本研究以心理實驗的方式，檢測臺閩語使用者的次音節結構單位是否如傳統語言 學者所認為的「聲韻」結構，還是「體尾」結構。實驗以觀念形成的方式進行，受試者分為 新兵組以及大學生組兩組，又各隨機分為八組，每組測試一種觀念，分別為：韻、體、聲、 尾、腹、調、腹及調、邊。測試分三個部份。第一部份是測試受試者辨音的能力，第二部份 是以觀念形成測驗的方式讓受試者偵測 /ki55 / 在變調位置（即 / ki33 ）的觀念， 如「 基礎」，使受試者能習慣觀念形成測驗的方式。 第三部份為實驗主體，共有 100 題，其中 50 題為正刺激項，50 題為負刺激項。 受試者利用這 100 題的機會來形成所預定的觀念。 我們的假設是在上述八個觀念中，原來就存在語言使用者腦裡的觀念應該最容易形成，反之 則否。 測試的結果顯示：不管是新兵組或大學生組，雖然八種不同的觀念的成績有差別，但這些差 別都不顯著。也就是說這兩組受試者都沒有顯示出明顯的次音節單位。但我們發現在這八個 觀念中， 「腹及調」也就是 [a51] 的這個觀念比單純的「腹」 [a] 或單純的「調」 [51] 都要好。從傳統語言學的「分析」的角度不容易解釋這樣的現象。本文的結論從「關連論」 的觀點，建議語言使用者是從音節與音節之間互相的關連建立起語音的相似性。因此 [a51] 比起抽象度更高的 [a ] 或 [ 51 ] 更容易掌握。也就是說語言使用者是掌握音節整體，而 不是掌握音節個別的部份。

　　　　　The study uses the concept formation technique to investigate the subsyllabic units of Taiwan Min to see whether the 'onset-rhyme' structure assumed by traditonal Chinese linguists or the 'body-coda' structure is realized by Taiwan Min users. Army recruits and college students were used as subject. Each subject was randomly assigned to one of eight test conditions: rhyme, body, onset, coda, uncleus, tone, nucleus and tone, and margins. The test was divided into three parts. In the first part, the subjects were tested whether they differentiate sounds in different places of the syllable. In the second part, the subjects were to form the concept [ki55] in sandhi position (which is actually [ki33]) to get familiarized with the concept formation experiment. The third part was the main test which contained 100 items. Among these stimulus items half of them (50 items) were positive tokens and the other half negative tokens. The subjects were to form the required concept within these 100 items. The assumption was that if the speaker had one of the concept in mind, the concept should be much more easily realized than others. The results showed that the eight concepts were not significantly different. That means the subjects did not show realization of sub-syllabic units in the test. But among these eight test conditions, we found that the concept 'nucleus and tone'(i.e.[51]) fared better than either 'nucleus' alone (ie、[a]) or 'tone' alone (ie、[a]). The phenomenon is rather difficult to understand from an analytic point of view. We suggest from a 'connectionist' point of view that the speakers first phonetic establish relationship among syllables befor they realize the component parts of the syllables. The concept 'nucleus and tone' is thus more easily realized than the more abstract unit 'nucleus' or 'tone'. That is to say, the subjects realized the syllables as a whole before they realized the component parts.