As a follow up to a recent question (1) where it is argued that the effect of substrate on the second language is most remarkable in phonology:
Do you know if this has biological reasons in production (bones, muscles, innervation, etc.) and perception (cognition), or what?
I have heard that children's ability to learn phonemes drops off after the first six months, but the causes and implications were left unexplained (Lingthusiasm Episode 12 [2], refering us for the age claim to Non-native phonemes in adult word learning: evidence from the N400m [2]).
Update, now that I reviewed the source:
Gretchen McMuloch (Lingthusiasm) explains:
So, there are pairs – so in English, for example, English babies retain the ability to hear the difference between “ba ba ba,” “pa pa pa,” but they don’t retain the ability to hear the difference between “ta ta ta” and “ʈa ʈa ʈa” and –
[...] the second ones were produced with my tongue curled back so that the bottom of my tongue was touching the roof of my mouth.
[...] I can produce them because I know where to put my tongue, but I can’t actually hear the difference.
I have listened to enough English speaking podcasters mentioning comparanda from German to know that Gretchen has with a relatively high degree of probabability not produced those sounds in a manner that would be unmarked to any Nepali speaker. This agrees with the premise of the related question (1), which is after all not in doubt.
It is arguable that the perception is relied on in practice. The language learner practicing a given realization has no way to reinforce their progress in training if they cannot perceive the sound. This does not answer the question to satisfaction, because it denies the purpose of practice. Even if they reach an acceptable stage where most of all minimal pairs could be distinguished, the vast majority of late second-language learners have a ''thick accent'', effectively creating allophones. Some idiolects can be close to incomprehensible, despite a certain degree of fluency.
Infants babble, progressing through a series of phonemes which are typologically common. I know I do when reading the description of a phoneme.
I expect that baby talk develops a representation of the sounds which may be rather abstract, and that the production conditions the cognition as well. It is no coincident that /m/ is among the simpler and most common phonemes (as has been argued for mama).