A quantitative investigation of English adnominal modifiers

A quantitative investigation of English adnominal modifiers Tsy Yih1 Haitao Liu✉ Zhejiang University Zhejiang University Department of Linguistics Department of Linguistics yezi_leafy@hotmail.com lhtzju@gmail.com 1 Tsy Yih is the transliteration of the name of the firth author in his mother tongue, Shanghai Wu Chinese. He is also known as ZI YE in Mandarin pinyin. Abstract The present study employs the valency theory to approach the internal structure of noun phrases. Specifically, we adopt The Georgetown University Multilayer Corpus (GUM) to investigate the valency distribution, mean valency, and valency pattern of three subclasses of nominal heads, commons nouns, proper nouns, and pronouns. It is found that: 1) The mean valency of common nouns is significantly larger than that of proper nouns, which is far larger than that of pronouns. 2) The Mean Valency Pattern (MVP) newly proposed in this paper offers a better comparative basis for the valency distribution of different categories than the original Probabilistic Valency Pattern (PVP). 3) The three nominal classes manifest different valency patterns. Common nouns have a strong combining ability with determiners, adjectival modifiers, prepositions phrases and nominal modifiers. The combining ability of pronouns is near zero for most groups except for a few cases in certain constructions, whereas the mean valency of proper nouns are not close to zero as expected due to the improper treatment of complex proper names in annotation. The findings are expected to pave the way for further quantitative studies into the frequency structure of noun phrases. 1 Introduction The noun is one of the most important lexical category in human language, which accounts for about 37% of word-tokens in texts (Hudson, 1994). The internal structure of English noun phrases has been a long and constant topic in theoretical linguistics, which has received much attention in the literature (Quirk et al., 1985; Rijkhoff, 2002; Keizer, 2007; Davidse and Breban, 2019). A popular theory is that the noun phrase has a layered structure (Rijkhoff, 2002) or an orbit structure (Lu, 1993). Seen from the coarsest perspective, English adnominal modifiers could be divided into prenominal lexical categories such as determiners, quantifiers, adjectives, nominal modifiers, as well as postnominal phrasal and clausal modifiers, including prepositional phrases and relative clauses (Quirk et al., 1985; Halliday, 1985). Some have studied the finer internal structure of the category, adjective (Halliday, 1985; Bache, 2000; Davidse and Breban, 2019), and even further splits might be employed to the inside world of descriptive adjectives (Dixon, 1977). Besides their own theoretical value, fine and elaborate classifications would also provide good basis for developing high-quality language resources for natural language processing, thereby being fundamental to computational linguistics. So far, all the above-mentioned studies investigate the theoretical structure of noun phrases, with special attention paid to the preferred order when multiple modifiers occurred. For instance, we have roughly presented the above-mentioned modifiers in the order with regard to their natural closeness to the head noun (demonstrative < quantifier < adjective <nominal modifier < head noun). However, there is a lack of quantitative studies on …

Content truncated to 500 words