Multiple hyphens make this phrase feel unwieldy... but are they right?

I'm writing a paper in which I refer to "natural-language-controlled robots" about thirty times. I'm curious about this phrase's hyphenation.

I would write

robots controlled by natural language

without hyphens, but when I turn the phrase around to avoid ambiguity, it turns into

natural language controlled robots

which seems to need hyphens all-upons! The robots are not controlled robots of the language variety, so I need a hyphen between language and controlled. They are similarly not language-controlled robots of the natural variety, so it seems I need hyphens between all three words.

However,

natural-language-controlled robots

feels unwieldy, which brings me here. Is this really how this phrase should be punctuated?

The alternative I've come up with is applying a partial initialism, yielding

NL-controlled wheelchairs

which seems better, but prompts similar problems; do I really want a hyphen attached to an initialism? That looks wrong. But the sentence seems semantically wrong without it!

I'm sure I'm overthinking this.

Solution 1:

There is a degree of flexibility in this, though as with other cases where there's a degree of flexibility, a given style guide may have a firm rule that you should follow rather than what anyone says here.

When creating a compound where one or more of your elements are itself compounds (in this case natural language or natural-language depending on whether you decide to hyphenate that as discussed further below), then many favour using an en-dash (–) over a hyphen, especially if the compound is open (with a space rather than a hyphen). I would strongly advise on following this style as a rule unless you've a guide that insists against it, as it adds a heirarchy of "levels" to your compounds. So the choice is between natural-language–controlled and natural language–controlled.

Now, I've already given my opinion here on the general question of when to hyphenate and I won't repeat it all there especially since some (but not all) of it is a matter of personal style.

I would say that in this case if I'd expect people that would be reading the piece to be familiar with the phrase natural language, then I'd favour the open form:

natural language–controlled

Because I would expect such readers to "pick up on" the phrase natural language as a unit.

If I would expect a large number of readers to not be familiar with the phrase, then I'd favour the closed form:

natural-language–controlled

Because such readers would be more likely to mis-read "language–controlled" as a unit that is modified by natural, even with the en-dash (after all the difference between - and –, is a subtle one).

Personally, I'd prefer to read the first of my two suggestions, but attention must be paid to likely audiences, so I'd go against my own preference if it was for a very wide audience including many unfamiliar with the term natural language.