It’s a fair question (raised by my also-interested-in-linguistics-wife). Here’s my short answer: a communicative approach will produce better readers, with better reading ability, faster.
To understand why I hold that position, we need two puzzle pieces: how Grammar/Translation thinks it creates readers, and how CI can be geared towards a text-oriented goal.
Grammar Translation tends to operate along these lines:
Front-load the explicit teaching of grammar so the learner knows all about how the language operates and how to analyse utterances. Teach them a lot of vocabulary by having them memorise L1 glosses. Have them translate sentences into their L1 to solidify grammar + vocab. Eventually let them loose on passages once they’ve accumulated enough of grammar + vocab.
This is why most G/T approaches don’t see students tackle extended connected text until late in a 1st year (if we’re talking about a tertiary education setting) course, and they don’t really get a huge amount of ‘reading’ (i.e. translation) until they hit second year. By this stage a “parse/gloss/translate” mindset is pretty set-in-stone and you can get through a whole 4-year university curriculum doing that and still not feel, or read, fluent(ly) – I certainly did, and I’m not alone.
Most graduates of a G/T approach will never make the transition to reading, with high accuracy and speed and without mental translation, their L2 texts.
Gearing CI to a text/reading goal:
It’s not at all the case that a communication-based approach needs to be all “may I go the bathroom?” and “A double-shot piccolo latte with a marshmallow on the side, please.” Indeed, learning such things is neither here nor there, a question that’s independent of CI.
While the very initial stages of CI will probably be physical, concrete, classroom-based, oral work, it doesn’t take that long until you can develop some structures and vocabulary to read simple texts. And once you do, you can introduce simple, but accurate, language to talk about texts. Whether that’s “subject/protagonist, theme, symbol, context” etc., or even grammatical, “(grammatical) subject, predicate, complement, verb, adjective, case”. If your end goal depends upon discussing the grammar of texts, there’s no reason you can’t do that in the L2. If your end goal is more ‘literary’, you can do that in L2, and neither of these necessarily depend upon “advanced”, or more accurately, technical, language. 4th grade kids discuss L1 texts, using 4th grade vocabulary. Post-beginner classical students can do the same in a classical language, if you give them the tools to do so.
The difference will be this, though : a CI approach that makes texts the topic of discussions, and encourages reading, especially extensive reading, is going to expose students to a ton of language, spoken and written, more than a G/T approach. Yes, it may take longer until they encounter/are able to read certain structures, because you haven’t front-loaded all the grammar. However, I think hour for hour the outcomes will be better, provided we are assessing the right thing.
I would love to hear from you if you have either personal anecdata on this, or links to peer-reviewed research.