Context can reduce accurate word learning

I’ve been reading an interesting 2017 dissertation by US researcher Reem Al Ghanem in preparation for this month’s DSF conference. It’s about how children learn to read and write polysyllabic words.

One section jumped out at me, because multi-cueing and the idea that phonics/word study should occur in context is still popular in many Australian schools:

“When poor readers rely on context to aid word recognition, they focus on selecting semantically appropriate words given the context clues rather than decoding the words through letter-sound conversion strategy.

When children utilize a compensatory strategy like contextual guessing rather than phonological decoding to aid their word recognition, their attention to word form is limited, resulting in poorer acquisition of word-specific representations, hence the negative context effects.

When poor readers are presented with words in isolation, they are forced to read them using phonological decoding. Although inefficient, their phonological decoding of the words increases their attention to the orthographic details of the words, resulting in acquiring higher quality representations for the words than when they are presented in
context.
” (p103)

Developing high-quality word representations is a challenging activity for struggling readers. Expecting them to only learn words in context is a bit like asking them to only learn to shoot netball or basketball goals during a real game, and discouraging goal-shooting and other skills practice.

As a weedy, unco, asthmatic kid keen to avoid on-court humiliation, I voluntarily did many hours of goal-shooting practice. Imagine if coaches discouraged such practice, and said sporting skills should only be learnt in the context of real games. We’d all stare at them. Then ignore them.

Al Ghanem’s dissertation goes on:

“While context clues can support comprehension, they are unreliable sources for orthographic learning. Teachers must select the instructional strategy that fits the goal of instruction, and presenting words in isolation appears to be the most beneficial when the goal of instruction is acquiring word-specific representations.” (p107)

15 thoughts on “Context can reduce accurate word learning

  1. Voon Pang

    Thanks so much for this Alison! I am working with quite a few students who are struggle with poly-syllabic words and even single syllable words due to the multi-cueing instruction they’ve received and single word testing has revealed faulty representations and guessing (e.g. “olive” read as “Olivia”, “mechanic” read as “machine”). This hits the nail on the head!

    Reply
  2. emily

    Thanks for sharing
    This is absolutely in effect in my child’s reading practice, he is a struggling reader. We focus on explicit synthetic phonics instruction at home. Occasionally a book comes home from school that he is supposed to read. Every time he spends so much energy trying to work out the context clues given in the whole language type materials. It’s like he’s trying to do so many competing things that none of it is done well and just ends in confusion and disillusion.

    Reply
  3. Kate

    Alison, the import and comment about context resonated with me when you wrote, “As a weedy, unco, asthmatic kid keen to avoid on-court humiliation, I voluntarily did many hours of goal-shooting practice. Imagine if coaches discouraged such practice, and said sporting skills should only be learnt in the context of real games. We’d all stare at them. Then ignore them.” Great analogy and goal in one 🙂

    Reply
  4. aJ

    Working in a school, and given 20 minutes once a week to teach goal shooting to kids who can’t catch a ball. As an OT and SP you can imagine my consternation.

    Reply
  5. Elsa Hodgins

    I agree with but…… always a but….. I have seen a client struggling with reading had and Liz in a decodable reader for weeks until we took a step back and discussed the name of the characters in the story and looked at had and its meaning which resulted in a better outcome the next time so.oooooo. I suppose one has to make a clinical judgement about when to try another strategy to achieve the golden aha moment for that reader and context and comprehension can help.

    Reply
    1. alison Post author

      Yes, we’re trying to unitise phonemes, graphemes, and morphemes/semantics, so sometimes it’s useful to think about why something isn’t sticking and then boost that layer, and that could be done by going back into connected text.

      Reply
  6. JOHN A MCCLUSKEY

    I used to tutor the MacqLit program. Students were able to read words in contextual sentences than in isolation. For example, the sight word: ‘said’. “In a sentence it rolls off the tongue so naturally it barely needs to be decoded”… he said. But on its own, ‘said’ doesn’t look right. Just look at it! the ai should make an ‘ay’ sound. Students can see the word ‘aid’ in there and it could easily be adjusted to ‘paid’, all strategies that will not help the learner here. Other high frequency words my students struggled with in isolation were ‘Mr’, ‘because’ and ‘where’.
    John

    Reply
    1. alison Post author

      Yes, but you teach the patterns and then explain/discuss unusual spellings, I would never leave a student to sink or swim with an unusually-spelt word on a wordlist. ‘Said’ was once pronounced ‘sayed’ as in ‘she said unto him’ and sometimes was even written ‘sayed’ but over time the pronunciation shifted, a bit like ‘again’ and ‘against’ now both have two possible pronunciations, more posh people say the ‘ai’ as /ae/ and the rest of us say /e/. ‘Mr’ is an abbreviation and can be taught with others like Mrs and Ms and Dr, ‘because’ was originally by (bi) + cause, and then the pronunciation shifted. I teach ‘here’, ‘there’ and ‘where’ as all being about places and the latter 2 contain the former word.

      Reply
  7. G

    Thank you Alison for your informative articles. Removing the context makes sense.
    When I sent my daughter to school with a decodable reader, her teacher had her read both books, the decodable and the regular levelled reader. She later mentioned to me that the decodable reader was harder for her. I got the feeling that she was implying that therefore the decodable readers are not as good. I had to have a little think on that before saying that yes, I think that is the idea. They might look more simple but she has to work hard to decode (read!) each word; there’s nothing there to help her guess it, and she’s great at guessing. I could see her thinking about that.

    Reply
  8. Jennifer

    Thanks so much for sharing this. It all helps when working with teachers who just don’t understand the benefits of explicit systematic synthetic phonics teaching.

    Reply
  9. John Walker

    Hi Alison,
    Thanks for drawing attention to this.
    What really stands out for me is the following, “When poor readers are presented with words in isolation, they are forced to read them using phonological decoding. Although inefficient, their phonological decoding of the words increases their attention to the orthographic details of the words, resulting in acquiring higher quality representations for the words than when they are presented in
    context.” And, not only does it increase ‘their attention to orthographic details of the words’, their skills become more and more automatic until they disappear under the level of conscious attention while their reading becomes more efficient. This is the kind of thing that K Anders Ericsson argued in many of his pieces on deliberate practice of the sort you did when you were a kid.
    I’m away from home at the moment so I can’t quote the paper (in Progress in Understanding Reading), but Keith Stanovich has said much the same in the past.
    Anyway, Al Ghanem adds yet another valuable bit of evidence for direct instruction.
    Best wishes to you.
    John

    Reply
  10. Tracey

    I’m a tier 3 interventionist and work with students that have been able to hide their compensatory habits of guessing unfamiliar words using context cues for many years. These students progress slowly but do not reach the optimum levels required for future skilled reading. These students are usually girls that are undiagnosed and have very competent language comprehension skills and therefore are expert at guessing. The way I have collected data on students to demonstrate to the teacher that the learner is guessing ( I have to work very hard in convincing some educators that students aren’t actually reading words as they have dyslexia) is by asking the student to read the same word in context and out of context. Today I chose a text with the word ‘old’ in it. Sure enough, the student read the word with confidence when asked to read it in context but failed to do so when it was presented to her in a list of words. Trying to encourage a student to decode unfamiliar words instead of guessing them for most of their school life is like trying to turn a super tanker around on the main highway.

    Reply
    1. alison Post author

      How frustrating! I agree, it’s crazy how far some well-behaved kids with good oral language can get through school without anyone noticing they have poor word-level reading. I think tests like the YARC feed into this, there are so many kids who get good reading comprehension scores on it, even though they are mostly figuring stuff out from context. There’s no way to capture their poor decoding on the test properly. That’s why we need a Year 1 Phonics check! I like to get kids to read lists of similar-looking words (split, splint, sprint, spit, spilt etc, or concrete, compete, complete) and show the adults who aren’t worried about their reading how inaccurate they are. Good luck persuading people! Alison

      Reply
  11. Nadine

    *sigh* this makes so much sense.
    Unfortunately I work in a school where the exec breathe down our neck in literacy time, and INSIST we teach and prompt for strategic activity during guided reading (using meaning and context) and not sounding out. My kids are just guessing over and over 🙁

    Reply

Leave a Reply

Your email address will not be published.