Supalla, T. 2003. Revisiting Visual Analogy in ASL Classifier Predicates.

In
Perspectives on Classifier Constructions in Sign Languages, ed. K. Emmorey,
249-258. Mahwah, NJ: Lawrence Erlbaum.

11
Revisiting Visual Analogy in ASL
Classifier Predicates
Ted Supalla
University of Rochester

In the 1970s, American Sign Language (ASL) research had made significant progress in demon-
strating that ASL is not a nonlinguistic system of gesture but a language built on universal gram-
matical principles shared by all human languages. The signs of ASL had been shown to be
composed of the equivalents of phonemes (Stokoe, 1960, Stokoe, Casterline, & Croneberg,
1965), the signs themselves were shown to not be completely iconic (Klima, Bellugi et al.,
1979), and to have evolved in consistent ways over time (Frishberg, 1975, Woodward, 1974,
1976). However, classifier predicates remained a challenge. They were the most highly iconic
signs, whose forms (handshapes, locations, and movements) appeared to vary with the continu-
ous variation in the world event that they represented. To ask a native signer the meaning of a
slight variation in a verb of motion, for example swerving a car to avoid potholes, the signer
could only say that the sign zigzagged to depict that motion in the actual event. And yet a hall-
mark of human languages is that they represent real world events using discrete rather than con-
tinuous forms. Do classifier predicates mark the point where ASL diverges from other human
languages, and lapses into the all-too-available gestural capacity of the visual medium to com-
municate these spatial events?
In this chapter I revisit the issue of classifier predicates, as they pose a significant challenge to
the investigator attempting to discriminate analogical representation from discreteness and
combinatoriality in sign language. I review my own early approach to investigating classifier
predicate constructions, including some description of the evolution of my own understanding
of these issues. I also attempt to break the study of classifier predicates from a single object of
study into a variety of issues: “grammatical” versus “ungrammatical” uses, by a variety of popu-
lations (native vs. nonnative signers, children vs. adults), for a variety of pragmatic purposes.
Each of these different situations offers a valuable perspective and a greater understanding of
whether these forms are formed from discrete components or whether they represent their refer-
ents analogically.

249
250 SUPALLA

In my dissertation and subsequent work (Supalla, 1978, 1982, 1986, 1990), I studied one sub-
class of classifier predicates, ASL verbs of motion and location, in order to confront the issue of vi-
sual analogy in sign language. I concluded that they conform to the same grammatical constraints
found in other languages: they are composed of discrete morphemes, which recombine and have
constraints on order of combination, number of morphemes, etc., similar to spoken language clas-
sifiers. However, the visual–analogic properties of these verbs are being questioned again in the
current sign language research community. I would like to revisit details of the methodology and
findings of these earlier studies and bring this information to the current debate, as I believe that
similar kinds of methodology are required for addressing these new and old controversies.
A significant challenge in research on verbs of motion concerns how to collect data that can
determine whether one view of their structure is empirically more accurate than another. In
1977, DeMatteo suggested that verbs of motion in ASL were depictions of events of motion,
varying continuously as did their referent events. More recently, Liddell (chap. 9, this volume)
and others have presented a similar view of motion and location in such verbs. In contrast, Cogill
(2000) suggested that ASL and nonlinguistic gesture are similar in the opposite way, that is, both
display discrete or schematic forms. Unfortunately, discussions of these views sometimes hinge
on whether constructions look analogue, numerous, or schematic to the investigator. A crucial
question concerns how to collect empirical data that can address and distinguish these views
more convincingly. This problem is not resolved by asking native signers whether forms can
vary. It is part of the culture of the American deaf community that deaf people’s folk etymology
relies heavily on explanations of the iconic roots of signs and parts of signs. Asking a signer why
one verb of motion with an upward trajectory uses a straight path while another uses a curvy path
will generally result in an explanation about the event pictured.
In my dissertation on verbs of motion and location, I used my linguistic intuitions about these
verbs and performed a linguistic analysis using the usual methods of contrastive forms. But I
also needed to confirm both the forms and the analysis with data from other signers. I created
brief animated film clips, “vignettes” of different toy objects moving through space. Subjects
viewed these stimuli, and after each one produced a sentence describing what they observed.
In order to verify the regularity claimed in my linguistic analysis, I asked whether native sign-
ers would produce the same linguistic forms regularly and consistently, even when the referent
event varied in its details. The animated, filmed vignettes each showed an object moving along a
certain trajectory path, or moving in relation to a stationary prop, such as a toy cow jumping over
a fence or a toy bird swooping upside down. The objects in motion, their trajectories, and place-
ments relative to other objects were manipulated, frame by frame, to elicit variants of a specific
verb, with particular handshapes, movements, and locations. For example, to analyze the sub-
jects’ selection of verbal classifier handshapes to represent the items involved in the event, I
carefully selected a range of props that should elicit a particular handshape. Each prop was used
only once in the test paradigm, so that I could interpret the repeated use of the same handshape
for different props, by a single subject or across subjects, as evidence for one form representing a
class of noun referents. Similarly, I produced a range of motion paths and motion manners, in or-
der to be able to see whether subjects would use the same path or the same manner across a class
of referent motions. Finally, over the length of the test, paths and manners were each produced
by a variety of objects, so that I could see whether the forms for motions and those for objects
combined orthogonally.
At the same time, Elissa Newport and I, along with graduate student Carol Schwartz,
(Schwartz, 1979), designed a different type of study to ask whether signers used handshapes to
 11. REVISITING VISUAL ANALOGY 251

represent a class of objects, or whether they varied their handshapes continuously and analogi-
cally to represent an object’s properties. In this study, we showed a group of signers a varying set
of objects (e.g., dots of varying sizes), and then asked the signers to describe the objects. The tar-
get objects were mixed with fillers with other characteristics to avoid forcing the signers to dis-
tinguish fine gradations of size if they would not ordinarily do so. (As in spoken languages, of
course, it is always possible to describe fine gradations, by special or periphrastic constructions.
Our question, however, was directed to ordinary ASL constructions, particularly those involv-
ing single signs.) These productions were then filmed and shown subsequently to another group
of subjects, whose job was to identify from an array of dots the precise one the signer had been
describing. If signs varied analogically, we reasoned, then the observer’s choices should show a
correlation with the dot size actually seen by the signer. We found, however, that only categories
of size were encoded by the signers; within a category, the observers’ choices were at chance.
This result shows that these dimensions of handshape are apparently not encoding continuous or
analogic variation, but only categories (see Emmorey & Herzig, chap. 10, this volume, for a re-
cent use of this same procedure, with similar results).
The same result appeared in subjects’ responses to my vignettes. Given different motions
across a movement or manner category, subjects produced the same path or manner in their sign-
ing. Locative relations between the moving object and the stationary object were also limited
and categorical. My linguistic analysis suggested that the ASL system of locative relations ob-
serves a “base grid,” in which secondary objects can only be marked at a few contrasting points
along the path of the moving object (e.g., at its origin, its midpoint, or its endpoint). Although vi-
gnette events had more continuous variation in the placement of secondary objects, signers pro-
duced only a few significant contrasts. Similarly, given different props across a semantic or size
category, subjects produced the same handshape repeatedly.
I also found that the handshapes the subjects used to represent the different props had several
different types of semantic and phonological properties. Some handshapes (which we later
called SASS classifiers), were composed of components that were used in other handshapes. For
example, the size-and-shape classifiers G, H, and flat-hand shape are three distinct classifiers
that represent relatively flat objects of varying widths. (Note that ASL does not allow the addi-
tion of a third finger to represent widths intermediate between narrow and wide—it uses only
these three classifiers to represent all possible widths. Other sign languages may use only two
such classifiers: narrow and wide.) They share flat shape (straight fingers) with one another, but
vary in the width of the object (marked by number of fingers—one, two, or four). A related set of
SASS classifiers, with baby C-, hC-, and C-handshapes, represent cylindrical objects of the
same widths (again, marked by number of fingers—one, two, or four). In short, then, these
size-and-shape classifiers represent classes of objects based on their physical dimensions, but
are composed of internal morphemes of form (e.g., straight vs. curved, and narrow vs. medium
vs. wide). These classifiers also share a set of constraints on how they may combine with certain
movement and orientation features. In contrast, other handshapes, like the “vehicle classifier”
(thumb, forefinger and second finger extended), are used to represent more abstract classes.
These semantic classifiers have more frozen handshapes; the subparts of the handshape cannot
be altered to represent variations in the object, such as a car with a broken door, and they cannot
be analyzed into subcomponent morphemes. Furthermore, they have a different set of con-
straints than do SASS on the movement and locative morphemes with which they can combine.
I interpreted these phonological and semantic differences as evidence for two different types of
classifiers, called SASS classifiers and semantic classifiers.
252 SUPALLA

Creating the vignettes as elicitation stimuli allowed me to constrain and focus the types of ob-
jects and motions under study. Real-world events have so many attendant and varying aspects,
subjects could choose to describe them in a great variety of ways, and this variety could be quite
separate from whether there was consistent grammatical structure for verbs of motion in the lan-
guage. Even in this study, I created some events with too many details. For example, I filmed a
doll that walked by moving her legs in a lifelike manner and swaying the body. I found that sub-
jects responded using a variety of descriptions, focusing on different aspects of the event. Some
subjects would respond with verbs that used the whole body (body classifiers) to represent the
manner of walking, whereas others would use a handshape classifier that represented the doll’s
body along with path and manner morphemes; and some would use both kinds of verbs, in se-
quence. These responses showed me that verbs of motion are composed of two different types:
verbs that used classifier handshapes for the nouns involved and could combine these with man-
ner and path or location morphemes, versus verbs that used body classifiers and marked the body
motions involved (but not the path). The responses to the highly detailed stimuli also provided
evidence that signers could not encode all these aspects of an event by adding an unlimited num-
ber of morphemes to a single verb. Rather, verbs of motion are constrained as to the number and
type of morphemes they could accept. If more details must be expressed, they must be encoded
in a second verb, with different types of verbs able to carry different types of morphemes. This is
how I discovered serial verbs (Supalla, 1990).
A goal of my dissertation was also to use these filmed vignettes to study several native sign-
ing deaf children, to see whether their responses would deviate from or omit any of the mor-
phemes used in adult native signer’s responses. For this study, I created 120 vignettes (later
reduced to 80 vignettes as part of a larger battery of tests of ASL grammar), which I designed to
elicit only a single verb, in order to focus on complex movements, locative relations, and types
of classifiers. The analysis of the children’s correct selection, omission, or replacement of
handshape, movement, and location morphemes demonstrated that different parts of the classifi-
catory morphology are acquired independently, which provided evidence of the psychological
reality of discreteness and componentiality in classifier predicate constructions, particularly in
the verbs of motion and location.
Another advantage of this methodology is that it allows repeating the same procedure with
different individuals with the context information under full control. We could therefore guaran-
tee that each subject would be asked for the same propositions because the same vignettes were
presented to all subjects. Eventually Elissa Newport and I realized the potential of this test para-
digm for comparing competencies across signers of different ages and backgrounds, which led
to the development of a full test battery testing a variety of grammatical structures in ASL (New-
port, 1990: Newport & Supalla, 1980).

Linguistic Innovation—Verbs of Motion and Location

This method showed that across varying events of motion and location, adult native signers
use a limited set of discrete morphemes that recombine to form verbs of motion and location.
But it also allowed us to study different populations of individuals, some of whose responses
deviated from the performance of adult native signers. In this section, I describe these other
populations and situations where I initially thought one could find signers using analogical,
gestural expression.
 11. REVISITING VISUAL ANALOGY 253

Adult native signers responses to the test stimuli were completely systematic; in fact, I based
my definition of full grammatical competence in the morphology of these verbs on adult native
signers’ consistent, patterned responses. I consider the adult native signers’ responses using
these classifier predicates to be systematic (or grammatical) innovations, because the informant
may never have produced that particular combination of morphemes before. In fact I designed
many of the stimuli as unusual events (e.g., a loop moving over the top of a tree) in order to elicit
a novel combination. By their nature, polysynthetic morphological systems allow novel combi-
nations of morphemes (innovations) that are completely grammatical.
In contrast, young native children’s responses to the same stimuli did not always conform to
the grammatical targets found in native adult signers. Were they using nonlinguistic gesture in
the place of language? Interestingly, even these “erroneous” responses revealed a set of discrete
forms. For young children, many of the morphemes required in the adult verb were omitted. One
type of error involved the replacement of a single morpheme, the classifier, with a handshape
from a relevant ASL noun sign. This is also sometimes done by adult second language learners
of ASL. In contrast, older children (approximately 6 years old) would sometimes use
handshapes that were modifications of the target classifiers. These modified handshapes were
the result of recombinations and resynthesis of the features or parts that comprise these classifi-
ers. One might, at first consideration, call these unsystematic innovations on the part of the chil-
dren, because they do not conform to the adult grammar. On the other hand, the handshapes may
be fully systematic with respect to the child’s developing linguistic competence. The children’s
ungrammatical predicate classifiers were still composed of discrete morphemes, and the re-
placements were constrained to handshapes borrowed or derived from other parts of the lan-
guage, not idiosyncratic or analogic forms.
There do exist situations where adult native signers use classifier predicates that violate the
grammatical morphology of these verbs: when they are making a joke, playing on words, or for
artistic effect. Perhaps, I thought, this is where we will find signers leaving behind the basic con-
straints of the language, and taking up the mantle of gestural representation. In 1980 I presented
an analysis of Eric Malzkuhn’s translation of Lewis Carroll’s Jabberwocky poem, where I
showed that he created “nonsense” signs by playing with the morphological system in a discrete
and compontential fashion (Supalla, 1980). The result, like the English original, was a signed
poem about unfamiliar beasts living in a fantasy world. Although his signs were unsystematic
innovations, they were still created using the components found elsewhere in ASL. They also
still conformed to the same constraints on phonological formation and syntactic agreement used
in grammatical ASL. I found that Malzkuhn had merely replaced conventional classifiers with
novel hand and body articulators. Otherwise, his ASL was highly systematic and grammatical.
Padden and Humphries (1988, chap.5) and Wilbur (1987, pp. 32–34) describe further how sys-
tematic modifications to the phonemes and morphemes of ASL have been used to create a vari-
ety of signing artistic expressions.
It is clear that the morphology of verbs of motion offers rich possibilities for linguistic inno-
vation. More importantly, the creation of novel classifier predicates for artistic effect is not id-
iosyncratic; instead, the artist provides powerful evidence for the existence of a morpho-
logical system by systematically manipulating and violating it. All these cases show that a
model of the grammar of verbs of location and motion can account for both grammatical and
ungrammatical innovation.
There is another setting I have considered in which I thought adult native signers might revert
to gesture: in communicating with signers of mutually incomprehensible foreign sign lan-
254 SUPALLA

guages. For this, I examined the lectures of two native signers of two different sign languages.
The lectures were delivered in International Sign (IS), at an international conference for deaf re-
searchers (Supalla & Webb, 1995). IS is especially interesting as a lingua franca because it lacks
of a standardized lexicon, which forces signers to innovate new ways to represent concrete and
abstract referents. Although the lecturers were experienced users of IS, I thought the setting
might force the lecturers to revert to pantomime or gesture, to communicate abstract concepts to
their audience from a variety of different sign language backgrounds, even though using IS with
a single interlocutor with a single native sign language might be more systematic. Yet even in
this communication situation, where the lecturer and audience do not share a single grammar, I
found that the signs and sentences of IS were structured in much the same way as in sign lan-
guages: composed of discrete components.
It seems that native signers find it very hard to completely abandon the linguistic constraints
of their language. As an aside, when I supervised a dissertation on the topic of the gesture used
by hearing people, hearing lab members pointed out that my attempts to reproduce these ges-
tures did not appear natural, the way hearing people made them. Could it be that a signer cannot
completely abandon the phonology of his native language, and imports a language “accent”
even when imitating the gestures of hearing people? This fascinating question remains open.

HANDLING AND BODY CLASSIFIERS

To this point I focused on a subset of classifier predicates: verbs of motion and location. How-
ever there is another type of classifier predicate where classification is based on kinesthetic
properties rather than shape or abstract classification. I consider these predicate classifiers to
include two subtypes. One type is handling classifiers, which represent an object with a
handshape that resembles the hands holding or manipulating the object. Years ago, while read-
ing the literature on classifiers in spoken languages, I found a paper that presented pictures of a
person holding different quantities of various objects, such as a person holding bundles of
sticks of different sizes, to illustrate the classifiers used in that language. I realized then that
ASL also has a limited number of discrete morphemes to refer to quantities of different sub-
stances; for example, classifier morphemes for quantities of stick-like objects are based on
how they are grasped: the pinching F handshape, the handling classifier for a single stick; the
C-handshape for a small bundle; the two-handed C-handshape, palms facing each other, for a
middle-sized bundle; and the two arms extended in front, with hands overlapping, for large
bundles. In Supalla (1986), I presented a paradigm of different handling classifiers of this
type. Although the gestural modality permits an iconic representation of these handling classi-
fiers in a way that the acoustic medium does not, ASL nonetheless does not allow continuously
varying gestures to encode small increments in size. Although the form of the classifiers
within or across individuals can vary slightly, ASL only has four classifiers, for single, small,
medium and large bundles of stick-like objects.
The second type of kinesthetic classifier is body classifiers, which are articulated by the
signer’s upper body in a posture that in some way resembles the referent, such as a posture repre-
senting a person holding and swinging a baseball bat; a person walking in a relaxed manner, or
the body of an animal. I have not yet been able to identify a limited set of discrete components of
the upper body articulations that are used in these signs. But this does not mean there is no such
discrete system. From my previous results my best guess is that my analysis is simply not yet
correct or complete.
 11. REVISITING VISUAL ANALOGY 255

One impediment to my investigation of whether body classifiers are limited in number and
form has been that traditional ASL transcription systems (e.g. Stokoe, 1960, Stokoe et.al., 1965)
do not recognize the body as an active articulator, only as a set of locations. However, Kegl and
Wilbur’s (1976) work on projected body pronouns has offered a first linguistic account of the
body as an articulator involved in body shifting for point of view.
Another approach I considered was informed by the prototype model of categorization
(Rosch, 1973), which claimed that people share basic-level concepts about the actions and pos-
tures used to manipulate various objects in the world. For example, most people share a general
concept or a schema for eating pizza that involves holding a triangular shaped piece, cut from a
round pie, and taking a first bite from the pointed end (Lakoff, 1987). Lakoff’s descriptions sug-
gested to me that signs could be based on these schemas, and could therefore be categorized as
well. I analyzed the postural representations for different animals into a limb classifier, where
the hands represent the animals limbs, with specific handshapes for claws, paws, hoofs, feet, or
fins; and size and shape classifiers that represent facial/head features, such as shape of antlers,
horns, ears, mouth, nose, and eyes. This analysis reveals that there are contrasts based on a single
component of body posture. For example, the postural representation for a rabbit and a puppy are
identical except for one component, the mouth. Both involve the limb classifier for paws, but
puppy has an open, panting mouth, whereas rabbit has a toothy mouth.
This analysis was incomplete and did not fully account for these kinesthetic representations
using the upper body, and I have not pursued it further. However, as mentioned earlier, in Supalla
(1990) I found evidence that serial verbs of motion are systematically composed of a body clas-
sifier, identifying the animate noun, followed by a second verb of motion or location, incorporat-
ing a classifier and morphemes for path and manner or location. In that analysis, I found
contrasting body classifier components across different verbs of manner (e.g., a body classifier
for a human flying—the Superman pose; a rabbit hopping by; a man limping). Further work is
needed to identify a full set of body postural classifiers, and to investigate whether they can be
fully analyzed into contrasting features.

CONCLUSIONS AND DIRECTIONS FOR FUTURE RESEARCH

In this chapter I presented different aspects of classifier predicates that I have considered over
the years in evaluating whether they are linguistic—that is, discrete and componential
forms—or analogical and nonlinguistic. At each step of the way, I have failed to find evidence of
purely analogical, iconic representations in ASL, either in the handshapes or the movements and
manners used in these verbs. This is surprising, given the apparent ability of the gestural medium
to iconically encode continuous variation in real world objects and events. However, further re-
search on this question is needed. One important topic for future research concerns the syntax of
classifier constructions. Most of the research on verbs of motion thus far has focused on the
structure of individual verbs. I believe this research has been valuable in clarifying the morpho-
phonemic components of these verbs as a class, and the dimensions along which they differ from
other types of verbs in ASL. However, although we know that these verbs are used in particular
types of syntactic constructions (with nouns set up in space first, followed by one or several clas-
sifier verbs expressing the motions and spatial locations of these objects with respect to one an-
other), virtually no linguistic research has focused on the details of this syntax and the
morphosyntax involved in marking the verbs in relation to the rest of the sentence. Why is this
important? Liddell (2000) has raised questions about how many positions in space are signifi-
256 SUPALLA

cant in ASL points, agreement verbs, and verbs of motion, and whether there is a limit on the
number of spatial locations such constructions utilize. In my opinion one important consider-
ation in addressing this question is not the superficial number of locations that can be used at dif-
ferent times (this is like counting the number of different acoustic tokens in a spoken language),
but rather the relations (or agreements) between one element in the sentence and another. Many
of the morphemes of verbs of motion are agreement morphemes: agreeing in spatial location or
in classification with other elements of the sentence. To analyze them correctly, then, one needs
to examine how these agreement features are marked and distributed over the sentence, and even
over several sentences within a discourse passage. My expectation is that this type of analysis
will show that ASL, like other languages, critically relies on contrasting a few categories and re-
lations in space, not on maintaining and contrasting large numbers of precise individual points.
But a definitive answer to this question requires analyses that have not been performed.
A second important topic for future research, and the topic on which my own current re-
search focuses, concerns pursuing the relations between nonlinguistic gesture and sign lan-
guages, particularly as young sign languages form. My discussion above has sometimes
presupposed that nonlinguistic gesture is, in fact, continuous and analogical, quite unlike the
categorical and discrete nature of language. But this also is not well studied, and much recent
research on nonlinguistic gesture suggests that it too may have more discrete and even
componential structure than we have recognized. For example, Webb’s (1996) dissertation, on
gestures used by hearing individuals while speaking, shows that hearing people may have a
limited and shared lexicon of metaphoric gestures, and that some of these gestures may con-
tain and contrast gestural subcomponents that become full morphemes in sign languages.
Work from Goldin-Meadow (personal communication) and Dufour (1992), on gesticulation
by hearing subjects asked to gesture in the lab without speaking, shows that many devices uti-
lized in systematic ways by full sign languages can appear (though typically not entirely sys-
tematically or consistently) in this type of gesturing. Finally, our own work on Amami Island
(Osugi, Supalla, & Webb, 1999) shows both the similarities and differences of gesture struc-
ture for hearing individuals who do not use gesture as their language and hearing and deaf indi-
viduals who do. Ongoing work on Amami Island and on historical change in ASL (Supalla, in
press) will, I hope, provide further insight into the features of nonlinguistic gesture that are
borrowed and exploited for a developing sign language, and the changes involved in
grammaticalizing these features as they are used in a mature language. In these situations, we
hope to see more precisely how constraints of human cognition and learning may direct the
evolution of a gestural communication system into a full linguistic system.

ACKNOWLEDGMENTS

The research described in this chapter was supported in part by NIH grant DC00167. I would
like to thank Elissa Newport and Karen Emmorey for helpful comments, and Rebecca Webb for
editorial work on an earlier draft.

REFERENCES
Cogill, D. (2000). Classifier predicates, linguistic structures, or templated visual representations? Manu-
script submitted for publication.
DeMatteo, A. (1977). Visual imagery and visual analogues. In L. Friedman (Ed.), On the other hand: Re-
cent perspectives on American Sign Language (pp. 109–136). New York: Academic Press.
 11. REVISITING VISUAL ANALOGY 257
Dufour, R.(1992). The use of gesture for communicative purposes: Can gestures become grammaticized?
Unpublished doctoral dissertation, University of Illinois.
Frishberg, N. (1975). Arbitrariness and iconicity: Historical change in American Sign Language. Lan-
guage, 51, 696–719.
Kegl, J., & Wilbur, R. (1976). Where does structure stop and style begin? Syntax, morphlogy, and phonol-
ogy vs. stylistic variations in American Sign Language. Papers from the 12th Regional meeting, Chi-
cago Linguistic Society (pp. 376–396). Chicago, IL: Chicago Linguistic Society.
Klima, E., Bellugi, U., Battison, R., Boyes-Braem, P., Fischer, S., Frishberg, N., Lane, H., Lentz, E. M.,
Newkirk, D., Newport, E., Pedersen, C. C., & Siple, P. (1979). The signs of language. Cambridge, MA:
Harvard University Press.
Lakoff, G. (1987). Women, fire, and dangerous things. Chicago: University of Chicago Press.
Liddell, S. K. (2000). Indicating verbs and pronouns: Pointing away from agreement. In K. Emmorey, & H.
Lane (Eds.), The signs of language revisited: An anthology to honor Ursula Bellugi and Edward Klima
(pp. 303–320). Mahwah, NJ: Lawrence Erlbaum Associates.
Newport, E. L. (1990). Maturational constraints on language learning. Cognitive Science,14, 11–28.
Newport, E. L., & Supalla, T. (1980). The structuring of language: Clues from the acquisition of signed and
spoken language. In U. Bellugi & M. Studdert-Kennedy (Eds.), Signed and spoken language: Biologi-
cal constraints on linguistic form, Berlin.
Osugi, Y., Supalla, T. & Webb, R. (1999). The use of word elicitation to identify distinctive gestural systems
on Amami Island. Sign Language and Linguistics, 2, 87–112.
Padden, C., & Humphries, T. (1988). Deaf in America: Voices from a culture. Cambridge, MA: Harvard
University Press.
Rosch, E. (1973). Natural categories. Cognitive Psychology, 4, 328–350.
Schwartz, C. (1979), Discrete vs. continuous encoding in American Sign Language and non-linguistic ges-
tures, Unpublished manuscript, University of California, San Diego.
Stokoe, W. C. (1960). Sign Language Structure: An outline of the visual communication systems of the
American deaf. Studies in Linguistics: Occasional Papers, 8. Buffalo, NY: University of Buffalo.
Stokoe, W. C., Casterline, D., & Croneberg, D. (1965). Dictionary of American Sign Language. Washing-
ton, DC: Gallaudet University Press.
Supalla, T. (1978). Morphology of verbs of motion and location in American Sign Language. In F.
Caccamise & D. Hicks (Eds.), ASL in a bilingual, bicultural context. Proceedings of the Second
National Symposium on Sign Language Research and Teaching (pp. 27–45).Washington, DC: National
Association of the Deaf.
Supalla, T. (1980, October). A preliminary field guide to playing with sign. Paper presented at the Third Na-
tional Symposium on Sign Language Research and Teaching, Boston, MA.
Supalla, T. (1982). Structure and acquisition of verbs of motion in American Sign Language. Unpublished
doctoral dissertation, University of California, San Diego.
Supalla, T. (1986). The classifier system in American Sign Language. In C. Craig (Ed.), Noun classification
and categorization (pp. 181–214). Amsterdam: John Benjamins.
Supalla, T. (1990). Serial verbs of motion in ASL. In S. D. Fischer & P. Siple (Eds.), Theoretical issues in
sign language research: Vol. 1. Linguistics (pp. l27– 152). Chicago: University of Chicago Press.
Supalla, T. (in press). Making historical sign language materials accessible: A prototype database of early
ASL. Sign Language and Linguistics.
Supalla, T., & Webb, R. (1995). The grammar of International Sign: A new look at pidgin languages. In K.
Emmorey & J. Reilly (Eds.), Sign, gesture and space (pp. 333–352). Hillsdale, NJ: Lawrence Erlbaum
Associates.
Webb, R. (1996). Linguistic properties of metaphoric gestures. Unpublished doctoral dissertation, Depart-
ments of Linguistics and Brain & Cognitive Sciences, University of Rochester.
Wilbur, R. B. (1987). American Sign Language: Linguistic and applied dimensions. Boston: Little, Brown.
Woodward, J. (1974). Implicational variation in American Sign Language: Negative incorporation. Sign
Language Studies, 5, 20–30.
Woodward, J. (1976). Signs of change: Historical variation in American Sign Language. Sign Language
Studies, 10, 81–94.