英语论文网

se. It can be illustrated as a coordinate with P on the horizontal axis and V on the vertical axis. V is the number of different types of a given affix in the corpus. The shortcoming of this measure is that 'in general it is unclear how to evaluate the different contributions of P and V to the global productivity when arbitrary affixes are compared' (Plag, 1999, p.32). P* should be calculated as the proportion of the number of hapax of a given affix and the total number of hapaxes of arbitrary constituency in the corpus. At last, most of the linguists suggest correlating the two measures. The use of P is to distinguish between productive and unproductive while P* is to rank productive process. The reason that P* can be seen as the measure of ranking productive process is that P* is seen as a method for the contribution of a particular morphological process to the extent use of the words.

2.3 Productivity as a psycholinguistic notion

Baayen also has a third measure of productivity, the activation level A (1993, p.195-196). It is defined as the number of tokens representing those types of a given affix whose frequency of occurrence is smaller than a threshold θ. The measure is motivated by the idea that people process complex words through two competing routes simultaneously: by parsing and by retrieving the whole word directly from their mental lexicon. Which route is faster depends on the frequency of the word: common words are readily available in the lexicon, while words rarer than the threshold value are processed by identifying and combining the affix and base. This process of parsing maintains the activation level of the affix; the level indicates how quickly the affix will be recognized and combined with the base. Baayen (1993, p.203) admits that choosing a suitable threshold value is problematic. He assumes that the only words that are semantically transparent will maintain the activation level. The higher θ is, the more semantically opaque words it will include: words are usually the more opaque the more frequent they are, because frequent words have accumulated more meanings or their meaning may have changed. Therefore, the threshold is best kept fairly small. Hay (2001) improves on this measure by introducing the concept of relative frequency; she shows that it is not the frequency of the affixed word alone that affects portability, but rather the relation between the frequencies of the affixed word and its base. Her work will not be discussed further here because it is not directly applicable to unlemmatised historical corpora, which would be a time-consuming task tracking down all of the different spelling variants of the base and searching for each of them. In addition to the semantic transparency mentioned above, phonological transparency has also been shown to affect productivity. According to Hay and Baayen (2003, p.105), there is a wealth of evidence that English speaking people use phonotactic probabilities when parsing words they hear. For example (2003, p.105), as the /pf/ transition in pipeful is unlikely to occur within a mono-morphemic word in English, people hearing this combination in running speech will immediately post a morpheme boundary between the pipe and ful, which will help them to decompose the composite word. These unlikely combinations facilitate the parsing route rather than the whole-word r本论文由英语论文网提供整理，提供论文代写，英语论文代写，代写论文，代写英语论文，代写留学生论文，代写英文论文，留学生论文代写相关核心关键词搜索。