Studying restrictions on patterns of word-formation by means of the Internet: 1875
Franz Rainer · WU Research · 2003
The Internet has not been conceived as a tool for linguistic research, but being a huge collection of texts it has turned out to be also an important source for linguists. Many studies exploiting this new source of information have been published over the last few years, and many more will undoubtedly see the light in the near future. The present study explores in how far Internet data may be fruitfully used to gain new insights about the nature of restrictions on patterns of word formation. As a test case, I have chosen the Italian intensive suffix -issimo, which I had already studied some twenty years ago on the basis of a corpus of a hundred novels and, to a certain extent, through the elicitation of acceptability judgements from native speakers. It is shown that, in the case of a highly productive suffix like -issimo, the absence from the Internet may be interpreted as a good indicator of low acceptability, an inference that is generally not allowed by smaller corpora. Another advantage of the Internet data is that it nicely brings to light the gradual nature of restrictions on patterns of word-formation. 1 1. The Internet as a research-tool for the linguist As is well-known, the Internet was originally designed as a means of making military communication less vulnerable and then put to use in scientific, commercial and private communication. None of those who developed this new medium ever had the slightest intention of facilitating linguistic research, but since the Internet constitutes a huge collection of texts it didn’t take long for linguists to discover its immense potential as a research tool. It is not my intention here to review the numerous ways the Internet has already been put to use in linguistic research over the last few years, but to explore one more possible application which, to the best of my mind, has not yet been exploited. It will be our goal to explore in how far data gathered from the Internet may be used as a substitute for acceptability judgements in word-formation and whether such data may help to improve the description of restrictions on patterns of word-formation. One of the notorious disadvantages of corpora is the fact that they contain no negative evidence. Now, the Internet is so huge a corpus that one might suspect that the absence of certain instantiations of at least