The paper investigates the way some Chinese weak forms in spontaneous speech are perceived when taken out of a wider context. These forms are the result of syllable initials and finals modifications as well as coalescence due to the drop of syllable constituents. The results of the previous acoustic study demonstrated that not only functional but notional words have weak forms. We hypothesized that a number of those forms appear to acoustically resemble other Chinese words or word groups giving way to potential homophony. To prove the hypothesis, a perceptual study was performed. The tokens presenting potential homophones were segmented from native female spontaneous speech of 6 subjects. Two groups of Chinese native speakers (20 males, 20 females) participated in the experiment performing a discrimination task. The obtained results prove the hypothesis as in both groups, only 16% of responses identified canonical syllables, whereas 84% of responses identified the homonyms. There was certain gender difference which, however, proved insignificant for three fourths of the responses – more male listeners preferred homophones to canonical words/word groups. As expected, the reliable preference for the homophone during the discrimination task was demonstrated in 64% of the responses classified as non-identification of the canonical word/word group. Chinese native speakers are aware of this homophony, which in some instances is further expressed in the change of hieroglyphs creating not just homophones but full homonyms that have to be considered in Chinese as L2 teaching and learning.