A tense consonant blocks compound tensification (Ito 2014, S. Kim 2016) but facilitates word-initial tensification in Korean (Kang and Oh 2016, H. Kim 2016). Previous studies examined such a tense consonant effect as dissimilatory in compounds but as assimiliatory in simplex words separately. The aims of this paper are twofold. First, we will test if a tense consonant is affected differently depending on the morphosyntactic domains by studying both compound and word-initial tensification processes, unlike previous studies. Second, we will investigate what motivates the asymmetry of the tense consonant effect between compound and word-initial tensification processes. We conducted a survey where 40 Korean speakers selected their own pronunciations for 120 compounds and 100 simplex words. We found that a tense consonant in fact behaves differently in compound and in word-initial tensification processes. We present two possible reasons why such a difference is observed. The perception view assumes that each tensification has its own perceptual function. Compound tensification is intended to make a compound perceptually distinguishable from each of its components. Compound tensification is then avoided before another tense consonant since a tense-tense sequence of consonants could make them perceptibly similar to simplex words. The production view of the tense consonant effect emphasizes the avoidance of consecutive long vowel-to-vowel intervals. Laryngeal cooccurrence restrictions attested in tensification in Korean will be discussed in light of both production and perception. (Chosun University, Assistant Professor and Chonnam National University, Professor)