本研究分析兩个世代攏總28位華台語雙語發音人、攏總14點鐘的自然語料。分析的重點是6个台灣話子音的誤讀情形。這6个子音是聲母［g］、［b］，佮韻尾［p］、［t］、［k］、［m］，6个攏是佮華語的音韻限制（phonotactics）有牴觸的音。結果顯示，老人組發音人雖然嘛攏真gâu講華語，毋過in的台灣話音韻系統猶原是相當穩定，因為誤讀音，差不多攏是少年發音人發的。除了單純的語音分析，本研究猶閣uì語言學的角度，來探討而且分析少年組發音人所有的誤讀音。結果發現這寡誤讀出來的音，毋是完全無規則通歸納的，顛倒有一寡脈絡會當參考：除了華語的第一語言轉移（L1 transfer）以外，本研究嘛針對應讀音佮取代音（substitute）之間的關係，歸納出4項原則。另外，台語語音除錯教學研究佮入聲韻尾的聲學特徵，嘛會當進一步解說入聲韻尾［p］的低誤讀率；啊「我」佮「阮」這兩字的［g］聲母流失率懸的原因，就可能佮詞頻（word frequency）有關。

This study analyzed the Taiwanese (also known as Taiwan Southern Min) spontaneous speech data of 28 Taiwanese/Mandarin bilinguals, balanced in two generations. The analyzed points were the mispronunciations of six Taiwanese consonants: the syllable onset [g] and [b], and the codas [p], [t], [k], [m]; all these six consonants were not found in Mandarin phonotactics.Results showed that the old speakers, despite of being able to fluently speak Mandarin, stably maintained Taiwanese phonological system. Nearly all mispronunciations were produced by the young speakers. This study further resorted to linguistic studies to analyze all the mispronunciations produced by the young speakers, intending to explore the mechanisms, if there are any, behind these mispronunciations.The results indicated that these mispronunciations did not occurr in a completely random manner. In addition to the influence from Mandarin as the L1 transfer, we proposed four principles to, to some degrees, predict the major substitute of the mispronounced consonants. In addition, the relatively low mispronunciation rate of the coda [p] may be related to its salient lip movement and the unique acoustic features. The high dropping rate of the initial [g] in the words gua and gun/guan, on the other hand, may be attributed to the high frequency of these two words.