Background and Aims: Children with developmental language disorder (DLD) benefit from word learning procedures that include a mix of immediate retrieval and spaced retrieval trials. In this study, we examine the relative contribution of these two types of retrieval. Methods: We examine data from Haebig et al. (2019) in their study that compared an immediate retrieval condition and a condition of spaced retrieval that also included immediate retrieval trials. Participants were 4- and 5-year old children with DLD and same-age peers with typical language development. Each child learned novel (made-up) words referring to unusual plants and animals in both conditions. We examined the phonetic accuracy of the novel words used during the final learning trial and during recall tests 5 min and 1 week after learning. Results: On the final learning trial, the children were more phonetically accurate in using the novel words learned in the immediate retrieval condition. However, recall tests after the learning trials revealed a decrease in accuracy, especially for the children with DLD. After one week, accuracy was much lower for words in the immediate retrieval condition than for words in the mixed spaced-plus-immediate retrieval condition. For words learned in the mixed spaced-plus-immediate retrieval condition, accuracy was very stable across time for both groups. Conclusions: Immediate retrieval boosts the phonetic accuracy of new words in the short term but spaced retrieval promotes stability and increases the likelihood that short-term gains are maintained. Implications: When novel word learning is assessed at the level of phonetic accuracy, children with DLD can show declines over time not characteristic of children with typical language development. Spaced retrieval procedures augmented by immediate retrieval opportunities during learning appear to prevent such declines, leading to longer-lasting gains.