Submitted by jenny on Mon, 05/07/2012 - 06:40
-
On the errors we make, and their patterns. We all make mistakes in our language—even the best of us. People who are learning a language make many errors. In most, if not all cases, there is a pattern to these errors.
-
On error analysis—theory, use, practical applications, and the limitations of reality. The field of “error analysis” (within the larger field of education) is a very big subject, beyond the scope of this paper. However, I will say this: you can determine something of a learner’s “process of learning” by studying their errors, for these errors often fall into a pattern. It is a little bit like an individual fingerprint. You can use this fact to find out where a student needs help: if you record every error a given student makes in an “error/correct answer” format, you have the raw information you need to help plan that student’s (remedial) education. It is very valuable information! However, since it takes a lot of time to write down, sort and analyze the many errors a student makes, it is not a realistic venture. Imagine the task of doing this work for every student in a 50-student class! Therefore, few teachers use error analysis to its full potential.
-
Error analysis work in the English language: potential and challenges. When one analyzes the errors a student makes, into what categories should the errors be arranged? The English language, having letters, can be broken down to individual letter sounds. English has 26 letters. English is about 85% phonetically regular, and 15% phonetically irregular. Therefore, there are many categories of errors one can put a student’s reading errors into, even if one only takes pronunciation into account, and not errors of grammar. If a patient teacher (with a pen, notebook, and plenty of time) is willing, then a student’s errors can be noted and analyzed. However, it is hard to program a computer using “voice-recognition technology” to do this task, because there so many variables to choose from! Therefore, a computerized, automated system of reading-error analysis is as yet unfeasible in the English language. (It would be wonderful if such a program were available.)
-
On error analysis in the Chinese language: potential and opportunities. With the Chinese language, this problem is greatly simplified. The Chinese language, having “characters” (Han Zi), exists at the syllable level. (Although the “romanization system”—Pin Yin—has letters, it too can be looked at in terms of syllables, not letters.) Therefore, Chinese is basically a language of syllables—and very simple ones at that. The Pin Yin “table of syllables” (Han Yu Pin Yin Biao) is quite large, especially when the four “tones” (Si Sheng) and the “neutral tone” (Qing Yin) are factored in. However, the number of existing syllables is finite. Pin Yin is also 100% phonetically regular; there are no phonetic irregularities under any conditions. Therefore, it is very easy to put a student’s reading error into a certain category. Comparing a student’s utterance with a finite range of choices is very simple, especially if one programs a computer using voice-recognition technology. Therefore, a computerized, automated system of reading-error analysis is, I believe, feasible in the Chinese language. It would make certain aspects of classroom management and reading assessment much, much easier, especially since there are so many students who need to be helped.
-
Proposal for an automated error analysis system to serve large populations of Chinese-language students, by adapting current “language-recognition” software technology. In “language-recognition” software, the computer “recognizes” something you say and writes it for you. Under this new system, the reverse happens. You start with a pre-determined text, say, 500 words in Chinese (Han Zi), which is already in the computer. The student reads the text. The voice recognition software compares the words the student reads with the text itself…word for word. If a word is right, then O.K., no problem. If the word is wrong, then the software writes down the error and the correct answer in the following format—(error/correct response). Pin Yin “tones” (Si Sheng) are also included. For example, if you read “lao” (third tone), but the word is “lou” (second tone), the results will be written (lao 3 /lou 2). It is a simple software programming challenge to come up with this result! This “raw data” –along with other errors – can now be sorted and put into error categories. This too is a simple software programming challenge. When you have a student’s reading errors categorized and quantified like this, it is then very simple to plan out remedial instruction –you now know where to look for trouble. Using this software, you can quickly survey a whole class of students, or you can get exhaustive data on a few students with reading problems. It is therefore useful for reading assessment or special education applications. I hope some educational software company, or someone from Qinghua University, or someone else designs such a product! There are many directions in which this software can be built upon and improved, but that is a problem for another day.
