Previous work has shown that emotional prosody, independent of the lexical carrier, activates words associated with the emotional information (Kim, 2015; Kim & Sumner, submitted). For example, hearing a non-emotional word (pineapple) uttered with angry prosody facilitates recognition of angry-associated words (mad). Building on this finding, the current study delves into the nature of the affective priming between emotional prosody and emotional words and tests if word length modulates affective priming. Word length is an important dimension in lexical processing, as longer words are shown to produce stronger lexical activation than shorter words (Pitt & Samuel, 2006). I hypothesize that social information shows a stronger effect in spoken word processing when lexical activation is weaker. Then we should find stronger affective priming with shorter words than longer words. This hypothesis was tested with a cross-modal priming experiment. The visual targets were 12 angry related words (e.g., mad, upset). The targets were preceded by two-, three-, or four-syllable non-emotional primes (e.g., atom,envelope, aluminum) spoken with angry prosody. Listeners recognized angry words faster after hearing angry prosody than after hearing neutral prosody when the prime words were short (2 syllables) but not when the prime words were longer (3-4 syllables). The current results provide evidence that social effects in word recognition are modulated by the strength of lexical activation.