This paper proposes that SOUND SYMBOLISM, and more specifically PHONEMIC ICONICITY, plays a role in conveying emotional weight in the context of poetry. Previous research has indicated that the ratio of plosives to nasals in poetry predicts overall perception of emotional affect, with plosives designating activity and pleasantness, and nasals designating inactivity and unpleasantness (Auracher, Albers, Zhai, Gareeva, & Stavniychuk 2010); however, this research has ignored the influence of such potentially mitigating factors as orthography and lexical meaning. The current study involves naive English L1 speakers listening to recordings of selected haiku from Matsuo Basho's Oku No Hosomichi (`Narrow Road to the Deep North') in the original Japanese, and as such, the potential of orthography and lexical meaning to influence perception of emotion is eliminated. After listening to each haiku twice, subjects were asked to rate the appropriateness of eight emotion words that ranged from active and positive to inactive and positive, and from active and negative to inactive and negative, on a five-point Likert scale. Emotion words were chosen on the basis of their respective positions on the Circumplex Model of Affect, in which each emotion is conceptualized in terms of its location along two intersecting axes measuring valence (negative - positive) and arousal (inactive - active) (Russell 1980). The selected words occupied regularly spaced positions along this two-dimensional circular model. Results indicate that plosive to nasal ratio may indeed play a role in the perception of emotion in poetry, particularly in the case of poems with high plosive to nasal ratios, which were perceived as markedly more active and positive than other poems. Wider implications of the discernible patterns of perception of emotional affect based on plosive to nasal ratio include the possibility that phonemic iconicity plays a role in general language processing. As this research involves Japanese L2 phonemic perception by naÃ&hibar;ve English L1 listeners, current L2 phonological perceptual theory is discussed, and taken into account in the analysis of the results. Specific consideration is given to the potential of English L1 speakers to perceive the Japanese rhotic /r/, which does not appear in English, as the plosives /t/ or /d/, and the Japanese affricate /ts/, which commonly appears syllable-initially in Japanese, but is much rarer in this position in English, as /s/ (Nozawa 2008). Future research on English L1 speakers' underlying perceptual categorizations of targeted sounds in Japanese is also proposed.