Sentences often require the manipulation of not just words but also Arabic digits. The present study aimed at discovering whether the subsequent reading aloud of a word stimulus followed by a number stimulus (and vice versa) had a switch cost, i.e., salient delays, in comparison to reading sequences of either only words or only numbers. For this, n = 18 healthy, German learners of Japanese were asked to read aloud strings of Hiragana words and Arabic multidigit numbers. No statistically significant differences in mean response times were found when comparing sequences of words, sequences of numbers, sequences of numbers followed by a word, and sequences of words followed by a number. Instead, mean response times increased considerably for lower language proficiency levels of the participants. Mean response times further saw a statistically significant increase the higher the syllable lengths of the words and numerals. Implications are discussed in juxtaposition with existing theories on the cognitive processes necessary to read aloud words on the one hand and numbers on the other. For word-reading, these theories consist of the dual-route cascaded model (Coltheart 1978) and the parallel-distributed-processing model (Seidenberg & McClelland 1989). For number-reading, the triple-code model (Dehaene 1992) and the model for multi-digit reading (Dotan & Friedmann 2018) are explored.