本篇論文是針對六位以台灣華語為母語的嬰幼兒，採長期觀察的方式，研究 台灣華語字彙產製的音韻變化性(phonological variability)，並詳細描述單音節詞和雙音節詞之中音節類型出現的頻率、變化性、以及代換模式。本研究同時要用Jakoson (1968)的音節標記理論來檢驗各種音節類型中的共通性。 本研究一共觀察了有六位年齡在十一個月至兩歲的嬰幼兒長達一年。以兩個 禮拜一次的頻率收集嬰幼兒和母親之間的自然對話，並利用錄製回來的影音檔做 譯寫和分析。 結果顯示小朋友的音韻變化是很常見的，且是有規則可循的。小朋友的音韻變異量的高峰(variability peak)會出現在當小朋友的音韻發展從一個階段進展到另一個階段的時候，而本篇論文顯示與當小朋友由單字期(one-word stage)進展到雙字期(two-word stage)以及字彙量有大幅上升的時期符合。華語音節習得的部分，結果顯示CV 是頻率最高、變化性最低，且最常被拿來替換的音節類型。CVG 也是頻率高的音節類型之一，但他的變化性也很高，主要是因為韻尾省略 (coda-dropping)的現象在小朋友的早期發展很常見的關係，所以CVG 雖然頻率高但是變化性也很高而且是最常被取代的音節類型之一。 最後，將所有的結果拿來檢驗Jakoson (1968)的音節標記理論，結果發現頻 率高以及變化性低的音節類型都是無標記(unmarked)的音節類型，相反的頻率低 以及變化性高的音節類型則都是有標記的(marked)音節類型，此外小朋友會用無 標記的音節類型來取代有標記的音節類型。 / The purpose of this study is to discuss the issue concerning phonological variability of children acquiring Taiwan Mandarin. Two aspects are including in the following: the phonological variability of words and the syllable types composed the words. The overall variability pattern, the frequency, variability rate, and substitution pattern of syllable type were analyzed. Six participants are investigated in the study, aged between 0;11 to 2;0. A longitudinal observation study is conducted by the author and the research team. The results showed that phonological variability is common in early phonological development. The increase in variability reflects the reorganization of phonological system, where children started to produce two-word utterances and the amount of different words was increased. As for the syllable type analysis, CV presented the highest in frequency, the lowest in variability rate, and also was used to replace other syllable type more often. CVG was one of the most frequently used syllable type; however, the variability rate of CVG was also high. The reason may due to the fact coda-dropping is a very common phenomenon in children’s development. The results in this study were examined in the markedness theory of syllable proposed by Jakobson (1968). The results showed that syllable types with higher frequency and lower variability rates were unmarked syllable types, while syllable types with lower frequency and higher variability rates were marked syllable types. Furthermore, children tended to use a more unmarked syllable to replace a more marked syllable.