@ Kayo Yoshida: I'm afraid I'm not sure what you're referring to here. Could you please specify what you would like an example of, and what you're struggling with?

@ ShamselDin: Thank you! I'm glad you found it useful!

@ Derek B: Homophony is definitely something that causes ambiguity. Here, my concern is about something that we often don't explicitly think about: the observation that what we *perceive* as an [s] or [z], or [f] or [v] is actually very context-dependent. Indeed, precisely the same acoustic waves that we perceive as an [s] in some environments can be perceived as [z] in other environments.

@ Jonathan: Thank you! I've found it useful to show spectrograms to students when highlighting contrasts like this (if only to offer scientific evidence that what we perceive isn't a constant function of what physical noise impinges on our sensory perceptors).

@Michaela: Thank you!

This is really an insightful overview of the canonical tones of Mandarin. I have a couple of (potentially hair-splitting) technical comments!

One small issue I have with the motivation part of this article has to do with the distinction between category perception (which is what the infant study you cite discusses), and category production (what the remainder of your article is focused on). It would provide a stronger motivation for your discussion on production if you could make a stronger link between category perception in children to how adults learn to produce meaningful phonological contrasts in their non-native language(s).

Also, to be a little more specific about the finding of this study, and more recent online studies of what is called the "perceptual magnet effect": the finding reported in the article you cite is that even prelinguistic infants show signs of becoming more sensitive to the sound features that underlie phonological contrasts in the language input that they have received (from their parents, for example). Similarly, even at an early age, infants show signs of steady desensitization to contrasts that they have been conditioned to ignore given the language data they have been exposed to. This distinction is important because there is no clear evidence to suggest that children know they are learning one language, or two, or 20. Children do indeed ultimately separate languages due to their ability to pick up on cues like prosody, rhythm, syllable structure, etc., that differ between the systems they are exposed to. My point here is that children bootstrap their first language(s) from input they receive, and part of this bootstrapping process involves deciding which sound components are meaningful, and which aren't.

Thanks for taking up the challenging task of explaining tone production in Mandarin -- this is a very helpful (and fun) guide!