Previous studies have demonstrated that young children use an argument structure in the input to learn words such as verbs and nouns, but they are unable to make use of this information if the language allows frequent the deletion of lexical items. These findings lead to the question of how Japanese case markers are learned because case markers are function words and are often omitted in speech. So far, few study has examined the acquisition of this unique category of words. In this study, we examined whether Japanese children could utilize the linguistic information (i.e., number of the argument and word order) when they learn artificial case markers (agent: “po”, patient: “bi”). Thirty-three five-year-old children were presented with either two arguments sentences (SOV: “rabbit po bear bi pushed”) or one argument sentences (SV: “rabbit po pushed”, OV: “bear bi pushed”) in the learning phase. After learning phases, they were given a comprehension test in the test phase in which the case markers were embedded in four different types of sentences, SOV, OSV, SV, and OV. The results showed that the children who studied one-argument sentences were more accurate in comprehending “bi” in OV sentences but not accurate in comprehending “po” in SV sentences, or “po” and “bi” in SOV and OSV sentences. In contrast, the children who studied two-argument sentences were more accurate in comprehending “po” in SOV and SV sentences, but not accurate in comprehending “bi” in OV or OSV sentences.