Does morphology support novel word learning through statistical learning?

Most novel words that speakers learn are morphologically complex (e.g. columnist, whistleblower). Rather than hindering learning, such complexity might provide facilitation. Affixes (e.g. pre-, -ness) could do so in at least two ways: either thanks to their meaning or because they are frequent clusters of letters. This information is statistical in nature and might thus be captured via statistical learning. To investigate this, Italian native speakers learned novel words with (i) existing suffixes (rugob-enza, akin to troft-er in English), (ii) non-meaningful endings matched in frequency (rugob-ondo, troft-an), and (iii) non-meaningful, low-frequency endings (rugob-allo, troft-ov). Participants also completed a visual statistical learning task. Results showed that items with suffixes and low-frequency endings were learned best, while also exhibiting the strongest correlation with statistical learning ability. These results highlight the role of statistical information in word learning, and particularly its connection with the effect of derivational morphology in the process.