Semantic clusters hinder adult beginners

Further to my post a couple of days ago re semantic clusters hindering learning, I found an interesting bit of research from the ELTJ March 2009.

The research concludes that....

adult beginners performed significantly better on the unrelated vocabulary test than on the related vocabulary test

children (intermediate level) showed no significant difference in test scores between related and unrelated vocabulary

This suggests tentatively that the presentation of unrelated vocabulary may assist learning of new L2 words more than related vocabulary only at beginners’ level (adults).

I am getting more and more interested in this subject, as it seems to buck the trend of vocab teaching and I have just started a new crop of courses of real / false starter adult beginners.
(Apology if this is old news to everyone and I've only just stumbled upon it!)

Has anyone out there (more knowledgeable and experienced than me) already changed the nature of their lessons in terms of giving vocab in non-related clusters?

If so, have you noticed any discernible improvement in Ss vocab recall after such a change??

I'd be very interested to get some feedback from anyone who has successfully taught vocab in this way and what the student reactions were?? Did the students think that there was no method and that they were suddenly presented with a load of random words???

Don't teach vocabs in associated groups!?!?!?!?

OK, so I've only just started studying the TESOL Diploma and maybe I should already know this, but I have just watched a lecture, where the speaker said that there is extensive research that shows that teaching associated vocab together is absolutely counter-productive to learning.

Here's the vid...

The lecture starts a bit slowly, but gets much better towards the middle and onwards.

When you go to any EFL lesson plans website, almost all of the beginners worksheets put vocab in related groups eg. fruit, family members, clothes etc

The speaker, Paul Nation, said that this info/research is from before 2000. SO maybe this is old news for more experienced teachers than me.
But this must surely be big news for a lot of teachers out there, who I think would teach vocab groups as the default method.

It seems incredible that this method of teaching can be 100% counter-productive to vocab memory recall (yep, that's what it says in the lecture) and I've emailed Paul Nation to get a link form him to be able to have a look at these research findings (which have been replicated a number of times with identical results apparently).

While I am eager to find out about this, I would think that it seems so ingrained in students' minds that you learn related vocab together that they might think that a teacher would not know what the hell they were doing if they only taught random words that had no interference with each other.

What would your students think?