Learning Grammar: The SuperMemo Method

New article by Dr. Piotr Wozniak: The true history of spaced repetition

When learning grammar, we address the 3 primary areas of skill and memory:

  • rules: the rules of the grammar that determine what constitutes a correct sentence in a given language
  • facts: facts of the grammar that include a usually huge body of exceptions (except for artificial languages), pragmatics, etc.
  • procedures: other memory components involved in composing correct speech incl. automaticity, phonetics and phonology, agglutination, etc.

The knowledge of rules and facts is declarative in nature, while the knowledge of the procedures falls into the area of procedural learning. This means that these should be treated in a different manner as far as the consolidation and retention of memory are concerned. The application of SuperMemo has one primary goal: the acceleration of learning as well as impeccable retention. Whereas children learn the grammar through innate neural inductive acquisition that makes it possible to extract, deduce, and abstract the rules of the grammar in a primarily procedural way, the SuperMemo method is based on the assumption that the student intends to achieve a substantial linguistic competence and fluency within a fraction of the time that it takes a child to learn a language. This is why the emphasis in learning shifts from the natural inductive procedural acquisition to an abstract and instantiated declarative learning based on spaced repetition.

SuperMemo can be used in both declarative and procedural learning. However, the true power of spaced repetition shows primarily in the declarative area. This is why the most efficient employment of SuperMemo in learning grammar involves the following areas:

  • learning the rules of the grammar via models, i.e. learning the rules and their implications in a theoretical textbook manner
  • learning the rules of the grammar via instantiation, i.e. memorizing a sufficient body of exemplary rule instances that amplify the theoretical knowledge
  • learning the body of facts associated with the grammar of a given language
  • learning the comprehension of grammatically correct sentences
  • learning the pronunciation and intonation of grammatically correct sentences

The above areas do not include a vital component of learning a language: the actual use of the language. This is where the use of SuperMemo is limited. The subdivision of knowledge and skills involved in learning grammar as well as the limitations of SuperMemo translate into the following formulation of the SuperMemo method of learning grammar:

  1. With SuperMemo: Formulate and memorize the declarative body of knowledge associated with a given grammar (rule models, rule instantiation, and the body of factual knowledge)
  2. With SuperMemo: Train the auditory and verbal procedures associated with the grammar
  3. Without SuperMemo: Use the acquired knowledge in writing, speech, comprehension, etc.

Examples

Note that all examples use English for learning English grammar. This is not an artifact reflection of the fact that this text is written in English. As in other cases of learning a language with the SuperMemo method, it is highly recommended the the entire instruction proceeds with the use of the target language (as early as possible). This is contrary to some other methods of learning grammar. Learning more than one language will always be affected by a powerful inter-language interference. It is therefore necessary to produce a strong monolingual conditioning to ensure easy and complete transition from one language to another. The more languages the student takes on, the greater the total interference (i.e. the extra body of knowledge and skill needed to separate all languages from each other at the neural level). Monolingual conditioning helps the inductive procedural learning that ensures that a separate set of neural procedures is applied for each independent language.

Learning the rules via models

An efficient way of learning the rules of any grammar is to use electronic source of knowledge and apply incremental reading.

In incremental reading, the following rule:

  • Adverbs are most usually placed at the end of a phrase. Time adverbs (yesterday, soon, habitually) are the most flexible exception. "Connecting Adverbs", such as next, then, however, may also be placed at the beginning of a clause. Other exceptions include "focusing adverbs", which can occupy a middle position for emphasis.

is converted to a cloze deletion:

  • Q: Adverbs are most usually placed at the [...] of a phrase. Time adverbs (yesterday, soon, habitually) are the most flexible exception. "Connecting Adverbs", such as next, then, however, may also be placed at the beginning of a clause. Other exceptions include "focusing adverbs", which can occupy a middle position for emphasis
  • A: end

which is incrementally converted to a typical SuperMemo item:

  • Q: Adverbs are most usually placed at the [...](position) of a phrase
  • A: end

Learning the rules via instantiation

Rule instantiation can best be accomplished by taking real-life examples and converting them into a Q&A format. For example:

  • Q: gram: This soup (taste) better if it had more salt in it
  • A: would taste
>Learning the facts

Factual body of grammatical knowledge will contain a great deal of exceptions. For example, irregular verbs:

  • Q: the past participle of: sting
  • A: stung

Learning the facts via examples

An irregular verb presented earlier should also be used in example sentences to consolidate various context permutations:

  • Q: Yesterday, I was (sing) by a wasp.
  • A: stung

Learning comprehension

SuperMemo provides a rich set of tools for learning comprehension. One of the recently added tools is incremental video. For the sake of the present article it is enough to note that incremental video makes it easy to extract portions of video for review, and subject them to a standard SuperMemo learning process. In practise, this means that a student can compile a body of favorite videos, and regularly review video fragments that contain phrases that are considered most valuable or difficult in boosting his or her own comprehension. Practise shows that extracting difficult fragments and reviewing them regularly leads to quickly plugging up the gaps in phonetic competence. It is hard to compile a body of audio materials to match the needs and interests of a given student. This is why incremental video with one's own audio collection is an example of procedural learning that can be extremely efficient on one hand, and unadulterated fun on the other.

Learning pronunciation

Incremental video can also be used to learn pronunciation. However, active learning approach is needed in learning pronunciation. This means that audio extracts should be used as answers to pronunciation questions posed by SuperMemo. Traditionally, question&answer approach was also used to learn pronunciation with the help of IPA (International Phonetic Alphabet). For example:

  • Q: pronounce: aldehyde
  • A: 'aldɨhaɪd

However, due to the importance of intonation, native audio extracts are strongly preferred when learning exemplary grammatical structures.

Knowledge set sizes

Even though a great store is attached to the theory of grammar, the theoretical section is actually the smallest in size. At least when measured by the grammatical competence of a well-educated native speaker. Theoretical knowledge on the order of hundreds of SuperMemo items is comparable to the knowledge of grammar of an average native college graduate. This body of knowledge can be acquired within a month, however, it is recommended that it proceeds in parallel with rule instantiation and a general use of acquired rules in practise.

The body of example items used to consolidate rule instantiation should be proportionally larger than the body of theoretical knowledge. This comes from a simple fact that each rule should be instantiated more than once, and important and/or difficult rules should be instantiated a dozen times or so.

Few people realize the enormity of the body of facts related to grammatical competence in any non-artificial language. The size of such a body can be easily estimated by students who compete to achieve native competence with the help of SuperMemo. Known cases of such efforts indicate that factual knowledge related to English grammar is on the order of thousands of SuperMemo items. If one adds the size of an average intelligent man's vocabulary, the total size of the linguistic knowledge related to learning a foreign language at the native level may approach 100,000 SuperMemo items. Mastering such a collection requires a few years of intense effort, and still is not sufficient without the actual use of the language on a regular basis and in all its practical aspects. In conclusion, knowing the limitations of SuperMemo and the human brain, few people can achieve native competence in more than 2-3 languages within a lifetime. Famous polyglots avail of the fact that there is an enormous effect of decreasing returns to scale when learning a language. It is far easier to learn many languages at the average level than to learn a foreign language really well.

See also

1.3.34