Chicago Press. Langacker, R. 1987. Foundations of Cognitive Grammar. Vol. I: Theoretical Prerequisites. Stanford: Stanford University Press. Liddell, S. 1995. Real, Surrogate, and Token Space: Grammatical Conse quences in ASL. In Language, Gesture, and Space, ed. K. Emmorey and J. Reilly. Hillsdale, N.J.: Lawrence Erlbaum Associates. Mayer, M. 1969. Frog, Where Are You? New York, Dial Press. McDonald, B. 1982. Aspects of the American Sign Language Predicate Sys tem. Ph.D. diss., State University of New York, Buffalo. McNeill, D. 1992. Hand and Mind: What Gestures Reveal about Thought. Chicago: University of Chicago Press. . 1999. An Ontogenetic Universal and Several Cross-Linguistic Dif ferences in Thinking for Speaking. Paper presented at the Sixth Interna tional Cognitive Linguistics Conference, July, Stockholm, Sweden. Schick, B. 1987. The Acquisition of Classifier Predicates in American Sign Language. Ph.D. diss., Purdue University. Slobin, D. 1996. Two Ways to Travel: Verbs of Motion in English and Spanish. In Essays in Semantics, ed. M. Shibatani and S. A. Thompson. Oxford: Oxford University Press. Slobin, D., and N. Hoiting. 1994. Reference to Movement in Spoken and Signed Languages: Typological Considerations. In Proceedings of the Twentieth Annual Meeting of the Berkeley Linguistics Society. Berkeley: Berkeley Linguistics Society. Supalla, T. 1982. Structure and Acquisition ofVerbs of Motion and Location in American Sign Language. Ph.D. diss., University of California, San Diego. . 1990. Serial Verbs of Motion in ASL. In Theoretical Issues in Sign Language Research, Vol. 1 : Linguistics, ed. S. D. Fischer and P. Siple, 127 52. Chicago: University of Chicago Press. Talmy, L. 1985. Lexicalization Patterns: Semantic Structure in Lexical Forms. In Language Typology and Syntactic Description Vol. 3: Grammatical Categories and the Lexicon, ed T. Shopen, 36-149. Cambridge: Cam bridge University Press. . 1991. Paths to Realization: A Typology of Event Conflation. In Proceedings of the Seventeenth Annual Meeting of the Berkeley Linguistics Soci ety, 480-519. Berkeley: Berkeley Linguistics Society. This content downloaded from 143.50.34.113 on Wed, 20 Mar 2019 20:02:48 UTC All use subject to https://about.jstor.org/terms