And even if Broca's region is necessary for speech, that does not mean it is sufficient. You wouldn't assume an electronic circuit is localized to one resistor just because it breaks when that resistor is removed.
That's a good question, I'd need to ask a neurologist friend of mine.
IIRC there are some areas that are localized almost in the sense of a circuit.
Hearing is an example, which needs to process sensory information much faster (and more direct) than ordinary pathways would, in order to be able to construct spatial representation from latency differences.
I suppose "more direct" can be interpreted differently, but to the best of my knowledge the signal from your cochlea takes more axonal jumps[0] than e.g. visual, which jumps "only" through LGM[1]. Even more direct is the olfactory pathway which skips the thalamus altogether[2].
However, in the context of the article you might be referring only to intertelencephalic processing, in which case I'd love to read a paper which compares the directness of hearing vs other modalities.
[0]: https://teachmeanatomy.info/neuroanatomy/pathways/auditory-p...
[1]: https://www.ophthalmologyreview.org/articles/the-afferent-vi...
[2]: https://www.frontiersin.org/articles/10.3389/fncir.2015.0004...
If you destroy an area and some people lose one particular function, while other people only lose small fragments of the function - which seems to be true of Broca's region - can the function really be said to be localised?