Basic skills training

From LearnLab
Jump to: navigation, search

French dictation training


The goal of this project is to improve the ability of students of Elementary French to write down sentences as they hear them. Mastering this ability involves improvements in vocabulary recognition, auditory range and precision, and control of spelling rules. Like other studies conducted by MacWhinney and Pavlik, this work emphasizes the role of scheduling in attaining mastery. This practice is now an in vivo component of French Online.


Research question

This research is designed to discover the best method of producing robust learning of French dictation. Toward this end, there is an emphasis on developing materials based on error analysis to configure correct stimulus sequencing. The specific online method examines the role of sequencing, explicit feedback, and lexical familiarity in promoting dictation skill.

An important component of this online tutor is its ability to provide direct feedback on student errors. This is done by placing asterisks over those segments of the student's dictation that fail to match the target.

Background and significance

Both L1 and L2 classroom instruction in French often uses a system of dictation that seeks to teach clear recognition of words, phrases, and sentences along with proper spelling. In classroom dictatin, the teacher will pronounce a series of phrases or sentences and ask the students to write them out. The sentences are typically designed to provide information regarding orthographic ambiguities such as the contrast between the masculine adjective intelligent, the feminine form intelligente, or the plural form intelligentes. This single classroom activity incorporates four types of learning at once:

  1. Students must be able to recognize the words between produced by the teacher. For L1 learners, this part of dictation is relatively easy, but L2 learners may find two types of challenges here. First, L2 learners often find that French is difficult to segment into words because of frequent elision patterns and the reduced, monosyllabic nature of high frequency forms.
  2. Second, word recognition may be difficult in some cases for L2 learners because they have not yet learned the relevant vocabulary.
  3. Once words are recognized, learners must be able to map them to standard orthographic patterns. For learners with English as L1, the use of diacritics and the proper encoding of vowel patterns is a particular challenge.
  4. Although French spelling is largely regular, many spelling patterns depend on grammatical features such as gender, number, and verb conjugation. In order to achieve proper dictation, learners must be sensitive to these features.

After pilot testing in Fall 2005, the FOL (French Online) course now incorporates a Java-based dictation template that logs directly to DataShop. The program gives immediate feedback regarding correctness, but we have not yet provided feedback regarding the technicalities or principles of spelling.

Dependent variables

Normal post-test measures:

  1. The dependent variable is percentage correct. Correctness is scored on the word level. There is no penalty for incorrect words. Scoring is done by a Perl program.
  2. In addition to the overall analysis for percentage correct, we are also closely tracking error types within words. We are interested in specifying closely the spellings rules and auditory patterns that are most difficult for students.

There are no explicit robust learning measures in this study. However, a major finding was that improvements from dictation training were across-the-board and not confined to sentences used in the training. Thus, it appears that dictation trains a totally general skill.

Independent variables

  1. We are using a pretest-posttest design to measure the overall effects of the online training. We compare gain scores from students in the traditional course with no dictation training with gain scores for students in the online course with dictation training.
  2. We are also tracking the effects of exposure to particular sentences. In each lesson, half of the students study one list and half study another. We then test generalization across lists.


  1. Our initial hypothesis was that exposure to a particular set of sentence would markedly improve the ability to spell these sentences in comparison with a parallel set of unexposed sentences. In the first round of testing, we found that this was wrong, indicating that the skill of dictation is not pegged to sentence level memories, but rather to the level of the phoneme and word.
  2. We have found a clear increase from pretest to posttest in dictation ability across the course.
  3. We further hypothesize that dictation training in French 1 will markedly improve dictation ability in French 2.
  4. We hypothesize that the most difficult dictation patterns will be those marked by use of diacritics, unreliable spelling patterns, the apostrophe, and difficult auditory sequences.
  5. When French words match up closely with English words in sound and meaning, but differ in spelling, we will expect some transfer errors.

These predictions derive from the Competition Model (MacWhinney, in press).

Errors in dictation often produce alternative sentences that are still fully meaningful. For example "Je ne nage pas, moi" (I don't swim) can be misproduced as "Je ne nage pas mal" (I don't swim poorly). Or "Il faut faire tes devoirs" (It is necessary to do your duties) is produced as "Il forte de duboise" (It strong of Dubois).


The Competition Model explanation for these effects emphasizes the role of L1 transfer, cue reliability, cue availability, and lexical learning as determinants of dictation learning. Availability and reliability are measured across the vocabulary. L1 transfer effects are predicted on the basis of a comparative analysis of French and English.


Annotated bibliography

  • Bonin, P., Fayol, M., & Pacton, S. (2001). La production verbale écrite: évidences en faveur d'une (relative) autonomie de l'écrit. Psychologie Francaise, 46, 77-88.
  • Bonin, P., Fayol, M., & Gombert, J. (1998). An experimental study of lexical access in the writing and naming of isolated words. Inernational Journal of Psychology, 33, 269-286.
  • Content, A., Mousty, P., & Radeau, M. (1990). Brulex: Une base de données lexicales informatisée pour le français écrit et parlé. L'Année Psychologique, 90, 551-566.
  • MacWhinney, B. (2006). A unified model. In N. Ellis & P. Robinson (Eds.), Handbook of Cognitive Linguistics and Second Language Acquisition. Mahwah, NJ: Lawrence Erlbaum Press.