<?xml version="1.0" standalone="yes"?>
<Paper uid="W97-0508">
  <Title>A Tutor for Teaching English as a Second Language for Deaf Users of American Sign Language</Title>
  <Section position="4" start_page="0" end_page="47" type="metho">
    <SectionTitle>
2 Literacy Issues for People Who
</SectionTitle>
    <Paragraph position="0"> are Deaf The problem of deaf literacy has been well: documented and has far reaching effects on every aspect of deaf students' education. Though data on writing skills is difficult to obtain, we note that the reading comprehension level of deaf students is considerably lower than that of their hearing counterparts, &amp;quot;...with about half of the population of deaf 18-year-olds reading at or below a fourth grade level and only about 10% reading above the eighth grade level...&amp;quot; (Strong, 1988) Some Deaf people use American Sign Language (ASL). 1 ASL is a visual-gestural language whose grammar is distinct and independent of the grammar  / of English or any other spoken language (Stokoe, Jr., 1960), (Baker and Padden, 1978), (Baker and Cokely, 1980), (Hoffmeister and Shettle, 1983), (Klima and Bellugi, 1979), (Bellman, Poizner, and Bellugi, 1983). The structure of ASL is radically different from that of English, being much more similar to that of Chinese or the Native American language Navaho. In addition to sign order rules (which are similar to word order rules of English), ASL syntax includes systematic modulations to signs as well as non-manual behavior (e.g., squinting, raising of eyebrows, body shifts, and shaking, nodding or tilting the head) for morphological and grammatical purposes (Baker and Cokely, 1980), (Liddell, 1980), (Padden, 1981), (Klima and Bellugi, 1979), (Kegl and Gee, 1983), (Ingrain, 1978), (Baker, 1980). The modality of ASL encourages simultaneous communication of information which is not possible with the completely sequential nature of written English.</Paragraph>
    <Paragraph position="1"> In addition to radical differences in the structure of ASL and English, another obstacle to the ASL user acquiring English is the unique processing strategies s/he brings to the task (Anderson, 1993). The cognitive elements used to store signs in short-term memory are distinctively different from those used with a spoken/written language. Also, hearers of spoken language buffer the speech in order to process it together in words and phrases, but the buffer for visually observed data has a much quicker decay time than that of auditory or visual data, which leads to repetition and redundancy in signed languages that does not occur in the same manner elsewhere. Moreover, long, involved utterances of a manual language are parceled into small parts that are recursively reinforced, referring back to previous details as each new piece of information is added, another characteristic atypical of spoken language.</Paragraph>
    <Paragraph position="2"> Adding to these difficulties is the fact that ASL has no accepted written form, eliminating the opportunity to establish literacy skills in a fluent native language and then transfer those skills to the new language being learned. Perhaps the worst difficulty for the deaf learner is that s/he has little to no understandable input in the language s/he is attempting to acquire. Thus, in addition to providing feedback on the student's writing, a tutoring system should be capable of offering sample understandable input using constructions that the student is currently attempting to master.</Paragraph>
    <Paragraph position="3"> We anticipate that our system will address the unique needs of the deaf population in other ways as well. For instance, this system would provide the focus has been on those people who are (near native) users of American Sign Language.</Paragraph>
    <Paragraph position="4"> user with feedback on his or her writing without involving a human teacher. Some students might prefer this mode of feedback since they would not risk feeling a &amp;quot;loss of face&amp;quot; as they might with a human tutor. The hope is that this will get the students to write more.</Paragraph>
    <Paragraph position="5"> In explaining the difficulties faced by the deaf learner of English, we do not propose that ASL natives are fundamentally different from other learners of English as a Second Language; rather, we want to stress the view that English is, for ASL natives, a fundamentally different and challenging language, motivating the need to adopt a Second Language Acquisition strategy toward facilitating the learning process. There exist many obstacles to this process, some which are shared with other native language populations and some which are unique, such as the absence of the opportunity to have English input tailored to the personal level of acquisition and understanding of the learner. The system we propose attempts to address these needs as closely as possible within its own constraints (i.e., without the ability to converse with the learner in his native language).</Paragraph>
    <Paragraph position="6"> We should note that while there are &amp;quot;style checkers&amp;quot; and &amp;quot;grammar checkers&amp;quot; on the market, these programs do not satisfy the needs of the deaf. Educators of the deaf (and other people working with deaf individuals) report that such checkers, geared toward the errors of hearing writers, frustrate deaf students. Tailored toward the writing style of fluent, native English speakers, they do not catch many errors that are common in the writing of people who are deaf, and, at the same time, they flag many constructions that are not errors. We ran some of our writing samples from deaf subjects through a few grammar checkers, and we judged the results to be consistent with these reports.</Paragraph>
    <Paragraph position="7"> There have been some attempts to develop &amp;quot;grammar checkers&amp;quot; for people who are deaf. Perhaps the most notable of these is the system named Ms. Pluralbelle which was developed and tested with students at Gallaudet University (Loritz, 1990), (Loritz, Parhizgar, and Zambrano, 1993). The work described here differs from this earlier work mainly in its emphasis on correction and on its model of the user's acquisition process.</Paragraph>
  </Section>
  <Section position="5" start_page="47" end_page="48" type="metho">
    <SectionTitle>
3 Overview of System Design
</SectionTitle>
    <Paragraph position="0"> Figure 1 contains a block diagram of the system under development. The system, called ICICLE (Interactive Computer Identification and Correction of Language Errors), is designed to be a general purpose language learning tutor. While our current focus is on users of ASL, and thus some of the modules</Paragraph>
    <Paragraph position="2"> will be specific to the errors and difficulties of this learner population, our eventual goal is to have the language-specific aspects of the system to be excisable, allowing modules for different native languages to be inserted, so the system would eventually be usable for any learner of English as a Second Language.</Paragraph>
    <Paragraph position="3"> The input/feedback cycle of ICICLE begins when the user enters a portion of text into the computer.</Paragraph>
    <Paragraph position="4"> The user's text is processed by the Error Analysis component which is responsible for tagging all errors. This component first performs a syntactic parse of a sentence using an English grammar augmented with error-production rules, or mal-rules (Sleeman, 1982), (Weischedel, Voge, and James, 1978). These mal-rules allow sentences containing errors to be parsed with the grammar, and enable the system to flag errors when they occur. The mal-rules themselves are derived from an error taxonomy which resulted from our writing sample analysis in conjunction with an analysis of how ASL knowledge might influence written English and other ASL information (Suri and McCoy, 1993). The initial taxonomy was developed from an analysis of forty-eight Freshman and Sophomore writing evaluation samples from Gallaudet University (a liberal arts university for the deaf), seventeen writing evaluation samples from the National Technical Institute for the Deaf (NTID, a deaf school in Delaware), and five letters and essays written by ASL natives and collected through the Biculturai Center in Washington, DC. In total, the samples represent about 25,000 words. The errors were hand-counted and categorized, leading to the development of the mal-rules which represent them.</Paragraph>
    <Paragraph position="5"> The possible effects of ASL on the errors identified are captured in the Language Model. The effects from the acquisition of English as a Second Language are captured in the Acquisition Model (described later in this paper). These two models affect a scoring mechanism which is used to identify a single parse (and set of errors) when multiple possibilities exist (McCoy, Pennington, and Suri, 1996).</Paragraph>
    <Paragraph position="6"> The error identification phase must also look for semantic errors (e.g., mixing of have and be), and for discourse level errors (e.g., NP deletions). Some of these errors will be flagged after syntactic parsing using independent error rules. Finally, the Error Identification module is responsible for updating any discourse information tracked by the system (e.g., focus information). Once this information is recorded, the next sentence will be analyzed.</Paragraph>
    <Paragraph position="7"> After all analyses are completed, the text, along with the error results and annotations from the error rules, will be passe d to the Response Generator.</Paragraph>
    <Paragraph position="8"> The Generator component processes this information (along with data from the User Model and possibly the History Module) in order to decide which errors to correct in detail and how each should be corrected (including what language level should be used in generating any required instruction). The decision as to which errors to correct in detail will be most influenced by reasoning on the Acquisition Model.</Paragraph>
    <Paragraph position="9"> The second decision that must be made in the Response Generator is which kind of correction strategy to use in actually generating the response. This decision is also affected by information stored in the User Model and History Module. The content of the response itself will be derived from the annotations on the errors that were passed from the Error Analysis component; additional content for the responses may be provided by the ASL/English &amp;quot;Expert&amp;quot; (Language Model) and influenced by the Acquisition Model. Finally, the responses will be displayed to the user who then has an opportunity to enter corrections to the text and have it re-checked.</Paragraph>
    <Paragraph position="10"> At the same time, information from the Response Generator will be used to update the recent and long-term &amp;quot;history&amp;quot; of the user. This knowledge can then be utilized to assess the user's second-language ability and other user characteristics, and to evaluate the success (or failure) of the correction techniques employed thus far.</Paragraph>
  </Section>
  <Section position="6" start_page="48" end_page="49" type="metho">
    <SectionTitle>
4 Implementation
</SectionTitle>
    <Paragraph position="0"> Our implementation to this point has concentrated most heavily on the analysis phase of processing. The user interacts with the system through  a windows-based interface 2 through which text may either be entered directly or loaded from a file. Once the text is loaded, the user may ask that it be analyzed by the system.</Paragraph>
    <Paragraph position="1"> The text is analyzed (one sentence at a time) by a bottom-up parser found in (Allen, 1995) using a grammar which has been augmented with mal-rules to capture errors uncovered in our analysis of writing samples. 3 The mal-rules are indexed with the errors that they realize.</Paragraph>
    <Paragraph position="2"> The following is an example of a mal-rule from the grammar currently in implementation:</Paragraph>
    <Paragraph position="4"> The rule shown is a simple sentence rule that states that an s is an np followed by a vp (where the vp is the head). The left and right hand sides of the rule are delimited by the rule name (-my01.2&gt; in this case), and each constituent has a set of features that are associated with it. This rule would recognize an error at the sentence formation level, in subject-verb agreement - specifically, an error where the subject is plural but the verb form is third-person singular, such as &amp;quot;We does...&amp;quot; or &amp;quot;They has...&amp;quot; By tagging the sentence parse with the feature (error-feature +), it is identified as containing an error, and the parse tree can be examined to discover the real-rule (in this case, my01.2) that was used in the parse.</Paragraph>
    <Paragraph position="5"> After all of the sentences have been parsed in this way, the current system displays the text with colored highlighting over all error-containing sentences (different colors are used for different classes of error, again as identified from the real-rules which were used). In addition, a color-coded menu appears which names the errors and associates them with the colors from the highlighted display. At this point the user may investigate the individual errors further. For example, s/he may click on a particular error name to get a (currently canned) explanation, or s/he may ask the system to mark the occurrences of a particular error only. In addition, the user may  their work on the grammar and Linda Suri for the writing sample analysis and development of the error taxonomy.</Paragraph>
    <Paragraph position="6"> edit particular sentences, which results in an immediate new analysis of the text.</Paragraph>
  </Section>
  <Section position="7" start_page="49" end_page="51" type="metho">
    <SectionTitle>
5 Accounting for the L2 Acquisition
</SectionTitle>
    <Paragraph position="0"> There are several reasons why a model of second language acquisition is necessary.</Paragraph>
    <Section position="1" start_page="49" end_page="49" type="sub_section">
      <SectionTitle>
5.1 Identifying Errors
</SectionTitle>
      <Paragraph position="0"> It is common for our system to find multiple possible parses of an input string, where some parses may contain mal-rules and others do not, some may contain different mal-rules than others, etc. Deciding between these multiple parses corresponds to deciding which errors (if any) the student made in the given sentence. One area of our current work concerns progress toward making an informed choice about which parse tree best represents the student's input.</Paragraph>
      <Paragraph position="1"> Our method is to develop a model of second language acquisition and use it for this task. For example, if we had a model of what the student had already acquired, what the student was currently acquiring, and what the student was most likely to acquire next, this could be used to select the most likely parse of the sentence in a principled fashion.</Paragraph>
      <Paragraph position="2"> A student is most likely to make errors in constructions s/he is currently acquiring (Vygotsky, 1986). Thus, given a set of parses, the one that is most likely to best describe the input is the one that contains mal-rules corresponding to errors in that realm of constructions (and that does not use constructions well beyond the student's current acquisition level).</Paragraph>
    </Section>
    <Section position="2" start_page="49" end_page="50" type="sub_section">
      <SectionTitle>
5.2 Focusing the Correction
</SectionTitle>
      <Paragraph position="0"> Once errors have been detected, the system must determine: * which errors to focus on in the correction * what basic content to include in the corrective response Our model of second language acquisition is crucial for these tasks as well. Research in second language acquisition and education indicates that as a learner is mastering a subject, there is a certain subset of the material that is currently &amp;quot;within his grasp.&amp;quot; This has been called the Zone of Proximal Development (ZPD) by Vygotsky (Vygotsky, 1986).</Paragraph>
      <Paragraph position="1"> This general idea has been applied to assessment and writing instruction by (Rueda, 1990), and second language acquisition by (Krashen, 1981). Intuitively the knowledge or concepts within the ZPD are &amp;quot;currently being acquired&amp;quot;.</Paragraph>
      <Paragraph position="3"> According to the above literature, instruction and corrective feedback dealing with aspects within the ZPD may be beneficial; instruction or corrective feedback dealing with aspects outside of the ZPD will likely have little effect and may even be harmful to the learning process, either boring or confusing the student with information s/he is unable to comprehend or apply. Thus the correction should focus on features at or slightly above the student's level of acquisition.</Paragraph>
      <Paragraph position="4"> Once an error has been identified and chosen for a corrective response, the system must also decide on the content of that response. Here again, where the user is in the acquisition process (and thus, why s/he made the error) is crucial. Consider the following example found in one of our writing samples: &amp;quot;My brother like to go...&amp;quot; This sentence appears to most of us to have a problem in subject-verb agreement. Because the subject is third-person singular, the present tense verb should be &amp;quot;likes.&amp;quot; Notice that there are several reasons why this error may be generated:  1. The student doesn't know that such agreement exists in the language. That is, the student may be unaware that the form of the subject has anything to do with the form of the verb in such sentences.</Paragraph>
      <Paragraph position="5"> 2. The student is mistaken about the syntactic form the agreement takes. In this case, the student is aware that s/he needs to mark subject-verb agreement, but does not know how to do so (or believes that s/he has already done so).</Paragraph>
      <Paragraph position="6"> 3. The student intended the noun to be in plural form (but mistyped).</Paragraph>
      <Paragraph position="7"> 4. The student intended the verb to be in singular  form (but mistyped).</Paragraph>
      <Paragraph position="8"> Notice that very different kinds of content would be required to effectively correct the above error depending on the actual reason for making it. In the first case, some general tutoring should be given, explaining that agreement exists in the language, the circumstances in which the agreement needs to be marked, and the iform the agreement should take.</Paragraph>
      <Paragraph position="9"> In case 2, only the form of the agreement needs to be explained. In cases 3 and 4, no tutoring should be given.</Paragraph>
      <Paragraph position="10"> Knowing where the student is in acquiring the second language can help a system distinguish among the cases above. If subject-verb agreement is something that the student has not acquired and is not about to acquire, case 1 is most likely. The student's placement in the model of acquisition can further direct our decisions regarding actions, because if this agreement is too far above the student's current level to be intellectually attainable at this time, we do not want to act on the error at all. If, on the other hand, it is currently within the ZPD (i.e., currently being acquired by the user), then case 2 is the most likely situation. Finally, either case 3 or 4 is likely if subject-verb agreement has already been acquired by the user.</Paragraph>
    </Section>
    <Section position="3" start_page="50" end_page="51" type="sub_section">
      <SectionTitle>
5.3 Modeling the L2 Acquisition Process
</SectionTitle>
      <Paragraph position="0"> We are currently developing a computational model that captures the way that English is acquired (as a second language) and gives us a framework upon which to project a student's &amp;quot;location&amp;quot; in that process. There is considerable linguistic evidence that the acquisition order of English features for second-language learners is relatively consistent and fixed regardless of the first language (Ingrain, 1989), (Dulay and Burt, 1974), (Bailey, Madden, and Krashen, 1974). In addition to studies concentrating on second language acquisition, research in language assessment and educational grade expectations (e.g., (Berent, 1988), (Lee, 1974), (Crystal, 1982)) also suggests that language features are acquired in a relatively fixed order. This research outlines sets of syntactic constructions (language features) that students are generally expected to master by a certain point in their study of the language. This work can be interpreted as specifying groups of features that should be acquired at roughly the same time.</Paragraph>
      <Paragraph position="1"> We have attempted to account for the preceding results in a language assessment model called SLALOM (&amp;quot;Steps of Language Acquisition in a Layered Organization Model&amp;quot;) 4. The basic idea of SLALOM is to divide the English language (the L2 in our case) into a set of feature hierarchies (e.g.~ morphology, types of noun phrases, and types of relative clauses). Within any single hierarchy, the features are ordered according to their &amp;quot;difficulty&amp;quot; of acquisition, reflecting their relative linguistic complexity. The ordering within feature hierarchies has been the subject of investigation in work such as (Ingram, 1989), (Dulay and Burt, 1974), and (Bailey, Madden, and Krashen, 1974).</Paragraph>
      <Paragraph position="2"> Figure 2 contains an illustration of a piece of SLALOM. We have depicted parts of four hierarchies in the figure: morphological syntactic features, noun phrases, verb complements, and various relative clauses. Within each hierarchy, the intention  So, for example, the model reflects the fact that the +ing progressive form of verbs is generally acquired before the +s plural form of nouns, which is generally acquired before the +s form of possessives, etc. Notice that there are also relationships among the hierarchies. This is intended to capture sets of features which are acquired at approximately the same time. These connections may be derived from work in language assessment and grade expectations such as found in (Berent, 1988), (Lee, 1974), and (Crystal, 1982). The figure indicates that while the +s plural ending is being acquired, so too are both proper and regular nouns, and one- and two-word sentences. While a learner is acquiring these features, we do not expect to see any relative clauses which are beyond that level of acquisition.</Paragraph>
      <Paragraph position="3"> We anticipate that SLALOM, when fully developed, will initially outline the typical steps in acquiring English as a second language. This model will then be tailored to the needs of individual students via a series of &amp;quot;filters,&amp;quot; one for each user characteristic that might alter the initial generic model. For instance, it is possible that the specific features of the student's Native Language (L1) will affect the rate or order of acquisition of the Second Language (L2). In particular, one would expect features shared in the L1 and L2 to be acquired more quickly than those which are not (due to positive language transfer). Another possible filter might reflect how various formal written-English instruction programs might alter the model, possibly stressing certain features normally acquired after others which remain unmastered.</Paragraph>
      <Paragraph position="4"> We are developing the initial language learning model and its filters based on acquisition literature.</Paragraph>
      <Paragraph position="5"> We expect to further solidify the model using the writing samples that we have already collected. We are currently performing statistical analysis on our growing body of hand-corrected samples to see what error classes co-occur with statistical significance.</Paragraph>
      <Paragraph position="6"> We also expect to seek input from English teachers of deaf students, to see how they rank their students' abilities based on assignments they correct.</Paragraph>
      <Paragraph position="7"> Once the SLALOM model is complete, we expect to rely on user modeling techniques to &amp;quot;place&amp;quot; the user within this model. This placement must be more sophisticated than simply looking at errors since some learners will avoid structures they do not know perfectly well in order to prevent error. Others will make heavy use of prefabricated patterns, such as the &amp;quot;tourist phrases&amp;quot; found in a travel book, whose use may precede a complete understanding of meaning or structure. Thus the placement algorithm must take into account both of these writing strategies.</Paragraph>
    </Section>
  </Section>
  <Section position="8" start_page="51" end_page="52" type="metho">
    <SectionTitle>
6 Generating the Response
</SectionTitle>
    <Paragraph position="0"> Aside from content, the generated response should have several other characteristics. In addition to providing examples of constructions the user is currently acquiring (as discussed earlier) the response should be organized so as to tie new knowledge into old knowledge thus facilitating meaningful learning as discussed by (Brown, 1994). When each new element is tied into already-learned data, and is presented so that pieces of new knowledge introduced together are related conceptually, the learning process gains a more significant meaning and new material is assimilate more quickly and entirely.</Paragraph>
    <Paragraph position="1"> In addition, responses should encourage both deductive and inductive learning (where in the former, a standard practice for many foreign language classrooms, the student is introduced to the rule and is expected to use it to construct specific examples; in the latter the student is not directly told the rule, but is encouraged to generalize to the rule from specific correct examples). Classrooms benefit from both forms, but the deaf learner has limited to no exposure to correct forms, so responses that encourage inductive learning may be particularly useful. We postulate that this technique may be best achieved by providing positive examples from the student's own work. We have investigated the possibility of doing a search on the parse trees of correct sentences in the writing sample in order to find those that most closely fit a desired template, perhaps based on a sentence the learner has written incorrectly elsewhere. null The Response Generator should also take into account that feedback to a language learner occurs at  two levels, affective and cognitive (Vigil and Oiler, 1976). The cognitive level is that which concerns the content of the feedback, or the part which addresses the intellect of the learner and either enforces the assimilation of the concepts involved, or tells the learner to retry his attempt at communication.</Paragraph>
    <Paragraph position="2"> The affective level is less explicit, expressed through nonverbal cues and tone of voice, addressing a less conscious aspect of the learner. Negative feedback in this area should be avoided, as it may result in an abortion of his attempts to communicate. Even when the cognitive content of the response is indicating that an error occurred, the affective feedback should always encourage the learner.</Paragraph>
  </Section>
class="xml-element"></Paper>