Diminutives in Child-Directed Speech Supplement Metric with Distributional Word Segmentation Cues

Article excerpt

In two experiments, we explored whether diminutives (e.g., birdie, Patty, bootie), which are characteristic of child-directed speech in many languages, aid word segmentation by regularizing stress patterns and word endings. In an implicit learning task, adult native speakers of English were exposed to a continuous stream of synthesized Dutch nonsense input comprising 300 randomized repetitions of six bisyllabic target nonwords. After exposure, the participants were given a forced choice recognition test to judge which strings had been present in the input. Experiment 1 demonstrated that English speakers used trochaic stress to isolate strings, despite being unfamiliar with Dutch phonotactics. Experiment 2 showed benefits from invariance introduced by affricates, which are typically found at onsets of final syllables in Dutch diminutives. Together, the results demonstrate that diminutives contain prosodie and distributional features that are beneficial for word segmentation.

Isolating words in continuous speech is a major challenge faced by language learners. Various informational sources-for example, rhythmic and prosodic patterns (e.g., Jusczyk, Houston, & Newsome, 1999; Morgan, 1996; Morgan & Saffran, 1995), utterance boundaries (Brent & Siskind, 2001), phonotactic regularities (e.g., Mattys & Jusczyk, 2001b), transitional probabilities between phonemes or syllables (e.g., Saffran, Aslin, & Newport, 1996), and context- or position-sensitive allophony (e.g., Jusczyk, Hohne, & Bauman, 1999; Mattys & Jusczyk, 200Ia)-have been shown to aid learners in word segmentation. As compared with adult-directed speech (ADS), child-directed speech (CDS) contains more acoustic cues to word boundaries due to exaggerated stress patterns (Fernald, 1989) and shorter utterances with longer and more frequent pauses. Computational studies have confirmed that CDS constitutes input for word segmentation models that is superior to that of ADS (Aslin, Woodward, LaMendola, & Sever, 1996; Batchelder, 2002; Brent & Cartwright, 1996; Christiansen, Allen, & Seidenberg, 1998).

Jusczyk (1997) and Echols, Crowhurst, and Childers (1997) suggested yet another reason why CDS may be beneficial for word segmentation. They hypothesized "that many diminutive forms in English that are used in addressing infants have strong/weak patterns . . . (e.g., 'daddy,' 'mommy,' 'doggie,' 'cookie,' 'kitty,' etc.). Consequently, it is not implausible that infants in Englishspeaking environments might develop a bias for trochaic patterns" (Jusczyk, 1997, p. 108). Given that diminutives are pervasive in the CDS of many languages (Gillis, 1998; Jurafsky, 1996) and are often derived by adding unstressed suffixes to word stems, it is possible that stress regularization in CDS through diminutivization is a more general phenomenon. In Dutch, a language with lexical stress, diminutives make up about 20%-30% of all child-directed noun tokens (uillis, 1997), which increases the frequency of stressed/unstressed nouns in CDS to 74% of multisyllabic word types (Taelman & Gillis, 2000). In Russian, approximately 45% of noun tokens in CDS are diminutives, as compared with only 3% in the ADS of the same Russian mothers (Brooks, Kempe, Fedorova, & Mironova, 2002). Connectionist simulations of word boundary detection based on stress information of the 200 most frequent Russian nouns show superior performance when networks are trained on diminutive, rather than simplex, forms (Kempe, 2004), indicating that diminutivization leads to stress regularization. In Spanish, many words with atypical stress assignment (e.g., teléfono, "telephone") have regular penultimate stress when diminutivized(e.g., telefonito, "small telephone"). Across languages, stress regularization is a common effect of diminutivization.

Regularized stress patterns facilitate the use of a metrical segmentation strategy (Cutler, 1994; Cutler & Morris, 1988), by which listeners rely on knowledge about predominant rhythmic patterns in their language to detect word boundaries. …