Research Bites: Learning Connotation through Data-Driven Learning (DDL)


Mansoory, N., & Jafarpour, M. (2014). Teaching semantic prosody of English verbs through the DDL approach and its effect on learners’ vocabulary choice appropriateness in a Persian EFL context. Advances in Language and Literary Studies, 5(2), 149-161. [link]

Twitter Summary

Mansoory et al found that corpus tools and DDL can lead to better understanding of connotation/semantic prosody for ELLs.


Semantic prosody is the connotation of words derived from their common collocations. For example, the verb cause has a negative semantic prosody because cause often collocates with negative words (e.g. problems, offence, distress). Vocabulary glosses and monolingual and bilingual dictionaries often only contain denotational information and are therefore not useful for learning connotation or semantic prosody. Through corpora, on the other hand, discovering semantic prosody is possible. Students can benefit from this in a classroom that utilizes data-driven learning (DDL).


41 Persian high school EFL students took part in this study. There was a control group (discovering semantic prosody through dictionaries) and a DDL experimental group (discovering semantic prosody through concordances). For the DDL group, the teacher/researcher took a scaffolded approach:

  • First, students were trained in the use of concordances.
  • Then they examined verbs found in their coursework, along with near-synonyms given by the teacher, initially using only a few examples.
    • During this stage, the teacher checked students’ analyses, had them compare answers, and finally provided them with the correct analysis – giving feedback along the way.
  • Eventually, they were required to find not only the semantic prosody of verbs given, but also their near synonyms.
    • The constant practice and use of concordances over time was though to lead to more autonomous corpus investigation.

Before and after 34 sessions (each sessions devoting 30 minutes to semantic prosody work), a gap-fill test was completed which asked students to select the most appropriate word from a synonym set. The sentence was taken from the corpus and the correct answer was considered the original term.

Example Test Item
Sentence: The constantly threatening nuclear war will ———.
Near-synonym set: start, develop, break out
Original term: break out


The researchers found that the DDL group’s posttest score improved significantly while there was little improvement in the control group. Therefore, a DDL approach to semantic prosody may be valid, in particular for verbs and their collocations.


This article serves as a reminder that connotation and semantic prosody are important things to teach and that any exploration of vocabulary may be incomplete without it. In addition, a structured/scaffolded approach to semantic prosody investigation through DDL may be a useful technique in the classroom. In this article, the researchers spent 30 minutes per class on semantic prosody. This type of time commitment is not convenient for most teachers. However, this type of activity can easily be done more quickly and less often. In other words, whether you take 2 minutes or 10 minutes and do it only a few times, showing students that it is possible to find the semantic prosody of words through corpora is a worthwhile endeavor.

That being said, it may not be possible to determine all cases of semantic prosody for every word. The data may not be there, or it may be lost in the ‘noise’ of concordances (the truncated sentences, disconnected language, stray XML formatting), especially if they are lower-level learners. Still, with a little proactive planning, this could be a useful technique.

Most students don’t know about corpora tools, so using the tools will likely spark interest and be a first step towards autonomous language investigations. Here are some suggestions I thought of:

  • Before class, find an interesting word, whether in a coursebook or one you can predict appearing in your class during conversation, and do a few KWIC searches on COCA or WordAndPhrase. Look for some good examples and put them in a slide/worksheet/notebook for use in class.
  • When these words come up, ask students about the connotation. Then, tell them you know a really cool tool to find out more.
  • Show students how and why you are using the corpus.
  • Prepare more examples or get students to elicit more examples that you can investigate together. You can also do this in the next lesson as a kind of reminder of why a corpus is useful. Think of this as slowly indoctrinating students into the wonderful world of DDL.
  • Set a corpus-based homework assignment
  • Use these techniques throughout your course. Plan or react to interesting vocabulary by using corpora often.
  • Remind students to look up interesting vocabulary and add this information to their lexical notebooks, vocabulary journals, or flashcards.

5 thoughts on “Research Bites: Learning Connotation through Data-Driven Learning (DDL)

    • Anthony Teacher says:

      Hi. Thanks for the comment. You are one of the co-authors, correct?

      Did I miss anything important?

      It would also be wonderful if you could provide an example of some of the materials you used, if possible.

      • mohsen says:

        Hi. Sorry for the delay. You are welcome. Yes, I am, and regarding your review everything was OK. what kind of materials? I did not get what you meant.

        • Anthony Teacher says:

          Sorry about being vague. In the article on semantic prosody, you provided the assessment tool (the test) in the appendix, but I was wondering if you also made any materials (worksheets, handouts) to assist with the DDL investigations you did with your students.

  1. Hi Anthony,
    I really like this approach to reading articles – it’s useful, as I tend to read blogs and can’t find time to access journals/articles at the moment. Can I just check – DDL is data-driven learning, right? It might be useful to put it at the start of the post, just in case people aren’t aware of the abbreviation.

Leave a Reply