Categorical Perception of Palm Orientation in American Sign Language
by Stephen Richard DeVilbiss Moss
Introduction
Categorical Perception
The study of human perception has often revealed that observations in the world are not always perceived in their genuine states. Scientific
research in many areas has revealed that the human brain perceives things in contrast to their nature based on individual experience, environment,
context, intuition and more. The human mind changes what is observed, and this is of practical concern to language researchers. One phenomenon
of note within both psychophysics and language research is known as Categorical Perception (CP).
Within language, this phenomenon is observed by looking at characteristics of the most basic distinguishable parts, or features, that make up units
of speech. Isolating these parameters allows observation along a single-dimentional continuum. Categorical Perception occurs when language
users receive linguistic stimuli along this continuum and—rather than perceiving the stimulus as varied along the continuum, therefore
creating an infininte number of unique utterances—categorize the elements as memebers of discrete language categories along the continuum. On its own,
this is unremarkable, but CP describes a specific type of warping of the perceptual range. One distinguishing feature of Categorical Perception is observed
when an experienced speaker is presented with paired stimuli along this continuum. If these stimuli are varied equally, the speaker will have more difficulty
distinguishing between each if the speaker perceives them to be of the same category and an easier time if the tokens cross boundaries and are perceived to
be in separate categories [6, 7].
The magnitude of CP effects can vary between different continua. Pronounced CP effects are observed in the production of stop-consonants along the continuum of time
delay between the burst and the subsequent onset of vocal fold activation. Listen to an example of this below! Weaker CP effects are found when measuring position of the tongue along the
two-dimensional range of space that is utilized in articulation of vowel sounds. These slighter effects leave room for discussion of competing models, e.g. the Perceptual
Magnet Effect (PME) described by Patricia Kuhl. The PME is characterized by the best exemplar for the category rather than by the category's boundaries. Testing shows
that the exemplar is effectively pulling the surrounding stimuli closer and making discrimination more difficult closer to the exemplar and easier when the stimuli are
further from the center of the category (Kuhl, 1991).
Beyond spoken language, CP has been observed in visual and non-linguistic realms as well: facial expressions [3], object recognition
[5], colors [10], and American Sign Language (ASL) [1,
2, 4, 8].
American Sign Language
American Sign Language (ASL) is the primary language for deaf and hard-of-hearing individuals in North America
[12]. Rather than taking advantage of the aural/oral pathways of spoken languages, ASL is produced with gestures of the hands and
body and received visually. ASL exhibits a full range of phonological, morphological, and syntactic structures as do spoken languages. [11
].
The phonological characteristics of ASL can be evaluated via several different phonological models. Though each model is different, each model takes into account different
parameters of each sign, which are comparable to features of spoken language phonology. These parameters may be summarized as: movement, location, non-manual signals,
handshape, and palm orientation.
Voice Onset Continuum
An audio demonstration of a voice-onset-time continuum. Notice, that the word deer begins the spectrum and the word tier is the final word.
The only manipulated element in this group of sounds is the duration between the burst of the articulation and the onset of of the voicing (when the vocal folds are activated).
The /d/ sound which begins deer occurs on the spectrum when the voicing begins simultaneously within 0 ms - 10 ms after the articulatory burst. The /t/ sound which begins
the word tier occurs when the voice beguns with a delay of approximately 70 ms after the burst. What American-English speakers will notice is that instead of hearing seven
different sounds, thus seven different word samples, they hear several utterances of the words deer and tier. (I hear five deers and two tiers)
At the time of this writing, studies have found CP effects in some but not all handshape continua [1,
4, 8, 9], and no CP effects in location continua (Emmorey, McCullough, & Brentari, 2003; Morford, et al., 2008;
Newport, 1982). While CP has been found in studies of emotional facial expressions (Cheal & Rutherford, 2013), the perception of linguistic non-manual signals remains an
open question. It remains open for movement and palm orientation as well.
In traditional studies of CP, an identification or categorization task is paired with a discrimination task to find the boundary between categories and then verify the
increased discriminability near those boundaries. In the identification component of spoken language studies, participants are asked to listen to a tone along an evenly
measured continuum and then identify it as a member of a category denoted by one of two endpoints of the continuum (need citation?). Because of the typical syntactic nature
of ASL, these tasks have been modified typically into an ABX format, where each of the endpoint tokens (A and B) are presented first and then the specific token from along
the continuum (X) is identified by the participant as a member of either category A or category B (need citation).
These identification tasks record successful categorization nearly all of the time for native signers, late signers, and sign naïve individuals (need citation?). However, it
seems unlikely that an individual with no experience in signed languages would truly identify phonological categories rather than simply grouping tokens into the nearest
group perceived. Additionally, these tasks do not accurately measure situations in which participants wish to identify more than two categories along the continua, nor
accurately describe differences in actual category continua from that predicted a priori by researchers. There are a few ways in which these studies may be improved to more
accurately reflect existing phonological categories in participants.
A long running controversial debate in the realm of sociolinguistics may offer a solution to these types of weaknesses. In their 1969 studies of basic color terms within
languages which contain fewer than 11 color terms, Berlin and Kay presented participants with an array of color chips (See Appendix A), and then ascertained from the
participant for each color term x: “(1) all those chips which he would under any conditions call x. (2) the best, most typical examples of x.” (Berlin & Kay, 1999). The
first task in essence requests category borders, and the second an exemplar for the category. The visual nature of ASL may allow a similar paradigm to account for variances
such as these as well as identify individual variances based on dialect, etc.
Assuming this new research method is successful, when paired with traditional discrimination tasks from ASL CP studies, we may begin to develop a basis for answering several
questions. What are the categories along palm orientation continua that are identified by native signers? Do sign naïve individuals continue to identify strong category
boundaries? Is there variation in discriminability along this continuum? Is it the same for native signers and sign naïve individuals? In addition to experience, does
language context influence this?
