• Keine Ergebnisse gefunden

I NFANT WORD SEGMENTATION FROM EXAGGERATED G ERMAN IDS

CHAPTER 3: LISTEN UP! DEVELOPMENTAL DIFFERENCES IN THE IMPACT OF

4.5.2. I NFANT WORD SEGMENTATION FROM EXAGGERATED G ERMAN IDS

Another important finding of the current study was that infants of 7.5 months of age showed a different neural response to familiarized compared to novel control words. In particular, they had a more negative neurophysiological response to familiarized words than to novel control words, which suggests that infants succeeded in recognizing the words previously presented to them in the familiarization phase. Importantly, infants succeeded in segmenting words only from the exaggerated IDS register.

In comparison with other ERP word segmentation studies, it appears to be the case that the time window showing significant differences between familiarized and control words is rather small. However, we note that other studies have also reported two separate effects similar to the ones in the current study (Junge, Cutler, &

Hagoort, 2014; Zangl & Mills, 2007). In particular, the N200-300 found in our study has been previously linked to word recognition and has been reported in other infant studies (Kooijman, Junge, Johnson, Hagoort, & Cutler, 2013; Männel & Friederici, 2013). Furthermore, the finding that the direction of this effect replicates that found in previous studies, with increased negative potentials to familiarized words relative to control words supports the comparison of the current findings to previous studies.

The results of successful segmentation only from exaggerated IDS underline the fact that German infants at 7.5 months might need additional cues in order to segment words from fluent speech. This finding is in line with previous behavioral studies on German infants’ speech segmentation. For instance, Altvater-Mackensen and Mani (2013) found that infants were only able to demonstrate segmentation of words from fluent speech if they had been previously familiarized with similar

CHAPTER 4: MATERNAL INPUT AND INFANT WORD SEGMENTATION sounding words. Similarly, even older infants at 9 months of age in Schreiner, Altvater-Mackensen, and Mani (2016) were only able to segment words from fluent speech when they had been familiarized with these words over a 6-week-period at home. Finally, infants at 7.5 months of age did not show successful segmentation of words from fluent speech even when tested on a more exaggerated IDS register, similar to that of American English, while 9-month-olds were able to (Schreiner &

Mani, 2017).

Hence, the finding of successful segmentation abilities from exaggerated IDS in 7.5-month-old German infants has important implications for our understanding of the linguistic development of German infants. First, it suggests that the cross-linguistic differences between American English and German infants’ ability to segment speech may lie in the difference of IDS modulations in the two languages. In addition, it appears that German infants just seem to be unable to demonstrate this ability in the context of the preferential listening paradigm (Schreiner & Mani, 2017).

Secondly, it underlines the fact that the use of different methodologies might yield different results.

The familiarity effect reported in the current study supports models of IDS as an attentional spotlight (Kuhl, 2007). That is, the use of a more exaggerated speech, such as a wider pitch range, which is very typical for IDS, may drive and maintain infants’ attention to the relevant stimuli and, hence, enhance infants’ ability to extract the individual words within a fluent speech stream.

The failure to find any potential relationship between infants’ segmentation abilities of fluent speech and later vocabulary outcomes (e.g., Junge, Kooijman, Hagoort, & Cutler, 2012) may, in part, be due to the weak reliability of parents’

reports on their infants’ comprehension skills (Friedrich & Friederici, 2006, 2010; Mills et al., 2005; Rämä et al., 2013; Torkildsen et al., 2006, 2008, 2009). We suggest,

CHAPTER 4: MATERNAL INPUT AND INFANT WORD SEGMENTATION therefore, that the absence of a significant relationship between infants’

segmentation skills and later vocabulary sizes has to be treated with caution.

Nevertheless, the marginally significant correlation between infants’

segmentation abilities of exaggerated German IDS and infants’ later language outcomes might suggest that infants who need a more exaggerated IDS in order to process segment words from fluent speech, may have smaller vocabularies later in life. Since these infants show improved recognition of words in exaggerated IDS, a register which does not exist in typical German adult-infant dyads, they might be slower in their vocabulary development – given that they are not exposed to such exaggerated IDS in their daily interactions.

4.6. CONCLUSION

The current study explored the potential impact of the quality of the maternal input on 7.5-month-old German infants’ word segmentation abilities using ERPs.

Significant correlations between infants’ segmentation abilities and the quality of the mothers’ input highlight the importance of IDS in early language acquisition, while improved understanding of the relationship between these two variables, we suggest, requires further analyses of longitudinal data.

In addition, contrary to previous research on word segmentation abilities in German infants, the results suggest that German infants at 7.5 months of age are able to segment exaggerated IDS. This finding suggests that the cross-linguistic differences in infants’ ability to segment fluent speech may lie in the difference of IDS modulations. In addition, it might also in part be due to methodological sensitivities, which should be considered in future research on infants’ segmentation abilities.

CHAPTER 5: THE IMPACT OF TEST REGISTER –