speech is: they are simpler than native speaker-native speaker speech, and are not "finely tuned" to the level of the acquirer, but one sees signs of "rough tuning". To illustrate this point, I present a brief survey of the literature available to me at this time, arranged according to input features:

1. Rate. This is evidence that caretaker speech is at least roughly tuned with respect to rate of speech. Broen (1972) compared rate of speech to 5- and 2-year-olds, and found that younger children received slower input. Also, Cross (1977) reports a 0.36 (ns) correlation between rate of input to fast acquirers and their age. On the other hand, Cross found no relation between rate of input speech and the children's linguistic competence, as measured by a test of comprehension (r = -0.12, ns).

For teacher-talk, Henzl (1973), investigating the behavior of Czech speakers, concluded that teacher-talk was slower than native speaker-native speaker speech, but did not look for signs of fine tuning. Trager (1978) confirmed Henzl's findings for ESL at USC, and found some signs of tuning, with beginning ESL students receiving slower input than intermediate and advanced students. A recent study by Brunak and Scarcella (1979) confirms that interlanguage-talk is slower than native speaker speech. Also, I know of no data on foreigner-talk.

2. Lexicon. Broen (1972) reports that younger children receive a lower type/token ratio, supporting at least rough tuning for age for caretaker speech. Henzl (1973) also found that teacher-talk had a lower type/token ratio than native speaker-native speaker speech but did not look for tuning. Trager (1978), in a small-scale study (two ESL acquirers and two native speakers), concluded that the less advanced ESL performer received a lower type/token ratio.

3. Well formedness. Caretaker speech, according to Newport et al. (1977), is "unswervingly well formed". Newport et al., however, report that there is no evidence for tuning: caretaker speech is more well formed than adult-adult speech, but less mature children do not receive better formed input than more mature children. Cross (1977), however, reports good correlations for well formedness and linguistic

129