4. Conclusion

This discussion has aimed at giving an insight into the basics of acoustic phonetics as well as providing the reader with a good understanding of the possible applications of Praat. I think that Praat is a very good program in terms of analysing speech data. It is well structured and after a few practice hours it can be easily handled.

Nevertheless, there are a few minor inconveniences, which may convey the impression one is working with a buggy beta version of the yet unfinished release version e.g. the deleting and zooming functions, which could have been designed in a more user-friendly fashion. And in a way this is true. Praat is a work in progress. The designers Paul Boersma and David Weenink are constantly working to make the program fit linguists' needs by correcting minor mistakes. For more detail on the latest improvements to Praat please visit http://www.fon.hum.uva.nl/Praat/manual/What_s_new_.html.

Obviously acoustic phonetics is a multifaceted sub-field of phonetics which provides linguists with numerous acoustic aspects of speech sounds to be investigated. Understandably enough, it was impossible for me to address all of these in great detail as this would have exceeded the scope of my work. In general, however, one must always decide which features he/she wants to focus on in his/her analysis, based on which information may prove significant for the particular research at hand. Here, as this paper has shown, Praat (Dutch = “talk”) comes in handy. It offers versatile and up-to-date methods of analysis. Some of its key features in terms of speech analysis are spectral analysis, pitch analysis, formant analysis, intensity analysis, jitter, shimmer, voice breaks, cochleagram and excitation pattern. But aside from these, the application offers a seemingly endless number of different options. It can help generate waveforms, wide and narrow spectograms, and pitch tracks. It shows an intensity contour, extracts sounds for further detailed analysis, measures the voice onset time and offers the possibility of enhancing specific frequency areas, labelling individual words, syllables or phonemes. Furthermore, the program can run on a wide range of operating systems (Unix, Mac and Windows) and, above all, is free of charge.  

As for the results retrieved from my own study, I have come to the conclusion that both speakers' vowel charts, which I was able to create based on the formant frequencies indicated in the specogram (formant analysis), differ very much. But despite this fact, the accent of informant B, the learner of English, shows only a very small number of features which can distinguish her as a learner of English. That may be due to her living in England for almost five years. And it is interesting to see that some features of our accent never really disappear.

One indication of this is that, in comparison to informant A, informant B articulates the vowel /e/ at a more fronted and more open position. A plausible reason for this phenomenon could be that Germans do not differentiate between /e/ and /æ/, but only use the intermediate vowel /ɛ/. The German /e/ is therefore articulated at a slightly lower position than the English counterpart. This may lead to L1 interference on the ELL's part, who is likely to transfer certain characteristics which are typical of the German language over to the English language.

The same may hold true for the ELL's extreme realization of the fronted /ɪ/. While the /ɪ/ vowel is common to both German and English, it is not phonetically identical in the two languages, with Germans exhibiting a slightly higher articulation than the English. Again, this may indicate an instance of L1 interference.

Finally, the lack of differentiation between /ɔ:/ and /ɒ/ could be attributed to the fact that in German there is only either the short, open /ɔ/ or the long closed /o:/, which may cause German learners of English to have problems pronouncing a long, open /ɔ:/.

Created with the Personal Edition of HelpNDoc: Full-featured Help generator