The aims of this paper are to explain the relation of a glottalization and the prosodic structure and to propose a rule of glottalization on the morphological boundary. Glottalizations on the morphological boundary are two kinds in korean. One is a glottalization in the compound word. The other is a glottalization at the back of an ending of an adnominal form. When an accent phrase is consisted of a prosodic word, the glottalization is realized in korean. Compound word has two prosodic words in the lexicon. Compound word that is realized the glottalization should be merged by using a glottal stop and a resyllabification and it can be a prosodic word. It is the degree of lexicalization that an underlying cause of merger two prosodic words; Compound word that is finished to be lexicalization is realized the glottalization by using a glottal stop and a resyllabification. On the contrary, compound word that is not finished to be lexicalization is realized a voicing. A speaker will recognize the former that is a rhythm unit boundary between a prosodic word and a prosodic word. The speaker will acknowledge the latter that is not the rhythm unit boundary between the prosodic word and the prosodic word. The prosodic structure of these words are different as following. Diagram 1. The prosodic structure in the compound word. <본문그림 삽입> The endings of an adnominal form are two kinds in korean; `-(i)l` and `-(i)n`. Glottalization is realized optionally at the back of the ending of an adnominal form `-(i)l` But it is not realized at the back of the ending of an adnominal form `-(i)n.` If a noun phrase must be consisted of an accent phrase, the glottalization can be realized. But, if the constituents of a phrase are made of the accent phrases each other, the voicing can be realized. Therefore, the speaker will pronounce the noun phrase that is realized the glottalization as one prosodic phrase because it is not a rhythm unit boundary in two prosodic words. But the speaker will pronounce the noun phrase that is realized the voicing by using two prosodic phrases because it is the rhythm unit boundary in two prosodic words. The prosodic structures of a verb phrase are different as following Glottalization : _AP[kiga] _AP[t∫ul¬t`onil] _AP[kiek¬handa] Voicing : _AP[kiga] _AP[t∫ul] _AP[tonil] _AP[kiek¬handa) According to above the passage, I propose the rule of glottalization on the morphological boundary. [-tense] → [+tense] / ([...C_l]_pw_[C_2...]_pw)_AP Condition : ① C_1 must be [+sonorant], [-released]. ② C_2 must be /p`, t`, k`, s`, t∫`/. ③ Rhythm unit boundary is not between prosodic word and prosodic word.