There were several very fascinating possible take-aways from this study. Firstly, vocabulary instruction in the primary grades was especially high yield. Indeed the effect sizes found for vocabulary here are higher than the effect sizes generally found for phonics, in the primary grades. While we know from other research that Whole Language approaches do not work, it does appear that we still need to teach some words, individually and directly. Secondly, while explicit instruction was unsurprisingly superior to implicit instruction (defined by the author as natural reading experiences), a combined approach (defined by the author as pairing actual reading practice with explicit vocabulary instruction) showed higher results than a purely implicit or explicit approach. I must say this confirms a personal bias of mine, as I have always found reading instruction worked best, when students received both reading practice time and explicit instruction. 

Also interesting, was the particularly high results found for teaching vocabulary to at-risk readers, which is significantly higher than the impacts of teaching phonics to at-risk readers. I think this result suggests that it is crucial that we teach both phonics and vocabulary to struggling readers. On this topic the authors wrote, ““There is no causal link between vocabulary and comprehension, (b) vocabulary instruction does not transfer beyond the taught target words and texts in which it is learned [......] Poor readers are likely to have difficulties with lower level skills such as decoding and quick access to word meanings (Mezynski, 1983). If students learn target words contained in the text, it may free up cognitive resources that can be allocated for higher level processes of integrating text (Mezynski, 1983; Perfetti, 1985). In comparison to students who do not have lower level deficits, poor readers will likely benefit more from learning vocabulary, because they can access words more quickly, thus alleviating cognitive resources and increasing their capacity to engage in the higher level skills required for comprehension.” (Marulis, 2010). 

However, absolutely crucial to this conversation was the impact of time on these results. As the authors found that vocabulary instruction showed very quickly diminishing returns. Indeed, the authors found the highest results for studies that included 5 or less instructional sessions and the lowest results for more than 18 sessions. While all results were high, there was clearly a direct correlative relationship, in which more time spent on vocabulary instruction meant lower results. With this in mind, I think it might be logical for teachers to teach the high frequency, irregularly spelled words and to move on. 

I think the 2009 meta-analysis by Dr. Amy Elleman, Et, al. Is very interesting in comparison to the Marulis study. This meta-analysis looked at 37 kindergarten to grade 12 studies, and specifically focused on comprehension outcomes. Their study found a mean effect size of .10 on reading comprehension results for teaching vocabulary. This was the only vocabulary meta-analysis that focused on comprehension. Moreover, it looked at higher grades than the other studies. For this reason, I think it is fair to say that the research evidence for teaching extended vocabulary, vocabulary in the later grades, or for comprehension purposes, appears weaker. Theoretically I think it makes more sense to teach vocabulary comprehension through morphology, however, the research evidence is also weak in this regard. My secondary meta-analysis of morphology instruction showed a mean ES of .18 for comprehension outcomes. That being said, the Elleman meta-analysis did show greater effect sizes for at-risk readers, with a mean impact of 1.23, further reinforcing the fact that vocabulary instruction appears to have a strong impact for struggling readers. Similarly to the Ellman and Marulis meta-analyses, the 1990 Marmolejo, the 1987 Foster and the 1986 Arnold meta-analyses also showed higher effect sizes for teaching vocabulary to struggling and disabled readers.