In this study, we used an unmasked constituent priming paradigm to test the effect of whole-word frequency and semantic transparency in the processing of Japanese compound verbs. Native speakers of Japanese saw a prime (related prime such as first constituent of the compound verb or an unrelated prime) overtly and then made lexical decision to a target compound verb. Results showed that target words with high whole-word frequency were significantly recognized faster than target words with low whole-word frequency, irrespective of prime type and semantic transparency. These results contrasted with English, in which robust constituent priming effects have been consistently reported. We contend that the processing of compound words is modulated by language-specific properties such as morphological properties, syntactic properties and orthographic properties.