Linguistic Knowledge can Improve Information Retrieval

William A. Woods Lawrence A. Bookman Ann Houston
Robert J. Kuhns Paul Martin
Stephen Green
Sun Microsystems Laboratories
1 Network Drive
Burlington, MA 01803
fWilliam.Woods,Ann.Houston,Robert.Kuhnsg@east.sun.com
fPaul.Martin,Stephen.Greeng@east.sun.com
March 10, 2000

Abstract
This paper describes the results of some experiments using a new approach to information
access that combines techniques from natural language processing and knowledge representation with a penalty-based technique for relevance estimation and passage retrieval. Unlike
many attempts to combine natural language processing with information retrieval, these results show substantial benefit from using linguistic knowledge.

1 Introduction
An online information seeker often fails to nd what is wanted because the words used in the
request are different from the words used in the relevant material. Moreover, the searcher usually
spends a significant amount of time reading retrieved material in order to determine whether it
contains the information sought. To address these problems, a system has been developed at
Sun Microsystems Laboratories (Ambroziak and Woods, 1998) that uses techniques from natural
language processing and knowledge representation, with a technique for dynamic passage selection and scoring, to significacantly improve retrieval performance. This system is able to locate specific passages in the indexed material where the requested information appears to be, and to score those passages with a penalty-based score that is highly correlated with the likelihood that they contain relevant information. This ability, which we call \Precision Content Retrieval" is achieved Lawrence Bookman is now at Torrent Systems, Inc. by combining a system for Conceptual Indexing with an algorithm for Relaxation-Ranking Specific
Passage Retrieval.
In this paper, we show how linguistic knowledge is used to improve search e ectiveness in this
system. This is of particular interest, since many previous attempts to use linguistic knowledge
to improve information retrieval have met with little or mixed success (Fagan, 1989; Lewis and
Sparck Jones, 1996; Sparck Jones, 1998; Varile and Zampolli, 1997; Voorhees, 1993; Mandalaetal., 1999) (but see the latter for some successes as well).
2. Conceptual Indexing
The conceptual indexing and retrieval system used for these experiments automatically extracts
words and phrases from unrestricted text and organizes them into a semantic network that inte-
grates syntactic, semantic, and morphological relationships. The resulting conceptual taxonomy
(Woods, 1997) is used by a speci c passage-retrieval algorithm to deal with many paraphrase relationships and specific passages of text where the information sought is likely to occur. It
uses a lexicon containing syntactic, semantic, and morphological information about words, word
senses, and phrases to provide a base source of semantic and morphological relationships that
are used to organize the taxonomy. In addition, it uses an extensive system of knowledge-based
morphological rules and functions to analyze words that are not already in its lexicon, in order to
construct new lexical entries for previously unknown words (Woods, 2000). In addition to rules
for handling derived and inflected forms of known words, the system includes rules for lexical
compounds and rules that are capable of making reasonable guesses for totally unknown words.
A pilot version of this indexing and retrieval system, implemented in Lisp, uses a collection
of approximately 1200 knowledge-based morphological rules to extend a core lexicon of approximately 39,000 words to give coverage that exceeds that of an English lexicon of more than 80,000base forms (or 150,000 base plus inflected forms). Later versions of the conceptual indexing and retrieval system, implemented in C++, use a lexicon of approximately 150,000 word forms that is automatically generated by the Lisp-based morphological analysis from its core lexicon and an input word list. The base lexicon is extended further by an extensive name dictionary and by further morphological analysis of unknown words at indexing time. This paper will describe some experiments using several versions of this system. In particular, it will focus on the role that the linguistic knowledge sources play in its operation.
The lexicon used by the conceptual indexing system contains syntactic information that can
be used for the analysis of phrases, as well as morphological and semantic information that is
used to relate more specific concepts to more general concepts in the conceptual taxonomy. This
information is integrated into the conceptual taxonomy by considering base forms of words to subsume their derived and inflected forms (\root subsumption") and more general terms to subsume more specific terms. The system uses these relationships as the basis for inferring subsumption relationships between more general phrases and more speci c phrases according to the in tensional subsumption logic of Woods (Woods, 1991).
The largest base lexicon used by this system currently contains semantic subsumption information for something in excess of 15,000 words. This information consists of basic \kind of" and \instance of" information such as the fact that book is a kind of document and washing is a kind of cleaning. The lexicon also records morphological roots and axes for words that are derived or inflected forms of other words, and information about different word senses and their interrelationships. For example, the conceptual indexing system is able to categorize becomes black as a kind of color change because becomes is an inflected form of become, become is a kind of change and black is a color. Similarly, color disruption is recognized as a kind of color change, because the system recognizes disruption as a derived form of disrupt, which is known in the lexicon to be a kind of damage, which is known to be a kind of change.

3 Relaxation Ranking and Specific Passage Retrieval
The system we are evaluating uses a technique called \relaxation ranking" to specific passages
where as many as possible of the different elements of a query occur near each other, preferably
in the same form and word order and preferably closer together. Such passages are ranked by a
penalty score that measures the degree of deviation from an exact match of the requested phrase,
with smaller penalties being preferred. Differences in morphological form and formal subsumption of index terms by query terms introduce small penalties, while intervening words, unexplained permutations of word order, and crossing sentence boundaries introduce more significant penalties. Elements of a query that cannot be found nearby introduce substantial penalties that depend on the syntactic categories of the missing words.
When the conceptual indexing system is presented with a query, the relaxation-ranking retrieval
algorithm searches through the conceptual taxonomy for appropriately related concepts and uses
the positions of those concepts in the indexed material to specific passages that are likely to
address the information needs of the request. This search can find relationships from base forms
of words to derived forms and from more general terms to more specific terms, by following paths in the conceptual taxonomy.
For example, the following is a passage retrieved by this system, when applied to the UNIX                operating system online documentation (the \man pages"):
Query: print a message from the mail tool 6. -2.84 print mail tool
Print sends copies of all the selected mail items to your default printer. If there are no selected items, mail tool sends copies of those items you are currently. The indicated passage is ranked6th in a returned list of found passages, indicated by the 6in the above display. The number -2.84 is the penalty score assigned to the passage, and the subsequent words print, mail, mail, and mail tool indicate the words in the text that are matched to the corresponding content words in the input query. In this case, print is matched to print, message to mail, mail to mail, and tool to mail tool, respectively. This is followed by the content of the actual passage located. The information provided in these hit displays gives the information seeker a clear idea of why the passage was retrieved and enables the searcher to quickly skip down
the hit list with little time spent looking at irrelevant passages. In this case, it was easy to identify
that the 6th ranked hit was the best one and contained the relevant information.
The retrieval of this passage involved use of a semantic subsumption relationship to match
message to mail, because the lexical entry for mail recorded that it was a kind of message. It used
a morphological root subsumption to match tool to mail tool because the morphological analyzer
analyzed the unknown word mail tool as a compound of mail and tool and recorded that its root
was tool and that it was a kind of tool modified by mail. Taking away the ability to morphologically analyze unknown words would have blocked the retrieval of this passage, as  would eliminating the lexical subsumption entry that recorded mail as a kind of message.
Like other approaches to passage retrieval (Kaszkiel and Zobel, 1997; Salton et al., 1993; Callan, 1994), the relaxation-ranking retrieval algorithm identifies relevant passages rather than simply
identifying whole documents. However, unlike approaches that involve segmenting the material
into paragraphs or other small passages before indexing, this algorithm dynamically constructs
relevant passages in response to requests. When responding to a request, it uses information in
the index about positions of concepts in the text to identify relevant passages. In response to a
single request, identified passages may range in size from a single word or phrase to several sentences or paragraphs, depending on how much context is required to capture the various elements of the request.
In a user interface to the specific passage retrieval system, retrieved passages are reported to
the user in increasing order of penalty, together with the rank number, penalty score, information
about which target terms match the corresponding query terms, and the content of the identified
passage with some surrounding context as illustrated above. In one version of this technology,
results are presented in a hypertext interface that allows the user to click on any of the presented
items to see that passage in its entire context in the source document. In addition, the user can
be presented with a display of portions of the conceptual taxonomy related to the terms in the
request. This frequently reveals useful generalizations of the request that would find additional
relevant information, and it also conveys an understanding of what concepts have been found in
the material that will be matched by the query terms. For example, in one experiment, searching
the online documentation for the Emacs text editor, the request jump to end of le resulted in
feedback showing that jump was classified as a kind of move in the conceptual taxonomy. This
led to a reformulated request, move to end of le, which successfully retrieved the passage go to
end of buffer.

4 Experimental Evaluation
In order to evaluate the effectiveness of the above techniques, a set of 90 queries was collected
from a naive user of the UNIX operating system, 84 of which could be answered from the online
documentation known as the man pages. A set of \correct" answers for each of these 84 queries
was manually determined by an independent UNIX operating system expert, and a snapshot of the man pages collection was captured and indexed for retrieval. In order to compare this methodology with classical document retrieval techniques, we assign a ranking score to each document equal to the ranking score of the best ranked passage that it contains.
In rating the performance of a given method, we compute average recall and precision values
at 10 retrieved documents, and we also compute a \success rate" which is simply the percentage of queries for which an acceptable answer occurs in the top ten hits. The success rate is the principal factor on which we base our evaluations, since for this application, the user is not interested in subsequent answers once an acceptable answer has been found, and finding one answer for each of two requests is a substantially better result than finding two answers to one request and none for another.
These experiments were conducted using an experimental retrieval system that combined a
Lisp-based language processing stage with a C++ implementation of a conceptual indexer. The
linguistic knowledge sources used in these experiments included a core lexicon of approximately
18,000 words, a substantial set of morphological rules, and specialized morphological algorithms
covering inflections, prefixes, suffixes, lexical compounding, and a variety of special forms, including numbers, ordinals, Roman numerals, dates, phone numbers, and acronyms. In addition, they made use of lexical subsumption taxonomy of approximately 3000 lexical subsumption relations, and a small set of semantic entailment axioms (e.g., display entails see, but is not a kind of see). This system is described in (Woods, 1997). The database was a snapshot of the local man pages (frozen at the time of the experiment so that it wouldn't change during the experiment), consisting of approximately 1800 les of varying lengths and constituting a total of approximately 10 megabytes.

5 Discussion
Table 1 shows that for this task, the relaxation-ranking passage retrieval algorithm without its supplementary knowledge sources (Recall II w/o knowledge) is roughly comparable in performance (42.9% versus 44.0% success rate) to a state-of-the-art commercial search engine (SearchIt) at the pure document retrieval task (neglecting the added benefit of locating the specific passages). Adding the knowledge in the core lexicon (which includes morphological relationships, semantic subsumption axioms, and entailment relationships), but without morphological analysis of un-
known words (Recall II w/o morph), significantly improves these results (from 42.9% to 50.0%).
Further adding the morphological analysis capability that automatically analyzes unknown words
(deriving additional morphological relationships and some semantic subsumption relationships)
significantly improves that result (from 50.0% to 60.7%). In contrast, we found that adding the
same semantic subsumption relationships to the commercial search engine, using its provided thesaurus capability degraded its results, and results were still degraded when we added only those facts that we knew would help find relevant documents. It turned out that the additional relevant documents found was more than o set by additional irrelevant documents that were also ranked more highly.

6.  Anecdotal Evaluation of Specific Passage Retrieval Benefits
As mentioned above, comparing the relaxation-ranking algorithm with document retrieval systems measures only a part of the benefit of the specific passage retrieval methodology. Fully evaluating the quality and ranking of the retrieved passages involves a great many subtleties. However, two informal evaluations have been conducted that shed some light on the benefits.
The rest of these was a pilot study of the technology at a telecommunications company. In
that study, one user found that she could use a single query to the conceptual indexing system to find both of the items of information necessary to complete a task that formerly required searching two separate databases. The conclusion of that study was that the concept retrieval technology performs well enough to be useful to a person talking live with a customer. It was observed that the returned hits can be compared with one another easily and quickly by eye, and attention is taken directly to the relevant content of a large document. The automatic indexing was considered a plus compared with manual methods of content indexing. It was observed that an area of great potential may be in a form of knowledge management that involves organizing and providing intelligent access to small, unrelated \nuggets" of textual knowledge that are not amenable to conventional database archival or categorization.
A second experiment was conducted by the Human Resources Webmaster of a high-tech company, an experienced user of search engines who used this technology to index his company's internal HR web site. He then measured the time it took him to process 15 typical HR requests first using conventional search tools that he had available, and then using the Conceptual Indexing technology. In both cases, he measured the time it took him to either nd the answer or to conclude that the answer wasn't in the indexed material. His measured times for the total suite were 55 minutes using the conventional tools and 11 minutes using the conceptual indexing technology. Of course, this was an uncontrolled experiment, and there is some potential that information learned from searching with the traditional tools (which were apparently used first) might have provided some benefit when using the conceptual indexing technology. However, the fact that he found things with the latter that he did not nd with the former and the magnitude of the time difference suggests that there is an effect, albeit perhaps not as great as the measurements. As a
result of this experience, he concluded that he would expect many users to take much longer to find materials or give up, when using the traditional tools. He anticipated that after finding some
initial materials, more time would be required, as users would end up having to call people for
additional information. He estimated that users could spend up to an hour trying to get the information they needed...having to call someone, wait to make contact and finally get the information
they needed. Using the conceptual indexing search engine, he expected that these times would be
at least halved.

Conclusion
We have described some experiments using linguistic knowledge in an information retrieval system in which passages within texts are dynamically found in response to a query and are scored and ranked based on a relaxation of constraints. This is a different approach from previous methods of passage retrieval and from previous attempts to use linguistic knowledge in information retrieval. These experiments show that linguistic knowledge can significantly improve information retrieval performance when incorporated into a knowledge-based relaxation-ranking algorithm for specific passage retrieval.
The linguistic knowledge considered here includes the use of morphological relationships between words, taxonomic relationships between concepts, and general semantic entailment relationships between words and concepts. We have shown that the combination of these three
knowledge sources can significantly improve performance in finding appropriate answers to specific
queries when incorporated into a relaxation-ranking algorithm. It appears that the penalty-based
relaxation-ranking algorithm figures crucially in this success, since the addition of such linguistic knowledge to traditional information retrieval models typically degrades retrieval performance rather than improving it, a pattern that was borne out in our own experiments.

Virtual Misadventures: Technical Problems and Student Satisfaction When Implementing Multimedia in an Advanced French Listening Comprehension Course

Alysse Weinberg
Second Language Institute
University of Ottawa

Abstract:
What are the advantages and problems related to using multimedia technology in an advanced French listening comprehension course? What are some of the typical vicissitudes, trials and tribulations encountered when teachers decide to introduce multimedia activities in their classes? This article illustrates the advantages and difficulties of using multimedia from the point of view of both the teacher and the learners. We first give a brief presentation of the context in which our experiment took place. We outline the difficulties encountered by the professor who had to select material, create activities, and administer the daily management of the course, touching on hardware and support issues. We then investigate the reaction of the students to the new media and the teaching approach.

INTRODUCTION

Enthusiasm and fear: Is this how language teachers feel when they approach the integration of the Internet and multimedia technology into their courses? Teachers are enthusiastic about Internet resources which can give their students immediate access to an immense pool of authenticmaterial (e.g., documents, music, and news broadcasts by native speakers). Learners may experience aspects of the target culture through the use of video and audio files, graphics, texts, and online tools without having visited the country (Garrett, 1991; Joiner, 1997; Pusack & Otto, 1997; Lafford & Lafford, 1997; Osuna & Meskill, 1998). However, while the process of integrating the Internet into the foreign language classroom curriculum stimulates learner interest and motivation (Sohonee, 1998), implementing a multimedia component in a language course involves a completely different set of skills than that of normal classroom teaching. Technologically inexperienced teachers, conscious of their technical inadequacies and their lack of control over capricious technology, often approach this task with trepidation merging into fear.

After a brief presentation of the context in which our experiment took place we address the following two questions:

1. What kind of technical difficulties are encountered by introducing Internet and multimedia activities into an advanced French listening comprehension course?

2. What are the perceptions and preferences of students presented with three different types of media—audio, video, and multimedia?

CLASSROOM AND TEACHING PROCEDURES: THE COURSE AND ITS SETTING

Between 1986 and 1994, the comprehension-based approach, as promoted by Krashen (1981), was widely used in the beginning level French courses at the Second Language Institute (SLI) of the University of Ottawa. This approach emphasized receptive skills—reading and listening— and de-emphasized productive skills—writing and speaking. In addition, the approach put greater focus on the students and their learning processes. Comprehension was perceived as the first step for language learning. This eight-year long experience required the creation of a specific curriculum, the development of listening and reading materials, and the implementation of testing procedures, along with studies of student satisfaction (Corbeil & Thérien, 1992; Cornaire & Tréville, 1992; Courchêne, 1992; LeBlanc, Duquette, & Compain, 1992; Compain, Dionne, Duquette, Rivas-Rivas, & Weinberg, 1995).

While many low-level comprehension courses have been abandoned, one advanced course in French listening comprehension was maintained. The description for this course states that "students will develop their listening strategies and skills and that they will be systematically exposed to a variety of accents, registers, environments, text types and delivery modes." The main objective is to prepare the student to easily understand a university course given in French.

No textbook is attached to the course, so the teacher is responsible for selecting the material at the appropriate advanced linguistic level to match the syllabus and course aims. The selection criteria for the material was based on content, authenticity, and variety as stated in the course syllabus. Content was chosen designed to enlarge students' knowledge of the contemporary manifestations and historical context of French culture. The main themes were similar to topics in introductory courses in fields of study such as sociology, history, and so on. All of the audio and video segments used in the course contained authentic French material taken directly from stations such as Radio Canada and SRC (Canada), TV5 (France), and from movies such as Le fleuve aux grandes eaux. Nothing was especially created for the course. A variety of speaking styles were selected to expose the students to various regional accents, gestures, rates of speaking, and discourse levels of the language as specified in the course syllabus. Represented were accents from Alsace, the south of France, Belgium, Switzerland, the Caribbean islands, Africa, and of course from French speaking Canada, especially Acadia, Québec, and Francophone Ontario. The media clips varied in length from 10 to 45 minutes.

The course description features listening strategies, and these strategies represent a major component of this course in order to increase the students' awareness about their learning processes. The definition of strategy here follows the commonly definition of a conscious mental operation applied by learners to help with their language acquisition (O'Malley, Chamot, Stewer-Manzanares, Küpper, & Russo, 1985; Chamot, Küpper, & Impink-Hernandez,1988; Oxford & Crookall, 1989). The typology proposed by O'Malley and Chamot (1990) classifies strategies in three large categories: (a) metacognitive strategies which assist learners in the evaluation and improvement of their own learning processes, (b) cognitive strategies which guide learner interactions with their text, and (c) social-affective strategies which help students with their foreign language interactions with others. By deliberately presenting and discussing learning strategies, the professor trains students to understand their language comprehension processes, to become more autonomous in their learning, and to use these strategies outside the classroom situation (Mendelsohn, 1994; Oxford, 1990).

All listening comprehension activities in the course included prelistening, listening, and postlistening components (Underwood, 1989; Joiner, 1997) and presented inherently interesting subject matter appropriate to the students' level. Working with previewing and prelistening activities reduce some of the difficulties inherent to the use of authentic material and thereby enables students to predict the course of a conversation based on their previous knowledge (Duplantie & Massey, 1984; Mendelsohn, 1995). Students were also taught viewing strategies such as segment planning, defining viewing goals, identifying problems, and evaluating the effectiveness of self-teaching methods.

The course is normally based only on the presentation of audio and video material, but, starting in January 2000, students were exposed to multimedia material as well. The class met twice a week for 90 minutes for 12 weeks. One of the two weekly classes was devoted to viewing video segments. This class was held in a small amphitheatre where the video was projected on a large screen (approximately 8 feet high and 12 feet wide). The second of the two weekly classes was devoted to listening to audio clips. Before this time, the audio portion of the course took place in a traditional language laboratory; starting in January 2000, the students went to a computer laboratory for the multimedia part. The audio and video segments presented in the same week had similar themes. The testing for the audio and video components of the course also presented material similar to the themes shown and discussed in class.

THE MULTIMEDIA ENVIRONMENT AT THE UNIVERSITY OF OTTAWA

The course was held in a brand new multimedia laboratory, opened in May 1999, equipped with 30 Windows 98 workstations each with sound cards and headsets. The laboratory is run from a Prisma Tandberg control panel, and the equipment allows the playing and recording of audiocassettes. The student workstations have limited but useful Internet connections. The firewall on the main multimedia server does not allow students to access sound and video files over the Internet, but these files may be accessed and displayed by the teacher working at the main console. This limitation was implemented to reduce the amount of bandwidth required to service the laboratory. The multimedia materials could be accessed only from the campus multimedia laboratory.

THE PARTICIPANTS

In January 2000, 11 students registered for the course and in the middle of the term another student joined the course, making at the end nine women and three men. All were strong students with marks of A or B from the prerequisite advanced listening and speaking course. The total number of 12 students represented a small number of students because there are normally about 25 students in the course. All the students had each spent at least one year at the university; they were in their second, third, or fourth year except for one, a special student over 70 years old who was just taking the course for his own benefit. Seventy-five percent

334

of the students had previously taken French courses in our department. None of the students was working towards a minor degree in French or had declared French as a concentrated area of study. All students were Canadians with a rich variety of ethnic backgrounds such as South-American, African, Italian, and Turkish. Some were very familiar with the Internet and the different formats of sound and video files. Others were true beginners with the technology and required regular review of the different steps to start the program, to enlarge the screen, and to move from one screen to the next. Only half of the students had access to the Internet at home. For all students, this course was their first multimedia language course. The students were highly motivated and attended 94% of the classes. The average student mark on the midterm test was 76%, while the average mark for the final exam was 80%.

INTRODUCING MULTIMEDIA FOR LISTENING COMPREHENSION

The inclusion of a multimedia component in the course added variety, quality, and interactivity to interesting topics. The following list of features highlights the advantages of using multimedia components which were not previously available in the traditional language laboratory. (Noblitt, 1990; Furstenberg, 1997; Desmarais, 1998; Lancien, 1998):

•a multimodal presentation combining texts, sounds and videos,

• high quality sound, video, and pictures,

• glossaries, grammatical and lexical reviews, dictionaries, and hyperlinks to related material,

•a variety of interactive question types such as fill in the blank, multiple choice, drag and drop, pairing, associations, and so forth,

•a pleasant-to-use and attractive program interface that allows easy navigation,

• controls to pause and rewind the sound track,

• useful feedback analyzing students' responses,

• different program branching procedures depending on level of language skill, and

•a system to help students monitor what exercises have been completed and what tests have been passed.

The following features of multimedia materials are especially useful in a listening comprehension course:

• The learning activity provides perspectives on different aspects of the same text in order to allow students to understand more fully what is being presented.

• The clip can be accompanied by a text script or presented without it. The script can include only keywords, the whole script, the whole script with certain keywords hyperlinked to explanations, or a translation of the script.

• The activity can include text, pictures, and diagrams to clarify the historical or cultural context.

• Finally, and most importantly, the audio track is recorded in a high fidelity format.

Implementing a multimedia component into a language course includes a variety of tasks such as courseware selection, installation, activity selection, and user support for the students. Hubbard (1988, 1992, 1996) proposed a framework for CALL courseware evaluation and development. Murphy (1995) listed the six essential characteristics of good software as good documentation, learner control, branching capabilities, portability, ease of use, and cost effectiveness. Robb and Susser (2000) made a study of how foreign language instructors choose courseware and found that, despite the availability of standardized techniques and checklists to select and evaluate courseware, a gap exists between the methodology recommended in the literature and the actual practice by the instructors.

Having decided to introduce multimedia in ones course, the language teacher then faces two options: to buy or to develop brand new material. For our advanced French comprehension course no commercial CDs were used. The ones we considered were flawed in some way that made them unsuitable. Most were aimed at the beginning level and therefore not relevant to this advanced course. Some had poor quality sound or confusing user interfaces. Some commercial CDs could not be installed on the local area network. Some emphasized only French culture from France and were not suitable for a Canadian university. While a single CD may be reasonably priced, the cost of a site license may be prohibitively expensive. We decided then to create our own activities designed specifically for the course. The next section details the challenges and the problems encountered in the development and implementation of the multimedia activities.

PROBLEMS WITH PREPARING VIDEO AND AUDIO FILES FOR USE IN THE LABORATORY

Multimedia File Formats and Student Access

The students needed to have reliable access to video or sound files used in the learning activities but were frustrated by the university Internet firewall that blocked downloading sound and video files to student computers. The professor's computer was also filtered by this firewall, but she finally obtained authorization to let sound and video files pass the firewall. These segments could then be displayed to all of the student terminals by mirroring the professor's workstation. Nevertheless, this solution was far from satisfactory for the students because they wanted to have complete control over the video rather than just having the professor play it for them.¹ For the first video segment, the students even had to memorize the information presented in the video in order to answer the subsequent questions on the text. After this first unsatisfactory session, a different approach was used.

Many video files on the Internet are stored as RealPlayer .ram or .ra formats. These formats, referred to as streaming files, have been engineered by the company to prevent Internet users from downloading local copies for copyright reasons. While the video portion cannot be downloaded, the audio portion can be saved while the video is playing. We saved the sound file as an independent file and stored it on the local network. The students could then view the video played by the teacher and also listen to the soundtrack that they could control as they wished.

Connections to Internet sites from the teacher's console, which one would not expect to be a problem, were also unreliable. The Internet site where the video was stored would sometimes return the infamous HTTP 404 error "Requested page not found" because the server was otherwise occupied or the location of the page had been changed. In the morning when network traffic was light, the files were usually easily accessible, but in the afternoon, when more people were using the Internet, the files would not load or would take a very long time.²

Multimedia File Quality

The inferior quality of the original video and audio tracks was another problem. Since file distributors want to minimize the amount of data being sent, they make many compromises in sound and video quality to reduce file size. The transmitted sound track might be encoded at 11 kilohertz, 8-bit mono (which should give adequate sound quality) and then highly compressed, resulting in very poor sound quality. Problems can also occur in the original recording. On one recording, for example, a verbal exchange was badly done because the microphone was held too far away from the people being recorded.

The video portion might also show signs of high compression rates which result in a very small, stamp size image (Godwin-Jones, 1997). Doubling the size of the image or enlarging it to the size of the whole screen makes the image so blurred that users cannot see any details, just colored pixels. Taken together, the video and audio problems make for a very unsatisfactory listening and viewing experience. When downloading multimedia files from the Internet, users encounter frequent pauses as the RealPlayer program waits for the data stream to come down. The video and sound track can be received in drips and drabs, full of stops and pauses, often in the middle of a word. Bush (1996) had already noted that students can become quite frustrated with the poor performance of the ongoing video stream especially because of the lack of synchronization between the lips and the sound.⁴

Size of Sound Files

Seven megabytes per song or video was generally the optimum size of file to be transferred within a local area network. If the size of the file was larger than seven megabytes, conflicts begin to interfere with the transfer when students all tried to access it at the same time. The Audioview 32 program, the sound editing software that was used for the recordings, had its default sound quality set to 44 kilohertz, 16 bit, stereo. These parameters created a sound file with size four times larger than it needed to be. A more satisfactory setting would be 44 kilohertz, 8 bit, mono, which created readily manageable files. Quite audible voices may be recorded at 11 kilohertz, and music pleasant enough to listen to may be recorded at 16 kilohertz. Bush (2000) offered a good suggestion that it is better to save audio or video files under the best format available for archive purposes and then make a second copy in lower quality for immediate distribution.

Problem of Copyright

Scinicariello (1997) pointed out that everyone working with multimedia has to be concerned with copyright issues (see also Pusack and Otto, 1997). The development and pervasiveness of MP3 files has further confused the already murky issue of copyright infringement. The wildly popular Napster and Gnutella programs make music files available to anyone without distinguishing those that are legally or illegally distributed. Downloading a music file for private listening that has also been made available by the copyright holders is certainly legal. However, copying the same file without permission to a local area network to make it available to students infringes on copyright. Some artists provide freely distributable MP3 samples of one or two of their songs. Many music files are made available on the Internet so that the music company benefits from advertising when visitors to their site download them. These files may only be downloaded from the distributor's site and may not be downloaded for any secondary use including educational purposes. As a general rule, it is prudent to assume that the use of any MP3 or video file infringes on copyright unless one has explicit permission to use it for educational purposes from the copyright holder. The Recording Industry Association of America (RIAA) (2000) states, "A copyright is infringed when a song is made available to the public by uploading it to an Internet site for other people to download, sending it through an e-mail or chat service, or otherwise reproducing or distributing copies without authorization from the copyright owner. In civil cases copyright infringement can occur whether or not money was exchanged for the music, and in criminal cases there only needs to be a possibility of financial loss to the copyright holder or financial gain to the infringer." (For an introduction to music copyright laws in the USA, see www.riaa.com.) As Dvorak, Charlotteaux, Gilgen, Herren, Jones, and Trometer (1995) noted the language teacher is not only faced with technology but also with the law.

DESIGNING MULTIMEDIA MATERIALS

It was necessary to find both audio and video multimedia files for use in the course. Three video segments were selected from the library of TV5 (www.funambule.com). These videos all dealt with the theme of le diable 'the devil:' L'Exorciste de Nice (an interview with an exorcist priest in the south of France), Le Violon du diable (a story about a Québecois fiddler), and the Les Sorcières de Marchiennes (about a festival of sorcery in the north of France). For the audio component, the following songs were chosen: L'Auvergnat by Brassens, Honte à qui peut chanter by Brassens but sung by Maxime Le Forestier, and Nuit et brouillard by Ferrat, all of whom are classic French folk singers.

While the activities for the songs were first being developed, an e-mail was sent to the singers' official web sites asking for permission to use their material in these activities. No response was ever received. As a result, these web pages were mostly used as proof-of-concept prototypes and only used once locally in a class to see whether students enjoyed working with them. Without the permission of the copyright holder, it is impossible to use these materials regularly or to post them to the Web. Consequently, new activities will be based on the music of less well known artists who are pleased to have the added attention. These new pages will only be developed when proper permissions are received. The web pages for TV5 and Funambule state that their videos may be downloaded and used locally for pedagogical purposes. In fact, their web pages were set up to promote learning the French language.

Outline of the Planned Materials

More activities were planned than were actually used in class. Originally, each class was to have an audio and a video activity, but time permitted only one or the other. The different activities developed around each audio or video segment took the students much more time than anticipated. (See the original schedule of multimedia programs prepared for the course and the changes that had to be made in the schedule in Appendix A.)

The students were introduced to the song with prelistening activities which were designed to activate their background knowledge and to give them information on the singer and the theme of the song. Students explored related links that were either stored locally or available on the Web.

The listening section had three parts. The first part was a true or false activity related to the general global comprehension of the song. The second part consisted of fill-in-the-blank questions on the lyrics. These questions were always accompanied by side notes giving historical, cultural, or literary details on the elements in the song, glossary explanations, texts and pictures, and Internet links to related topics. The third part of the section consisted of multiple choice questions focusing on detailed comprehension questions.

Developing Tools for the Server

Two different approaches were used to develop the language activities. Activities related to the video components were developed using the authoring tool Ficelle (aix1.uottawa.ca/~weinberg/ficelle.html), while all the audio activities were encoded in HTML and JavaScript.

Some authoring systems (e.g., ToolBook and Authorware) are difficult to manage without more advanced technical skills and support. Others (e.g., WinCalis from Duke University and Dasher form the University of Iowa) offer simple templates for the language teacher. The Ficelle program was developed in a partnership among Michael Davidson, Alysse Weinberg, and Hélène Knoerr to support the development of multimedia activities for second language learning. The program makes it very easy and fast to develop sophisticated fill-in-the-blank and multiple-choice activities for reading, listening, and grammar practice complete with automated generic feedback or customized context-sensitive feedback. The program allows developers to insert sound files, video files, pictures, review links, glossary items, and phonetic characters in lessons. Students can record their own voice and compare it to a standard recording. The program has been used extensively for language activities and for testing at the University of Ottawa where it has been installed on a multimedia network. (For a review of Ficelle, see Saury, DiBella, & Berg, 2001.)



Linguistic reality and technological limitations - brains vs chips

Thomas Hansen
University of Southern Denmark
Natural Interactive Systems Lab
thomas@nis.sdu.dk

Abstract

Behind the somewhat pompous name of this paper hides a much
more modest aim. It is an attempt at clarifying one of the major
problem areas found in the field of Computer Assisted Language
Learning (CALL), specifically the use of Automatic Speech
Recognition (ASR). The aim of CALL is to produce tools aiding
the acquisition of foreign languages and, long-term, to produce
full-fledged, virtual, multi-modal language tutors. In essence, the
teacher in the machine.
Focusing specifically on Computer Assisted Pronunciation
Training (CAPT) I intend to present a, hopefully, pedagogic
picture of how the mind works in relation to adopting the sounds of
a foreign language (L2), both perception- and productionwise, and
illustrate the shortcomings of ASR in comparison.
I also intend to present you with some rather unsubstantiated ideas
on language learning which can hopefully generate a wholesome
debate.
1. Introduction - and some fairly unsubstantiated claims open for discussion
Learning a foreign language is by no means a task which is accomplished easily. In recent years a trend has become visible which, within school systems, exposes children to foreign languages
earlier than ever seen before and the immigration politics of different countries focus much more on adult immigrants acquiring the native language as a prerequisite for permanent residence.
As the demand for faster, better and more autonomous language study increases, scientists have
begun looking at the computer as a potential aid in language acquisition. Some even speculate in the idea of an autonomous interactive, multi-modal language teacher/tutor. Although I find myself convinced that this will someday be reality, there are, as yet, multiple areas in which technology needs to improve in order to qualify for such a label, ASR not the least.
In my opinion the area in which computers can, presently and long-term, benefit learners the most is within the area of pronunciation. Very often in classes consisting of 20-40 people teachers will not have the available time to practice with and correct individual student pronunciation. Areas such as vocabulary and grammar tutoring lend themselves much more easily to a (blackboard) class-based teaching environment. In my opinion languages are not learned in class. Only building blocks aiding the learning of the language are provided within such an environment. One does not fully learn a language until one has to use the language in an actual situation with native speakers.
Furthermore, using a language entails possessing some rudimentary communication skills. It is my firm belief that three different priorities for language acquisition can be established1:
1. Vocabulary
2. Pronunciation
3. Grammar and all the rest.
 This claim was partly reached through a conversation with a Chinese test subject who said, “I know everything about
the grammar, I just can’t for the life of me pronounce it.”
In order to be able to speak a language a learner must know some basic words. Without a
vocabulary nothing else will work. Again, in order to use that vocabulary a learner needs to be able to pronounce the contents. Subsequently learners can be taught to arrange words in their proper
position. Whether a student utters the sentence: 1) “Me beer want,” or 2) “I would like a beer,” he or she will still be understood. This is comparable to tourists visiting foreign countries and availing themselves of standard phrases which have been picked up in one way or the other.
Learning a language in class is much like learning to drive a car. Initially you are presented with all the rules and interpretations of the signs, but it is not until you find yourself alone, driving the car on your own in actual traffic that you really learn how to drive.
Computer Assisted Language Learning, in time, carries the potential to provide you with your own car, but there are many hurdles which need to be overcome before this becomes reality.
2. Computer Assisted Language Learning
Computer Assisted Language Learning (CALL) and Computer Assisted Pronunciation Training
(CAPT) are areas in the academic field that explore the role of information and communication
technologies in language learning and teaching. This includes: materials development, pedagogical practice and research. Both are interrelated areas that have experienced rapid growth in recent years.Especially the introduction of the internet and the computer as a household item has spawned the possibility of online learning and computer applications in language learning [Levy, 1997].
Today CALL/CAPT have established themselves as prolific areas whose advantages are well
known to language educators. As a matter of fact I believe that this area should be viewed as a branching tree with a headline and several subdivisions, depending on your focus. CALL seems to me to be initial node, CAPT comes directly beneath it on its own branch. Other branches could be areas of grammar learning and vocabulary learning.
 See for instance (Neri, Cucchiarini, Strik & Strik 2002).
1) Training available at any time the learner wishes – (install on laptop and transport)
2) More and faster individualized feedback for the learner. – (only the learner uses it)
3) Stress free environment for the learner to practice in. – (nobody listens, no time limit)
4) More practice time than in a classroom setting. – (available anytime)
The above four points should be viewed in terms of a class setting versus a non-class setting.
But, as pointed out by Neri [Neri, 2001], many are concerned with the missing link between
technological advances and Second Language Acquisition (SLA) research. One task at hand lies in bridging the gap between proceduralists and formalists, those who wish to build applications versus those with knowledge of language [Chambers, 2001]. In essence a method for pronunciation training is needed which focuses on both the linguistic perception and production facets of learning a language as well as taking into consideration the shortcomings and possibilities of present day technology.
Perception and production
Nearly all people are born with the capacity of speaking one language with native competence (L1), but children and adults differ greatly in their ability to learn a second language (L2). To find bilingual children with the competence of speaking both languages with native qualities is not unusual and even throughout their initial years of education they may acquire a third or fourth language where they attain near-native qualities. Research shows that around the age of thirteen the capacity for learning a language seems to diminish, otherwise known as the Critical Period
Hypothesis (CPH) [Lenneberg, 1967; Flege, 1987]. Upon reaching adolescence our capacity to
interpret new phonetic input (i.e. sounds not indigenous to our native language) seems to fossilize. This is evidenced in many adults attempting to acquire a second language but producing it with heavy accent coloring. In 1957 Noam Chomsky coined the term Universal Grammar (UG) [Chomsky, 1957] arguing that all people possessed an underlying language module which acted as a blueprint for acquiring language. Once all the sounds of the native language have been ‘put into place,’ they are grouped in acoustic categories appropriate for the L1.5 Heavy accent coloring in L2 acquisition has then hitherto been explained via Equivalence Classification [Flege, 1987]: Upon attempting to acquire the sounds of the L2, these pass through the ‘phonological filter’ of the L1 and are assimilated to the sounds already known. In other words, the learner will attempt to compare the new sound with an already existing one in his or her native language, and produce the already known sound rather than the new one. This is also known as Acoustic Assimilation. Hence, a necessary component of speaking a foreign language with native quality is the ability to perceive the phonetic differences between the mother tongue and the target language. Additionally the ability to produce the sound is dependent on the ability that a learner must be able to perceive the difference in acoustic quality before he/she can produce it and thereby make room for new acoustic categories for the language in question.6
There are, of course, numerous other processes involved in the acquisition of a foreign language
besides the ones that relate specifically to speech viewed as a simple acoustic signal. These relate to extra-linguistic factors such as body movement, but are not treated in this paper.
It should be noted that the theory of Universal Grammar is highly controversial and opposing viewpoints can be found in most of the linguistic literature by J.R. Firth and M.A.K. Halliday. I am not an avid proponent of UG theory, but it does seem to provide a pedagogically sound picture of the elements involved in language acquisition. Acoustic categories can here be thought of as those ‘vowels’ and ‘consonants’ that we view as a fixed inventory for a particular language. Other languages will posses different acoustic qualities.
There are opposing views on the matter such as Borrel [Borrel, 1990] who claims that production precedes perception. Mastering the motor mechanics (in relation to the tongue), should facilitate the perception factor. My personal view, based mostly on simple intuition, disagrees with that statement.
4 CALL – ASR and all the problems we encounter. Today the most widely used piece of technology in CAPT applications are Automatic Speech
Recognition (ASR) systems. ASR allows the L2 learner to interact with the constructed software as well as provides the possibility of the learner’s pronunciation to be understood and evaluated
immediately, by analyzing the acoustic signal provided by the learner. In an ideal CAPT application the level of feedback should emulate that provided by human teachers in terms of, for instance, segmental correction. A teacher standing in front of a student is able to offer corrective feedback in respect to motor movements of the mouth, articulatory advice, guide the student on the phonetic as well as the sentence level. Hence, a non-native speaker pronouncing the word as should be informed that the second segment should be pronounced differently, followed by exemplification. Currently the best available option in terms of computer feedback is provided through the use of ASR systems. However, numerous problems are encountered when employing ASR technology in CAPT systems. These problems are manifold and can be exemplified on, at
least, two separate, yet interdependent, levels:
1. Single word pronunciation
2. Sentence level pronunciation
ASR systems, unlike the human auditory system, are currently incapable of distinguishing between and non-speech sounds. Hence a cough, a sigh, a hiccup, the slamming of a door, the chirp of a bird, all these sounds will be interpreted by the ASR system as constituting a speech signal and the system will attempt to match this signal with an already incorporated vocabulary. Even the possibility of having a faint echo in the room when producing an utterance will potentially be misinterpreted by the ASR.
Limited vocabulary, or simply out-of-vocabulary, constitutes another problem for ASR systems.
Most often the training material available to the student is fairly limited since not all words,
counting inflections and similar, are part of the trained or inbuilt vocabulary. This in turn brings up a related problem, namely that of dialectal variation and speaking style. Dialects can in some cases vary to such a degree that they can almost be considered languages within languages. Even the human auditory system can run into trouble at this level. Differences are also visible in terms of physiological differences in the human vocal tract.
In addition to dialects, we also encounter differences in speaking styles as well as differences
between the acoustic signals of men, women and also children. In many cases ASR systems are not capable of handling these differences unless specifically designed to cater to a particular group.
Furthermore ASR systems often run into trouble when the vocabulary becomes too similar. Hence, minimal pairs such as and are very difficult to distinguish. Most ASR systems seem happier when the phonetic material provided is polysyllabic. Even more difficult, if not entirely impossible is the aspect of getting ASR systems to provide segmental feedback. Segmental feedback requires the isolation and identification of single-phones in a context of other phones. Hence the system should ideally be able to recognize whether a [k] is pronounced with aspiration or without aspiration, or even as non-released. ASR systems are composed, mostly, of diphones or triphones, hence eliminating the possibility of single-phone identification. Creating an acoustic model which would be able to distinguish between all the different phonetic realizations of sounds does indeed provide a daunting task.
The already mentioned problems are then even further magnified when examining continuous speech. Continuous speech has no apparent boundaries from which the ASR system can solidly deduce single words. This can be seen from perhaps the most widely used example listed below.
How is an ASR system to know whether a speaker says:
1. Recognize speech
2. Wreck a nice beach
As human listeners we have the advantage of context and reasoning, an ASR system does not.
Further examples can be found in the area of homophones, which are words that sound the same but are spelled differently i.e. sale versus sail. All in all speech recognition still suffer from severe shortcomings in terms of functioning as substitute and autonomous language teachers.
5 Conclusion
In this paper I have attempted to outline two very different processes, which aim at achieving the
same goal, namely that of creating new speech sounds in learners of a foreign language, perception and production of the mind versus perception in ASR.
As we have seen, acquiring new speech sounds is by no means an easy task once a native language has become fossilized. Human teachers have the ability to tutor and monitor individuals as well as correcting minuscule differences between the mother tongue and the L2, but to some degree they suffer from not having enough time to devote to each individual students. ASR, if viewed as part of a larger package, does contain the possibility of providing students with more time for learning and individual practice and correction, but are simply not able to generate feedback which is finegrainedenough to highlight small differences in language sounds. Furthermore, ASR systems as standalone applications are entirely unable to provide the learner with the perception part of language learning. ASR does currently provide the best option in terms of generating feedback, but we are a long way away from creating alternatives to real-life teachers.

Applied linguistics

Main article: Applied linguistics

Whereas theoretical linguistics is concerned with finding and describing generalities both within particular languages and among all languages, applied linguistics takes these results and applies them to other areas. Often applied linguistics refers to the use of linguistic research in language teaching, but this is just one sub-discipline:

• Research in language teaching: today, 'applied linguistics' is sometimes used to refer to 'second language acquisition', but these are distinct fields, in that SLA involves more theoretical study of the system of language, whereas applied linguistics concerns itself more with teaching and learning. In their approach to the study of learning, applied linguists have increasingly devised their own theories and methodologies, such as the shift towards studying the learner rather than the system of language itself, in contrast to the emphasis within SLA.[13][14]

• Applied computational linguistics: two computer applications are speech synthesis and speech recognition, which use phonetic and phonemic knowledge to provide voice interfaces to computers. Machine translation, computer-assisted translation, and natural language processing are fruitful areas which have also come to the forefront in recent years.

• Clinical linguistics entails the application of linguistics to speech-language pathology. This involves treating individuals whose linguistic development is atypical or impaired.[15] This branch of applied linguistics may also involve treatment of specific language impairment, where one aspect of language develops exceptionally.[16] The field has also adopted existing ideas which have have not become 'mainstream' in theoretical linguistics. For example, both behaviourism[17] and natural phonology[18] have appeared in the literature.

Man and the Media II: Media and Cross-Cultural Communication in Foreign Language Learning A Conference Report

Hans-Wilhelm Kelling
Brigham Young University
Henry Niedzielski
Universität Bonn

This year the second annual symposium Man and the Media cosponsored by AILA and CETALL was held at the Institute for Romance Languages at the University of Saarbrücken, Germany from September 16 to September 19. The symposium was convened by Udo Jung of the University of Marburg and organized by Wolfgang Bufe of the University of Saarbrücken and Viktor Burowsky of the University of Bonn. Because of its emphasis on cross-cultural communication and Saarbrücken's immediate proximity to France, the conference attracted a large number of French speaking participants from France, Switzerland, Denmark, Hawaii, Hungary, the Netherlands, and Poland as well as German and English speakers from Germany, Great Britain, Ireland, Malaysia, Sweden, and the United States.

Presentations during the first day concentrated on comparisons between German and French news coverage on TV and in the press. Because of his multifaceted experience as a language teacher at the University of Limoges, as a research specialist at the B.E.L.C., and especially as a journalist attached to the national France-Presse Agency, Amr Ibrahim shared many ideas which are not usually considered by members of the academic community. He explained, for instance, how most international information relayed by mass media is standardized for all languages and, therefore, void of any cultural content. This presents a problem for language educators and also for the average viewer who becomes more and more attracted to extremely demagogical politicians who emphasize local or national ideologies.

Charles de Margerie, a foreign language teacher trainer and research associate at the C.R.E.D.I.F. addressed some aspects of interactivity—or the possible lack of it—in language learning with computers and other machines Because the individual learner usually interacts only with the program, in computer assisted instruction, he suggested that new strategies should be developed to generate inter-group interaction.

Ekkehard Eggs and Beate Gerken, both from Hannover, addressed the problem of plausibility and credibility in television reporting and compared standard news and interview coverage of French and German TV. Eggs contended that credibility is produced by means of techniques which take into consideration the sociological and cultural tradition of the country. Gerken demonstrated such techniques with an example of a political interview on camera of a popular French and a popular German politician by a French and by a German TV crew. She pointed out special effects created by linguistic, gestic and mimic devices employed by the politicians with the aim of producing credibility. Richard Batz of Berlin continued the discussion of television news coverage by pointing out the didactic nature of French news. He had analyzed approximately 80 French news reports and discovered certain didactic devices used by French TV which were used to inform the public and keep up interest. One interesting example showed how the didactic intent was so overwhelming that the TV news program showed in great technical detail how a gang of bank robbers was able to gain entrance into bank vaults. The explanations and visual demonstrations were so detailed that they could be considered a most useful model for robbing a bank. Interesting was also the technique of introducing the major news items at the front of the newscast in the form of riddles which invite the viewer to become actively involved in the solution of the riddle by watching the ensuing coverage.

Georgette Stefani-Meyer of Saarbrücken, restated the importance of the press in the acquisition of communicative competence. She emphasized the role of a skill that is much too often ignored in foreign language teaching, and yet indispensable for spontaneous and natural communication: the language socializing mode.

Letitia van den Heuvel of Tilburg offered a new too] for the teaching of language, culture, and literature: literary criticism published in newspapers and journals, She researched this subject extensively and involved the participants through a short questionnaire, asking everyone what had determined the choice of their most recent reading.

The day ended with an evening reception by the Vice President of the University of the Saarland.

The second day was opened with a lecture by Alois Palzer of Mainz on the subject of the importance of both the eye and the ear in language acquisition. Recent research has shown that language learning already begins in the mother's womb from the seventh month on and continues most effectively until the child is seven years old. Palzer pointed out the great importance of media in addressing both the eye and the ear in teaching language. He added, however, that teachers and researchers must also take into consideration emotional factors on the part of the learner as well as classroom structure and teaching techniques. We do not yet know enough about the emotional background of the learner to help us in the development of better teaching techniques.

Claude Gauthier and Beat Münch of Basel presented an interesting TV program showing how German speaking students of the French language planned and produced cultural vignettes in French-speaking Switzerland. The various interviews were made with individuals from several social and professional groups and were of high technical and pedagogical quality.

Isabelle Durousseau of Copenhagen continued the discussion and reported how she also involves her Danish students in planning, creating, and presenting a French newscast for the French TV programs in Denmark. The students find their material in French newspapers and on Danish TV. Their styles of presentation imitate those which they have observed on video tapes of French newscasters. Both Gauthier/Münch and Durousseau report that their students are highly motivated.

Thomas Szende of Budapest has obtained a series of VHS videotapes containing sequences of newscasts from the French TV network TFI directly from the Institut National de la Communication Audiovisuelle. Szenda has excerpted some scenes in order to isolate certain sociocultural features and thus to facilitate the socio-linguistic learning process in its evolution from a controlled situation to creative, spontaneous communication.

Pawel Plusa of Czestochowa reviewed the teaching of French as a foreign language on Polish public television since 1965. He compared several courses which were broadcast during this twenty-year period and showed how they were adopted in Polish secondary schools. The latest courses have been copied on videotapes and can be used both at school and at home in conjunction with public television.

In the late afternoon the group had the opportunity to participate in a sight seeing tour of the city of Saarbrücken.

The third day was opened with a report by Wolfgang Bufe of Saarbrücken on his investigation into the phenomenon of anxiety in foreign language acquisition by adult learners. He based his observations in part on his experiences with German evening school students who are studying French. As part of the acquisition of the French language, Bufe takes students across the border into France and affords them the opportunity to function in the country of the target language. With the help of a video film he showed that anxiety can be reduced by exposing learners quite early during their schooling to the foreign culture on location in the country. He found that in his experimental group, anxiety was reduced through personal contact with Frenchmen living in a small farming community. This was especially the case when the German students began to realize that the French also experienced a certain amount of anxiety about meeting German "intellectuals."

Relying on 15 years of experience with the British and German broadcasting media, Leo Koesten of Saarbrücken, together with professionals from the Westdeutsches Fernsehen has produced five 30 minute documentaries on Brittany. Because he established an excellent rapport with the natives before he began the actual filming, the final product is completely authentic. Through these films Koesten is able to share his own views and feelings for this part of France.

Adile Challe of Paris, a successful author of French textbooks, gave a fascinating presentation of her polychromatic review of French TV programs and the numerous possibilities they offer to teachers and learners of French. Henry Niedzielski of Hawaii and Bonn reported a successful experiment he has conducted with his students at the University of Hawaii on teaching genuine French gestures and facial expressions with the aim of fostering better understanding of the French language and culture. He teaches this technique with the help of videos and slides. Ulrich Hann of Bonn showed how video films can sensitize language learners and teachers about cultural differences between countries. With film clips taken from Pakistani television programs, he demonstrated how the perception of Moslem students from Pakistan concerning the subjects of religion, eating and drinking, and the relationship between the sexes varies quite considerably from the perception of German viewers. He raised the question as to how we are to present the native culture to learners coming from a greatly different culture and what type of video material provided readily by information agencies might be appropriate.

David Little of Dublin introduced the autotutor which he and others have been developing, a video device for self instruction. This device allows for computer interaction with a video component and has interesting possibilities for both the teacher and the learner of a language. Zofia Jancewicz of Warsaw contended in her lecture that intercultural understanding releases and facilitatescommunication. She showed English language material produced by the BBC which she adapted for language learning in Poland.

On the last day, Sarah Gill of Malaysia opened the proceedings with a report on research she is conducting as part of her Ph.D. dissertation at the University of London. She feels that traditional video materials are not necessarily adequate in teaching English as a second language to students in Malaysia because the cultural content may be distracting to them. Such culture-based materials may not be appropriate in countries like Malaysia where English is used as a language of communication between speakers from various Asian cultures. What kind of video materials might be best suitable and motivating to language learners who are not primarily interested in the culture of the target language but who want to use the language as a means of international communication?

According to Soia Koester-Thoma considerable difficulties are encountered by the Slavic Language Institute of the University of Berlin in finding adequate video materials for teaching Russian. She rejected most available material from the Soviet Union because of its obvious propagandistic nature.

Philippe Durant de Saint Andre of Paris and Poitiers discussed various exercises presently used in CAT and showed how they, unfortunately, limit interactivity and creativity, two fundamental elements of communicative competence. He proposed new activities partly based on cloze techniques which two French research groups are currently developing for the teaching of German. They will enable the learner to use newly acquired vocabulary and to compose paragraphs creatively. The available software can presently be used with Apple Writer, MacIntosh, and Word Star.

A clever analysis of German Videotext was presented by Sonja Klug of the University of Bonn. In her research she is analyzing German Videotext to determine how difficult and how appropriate the text is for language learners. She wants to find out which linguistic aids a teacher must give to his students before they can understand selected television programs. Her study will also determine if television is more appropriate for intermediate or advanced learners and if there exists a relationship between regular newspaper reading and the understanding of Videotext. Robert Baker of Southampton reported on his involvement with the development of both hardware and software materials suitable for teaching English to deaf children in Great Britain. He reported that the cost for video equipment has been substantially reduced so that it is now more readily available for teaching the deaf. Television and video recorders are a distinct blessing to the deaf whose chances of learning language are greatly enhanced through the use of these devices. The use of word processors and interactive video will even be a greater blessing to these handicapped persons in the future.

The final presentation was made by Hans-Wilhelm Kelling, the only participant from mainland USA. He demonstrated a computer interactive video model that was developed by Brigham Young University and gives the learner the option of constant interaction with a video. The video films, shot on location in Germany, present everyday situations which the foreigner encounters. A computer is connected to audio and videodisc players in such a way that selected audio and video sections can be quickly and precisely accessed as needed. The computer accepts the student's input and determines what should be presented to each student. The interaction between the student and the computer is tailored to each student's need, and the individualized instruction functions like a very good tutor.

Thanks to excellent preparation and sensitive guidance during the symposium primarily by Wolfgang Bufe, a pleasant spirit of interaction and learning developed during periods of questioning, commenting and breaks. Man and Media II was a resounding success, and we are looking forward to next year's conference in Australia.

Article: Linguistic-No Easy Answers in Evolution of Human Language

March 9, 2008 

Linguistics Anthropology is the study of human language since that is a characteristic that makes humans unique. This article shows how bird songs in some birds are now helping in gaining a better understanding of the development of human language. This brings together biologic, culture, archaeological, and linguistics together with science to learn more about the complex social interaction of human language and communication. ENJOY, I did and learn about new bread discoveries in linguistic anthropology.
ScienceDaily (Feb. 21, 2008) — The evolution of human speech was far more complex than is implied by some recent attempts to link it to a specific gene, says Robert Berwick, professor of computational linguistics at MIT.
Berwick will describe his ideas about language in a session at the annual meeting of the American Association for the Advancement of Science on Feb. 17. The session is called “Mind of a Toolmaker,” and explores the use of evolutionary research in understanding human abilities. Some researchers in recent years have speculated that mutations in a gene called Foxp2 might have played a fundamental role in the evolution of human language. That was based on research showing that the gene seems to be connected to language ability because some mutations to that gene produce specific impairments to language use, and because our closest living relatives, the chimpanzees, lack both these gene mutations and the capacity for language. But the claim that the gene mutation is directly connected to the development of language is very unlikely to be right, says Berwick, who holds appointments in MIT’s Department of Electrical Engineering and Computer Science and Department of Brain and Cognitive Sciences. “This kind of straightforward connection is just not the way organisms are put together,” he says. When it comes to something as complex as language, “one would be hard-pressed to come up with an example less amenable to evolutionary study.” And the specific Foxp2 connection is based on a whole chain of events, each of which is speculative, so there’s little chance of the whole story being right. “It’s so chaotic, it’s like weather forecasting,” he says. “The noise overwhelms the signal”. Rather, language is almost certainly the result of a far more complex and subtle interplay among a variety of factors, Berwick says, and it may never be possible to connect it to specific genetic changes. “There are some things in science that are very interesting, but that we’re never going to be able to find out about,” he says. “It’s a sort of romantic view some people have, that anything interesting can be understood.”
Even defining something as complicated as language in a precise way is daunting, as ongoing disputes over the significance of language experiments with apes, parrots and dolphins have made clear. Berwick says, “If you can’t define what it is, why study it from an evolutionary point of view?”
It’s more likely, Berwick says, that the role of the Foxp2 gene in language is somehow peripheral to the capacity for language itself. He compares it to a printer in a computer system–its part of the overall system, but it’s not fundamental. Berwick thinks a more productive approach to studying the evolution of language is to examine it in terms of deeper, internal mechanisms. In his own research, Berwick has compared the structure of languages with the structure of bird songs, and has found interesting connections that may lead to a better understanding of some aspects of language.
Both bird songs and all human languages seem to share some underlying characteristics related to their metrical structure, Berwick s ays. There’s an underlying sing-song beat that is pronounced in poetry, music and in the songs of birds that may reveal a fundamental aspect of how our brains process language. Future research could probe this link further, even looking at possible connections between other specific genes, in both birds and humans that might be connected to this sense of metrical structure. Ultimately, the important thing is to understand that language is, at bottom, something that takes place inside the human mind and is independent of any particular sound, sight or motion. The same internal mental construction could be expressed through verbal speech, through writing or through sign language without changing its basic nature, Berwick says. “It’s not about this external thing you hear,” he says. “It’s about the representation inside your head.”
Adapted from materials provided by Massachusetts Institute of Technology