A Polish-English Contrastive Study of the Order of Premodifying Adjectives: A Procedural Model Account

Some restrictions on the order of English premodifying adjectives were already pointed out by Whorf (1956). The first manual corpus studies of the order of premodifying adjectives were carried out by Goyvaerts (1968), Vendler (1968), Quirk and Greenbaum (1973), and Dixon (1982). In all, I have encountered over a hundred of studies concerning the ordering of premodifying adjectives. As a result of such studies, linguists agree that in case of the need to premodify a noun with more than one adjective, each of them representing one of the semantic categories: 1. “opinion”, 2. “size”, 3. “shape”, 4. “age”, 5. “colour”, 6. “origin”, 7. “religion”, 8. “material”, the adjectives must follow the order above mentioned. Interestingly, similar types of restrictions have been reported in numerous unrelated languages such as Hungarian, German, French (where the order of modifying adjectives is a mirror reflection of that in English), Chinese, and many others, which indicates a universal character of the phenomenon. Polish grammar books claim, however, that there are no similar restrictions in Polish. Below, in section one, I report on a contrastive Polish-English study, which shows that in Polish restrictions on the order of the semantic classes of premodifying adjectives similar to those in English are clearly visible statistically, while in English they are not as strict as it is commonly believed. In a very short section two, I mention some classical explanations of the phenomenon. In section three, I account for the phenomenon studied based on the procedural model of language introduced by Zielinska (2007a, 2007b), an approach, which explains the issue better than the previous research known to me has done. 1. A Polish-English Contrastive Study of the Order of Premodifying Adjectives The English data presented in this paper have been taken from the BNC via the BYU interface (the interface created by Mark Davies of BYU). The interface allowed me to create lists of English adjectives representing the semantic categories listed in the previous section and next have the computer check for me all instances of two adjectives following each other – the first one from one list and the second one from another list. This was done for all possible list combinations. I also searched the instances of the adjectives from the lists considered being separated with one, two, or three other adjectives, so as not to overlook situations when the adjectives of the categories 1 Department of English, The Jagiellonian University e-mail: [email protected] considered are separated by another adjective, which fact does not change the relative order of the categories investigated. So as not to overlook situations in which an additional adjective separates the given two adjectives from the noun they modify, I did not ask the search engine to list only the adjective strings immediately followed by the noun, but checked manually which out of the items found represent strings of adjectives premodifying a noun. To collect Polish data, I used the IPI Korpus – the corpus of the Polish language collected by the Institute of Computer Science Foundations of the Polish Academy of Science. Since it is not possible to search with lists there, I queried the Polish corpus by asking for strings, e.g. [a given adjective + any adjective+ (+any adjective) + noun], and [any adjective + the given adjective+ (+any adjective) + noun] etc. in which “the given adjective” was an adjective representing one of the categories investigated. Next, I manually subcategorized the data found according to the semantic category of the second adjective (referred to as “any adjective” in the search request above) into 7 categories: “opinion,” “size,” “shape,” “age,” “colour,” “origin,” “material.” Finally, I counted the number of occurrences of strings of adjectives premodifying a noun in which the first of the adjectives represents the semantic category “a,” while the second the semantic category “b,” where “a” and “b” stand, in turn, for each of the seven categories just mentioned. The data collected has been presented in tables 1 to 9. Tables 1 and 5 contain the numbers of noun phrases modified by at least two adjectives of the relevant categories, in English and Polish respectively. In tables 2 and 6, the cell (a, b) contains the percent of occurrences of noun phrases in which an adjective belonging to the semantic category “a” precedes an adjective representing the semantic category “b” in relation to all the occurrences of noun phrases in which a given noun is premodified by adjectives representing the semantic categories “a” and “b” – attested in the same two corpuses – along with the relevant statistical error. Tables 3 and 4 present the English data derived from table 1 while tables 7 and 8 the Polish data derived from table 5 after grouping the categories concerned into the following joined categories: “opinion/size”, “shape/age/colour” and “origin/material”. It turned out that combining categories in such a way shows a stronger preference for maintaining the order between such joined categories than between the categories within those joined categories. Table 9 compares the degree of predominance of the order of the selected categories of adjectives in English in comparison to Polish.