Associative learning and self-organization as basic principles for simulating speech acquisition, speech production, and speech perception

EPJ Nonlinear Biomedical Physics

Table 3 Training results for 10 most frequent syllables

Syllable	Number of model neurons (exemplars)	Mean variance of exemplars	Mean variance of training items
‘?Unt	13	0,004114	0,004037
‘di:	11	0,001899	0,004117
‘zi:	10	0,002830	0,007564
t@	10	0,002761	0,003878
‘dE6	10	0,001903	0,008438
‘?E6	8	0,002896	0,013582
n@	7	0,002053	0,005684
g@	6	0,001418	0,003845
n@n	5	0,002184	0,005175
b@	3	0,001906	0,001881

First column: phonetic transcription (SAMPA) of the 10 most frequent syllables of our children’s book corpus; second column: number of neurons representing all realizations of these syllables within the phonetic map (number of exemplars); third column: mean variance of the auditory states of all realizations by activating specific neurons within the phonetic map, representing exemplars for a syllable; fourth column: mean variance of the auditory states of all realizations activated by the training items directly.