Research Bites: Grammar Beyond the Sentence


Granger, S. (1999). Uses of tenses by advanced EFL learners: evidence from an error-tagged computer corpus. Language and Computers, 26, 191-202. [link]

Twitter Summary

Granger (’99) says that too much focus on sentence-level grammar may neg. affect grammar skills. #researchbites


Many learner tense errors may be the result of focusing too much on clauses and sentences rather than at the discourse level. This is likely due to teacher’s focusing on grammar at the sentence level.


In this study, Granger looked at two 75,000 word corpora made up of French university EFL students’ writings. The corpora had already been error tagged previously using the manual Computer-aided Error Analysis system. She analyzed the corpora for verb tense errors, looking for patterns of misuse or nonuse.


Granger found that these learners had most problems with certain pairs of tenses, often confusing the first with the second:

  • the present simple for the simple past <– this was the most prevalent kind of error
  • the present continuous for the present simple
  • the simple past for the present perfect
  • the simple present for the present perfect
  • the past perfect for the simple past

Granger claims that both sentence-level focus and L1-L2 interference are to blame for this. I will focus on the first claim: she noted that incorrect verb tenses were often correct when the sentence was taken out of context, but in context the tense was incorrect. In addition, over-reliance on time markers (e.g. for, since, ago) which automatically trigger a certain tense caused incorrect choices in the sentence’s context.

She claims this is likely “teaching-induced” and points to grammar guides stating rules without attending to important principles such as “tense continuity” (i.e. cohesion; pp. 5-6).

Practical Applications

Two practical applications stand-out: 1) don’t teach grammar at the sentence level and 2) don’t teach trigger words (e.g. for, yet, since, never) as hard and fast rules. She also states that, at the advanced level, tenses need to be taught contrastively.

Personal Reaction

Granger is a well known researcher in corpus linguistics and data-driven learning. What’s interesting here is she is suggesting that data-driven learning (in the form of KWIC concordances) may not be suitable for learning tenses because only a truncated context is shown. So, perhaps DDL is more useful for less cohesive or discourse-related linguistic items. Indeed, in my own research on adverbial connectors, DDL did not necessarily cause an increase in effective use of them.

Questions for Discussion

Regarding discourse-level grammar teaching, I have to admit I am guilty of focusing mostly on the sentence, and teaching trigger words too. So, what are some useful techniques, ideas, and activities for teaching grammar beyond the sentence level?