Education, Language and Sociology Research ISSN 2690-3644 (Print) ISSN 2690-3652 (Online) Vol. 4, No. 1, 2023 www.scholink.org/ojs/index.php/elsr 21 Original Paper A Comparative Analysis of Except and Except for in the COCA Namkil Kang1 1 Far East University, South Korea Received: February 17, 2023 Accepted: February 28, 2023 Online Published: March 9, 2023 doi:10.22158/elsr.v4n1p21 URL: http://dx.doi.org/10.22158/elsr.v4n1p21 Abstract The ultimate goal of this paper is to show how similar except and except for are in the Corpus of Contemporary American English (COCA). A point to note is that except and except for exhibit the same pattern in six genres, whereas they show a different pattern in two genres. What this suggests is that except is 75% the same as except for in the genre analysis of the COCA. A further point to note is that except and except for exhibit the lowest similarity in the fiction genre, whereas they reveal the highest similarity in the spoken genre. The COCA clearly shows that except people is the most preferable one among Americans, followed by except death, except Mr, and except water, in that order. The COCA further shows that except for people (45 tokens) is the most preferable one for Americans, followed by except for Mr, except for cases, and except for things, in that order. Finally, this paper argues that 21.95% of 41 nouns are the collocations of both except and except for. This amounts to saying that except is 21.95% the same as except for in the analysis of 41 collocations. Keywords Euclidean distance, token, COCA, collocation 1. Introduction The main purpose of this paper is to provide a comparative analysis of except and except for in the Corpus of Contemporary American English. First, we aim to compare the use of except and that of except for in the eight genres of COCA. We, by comparing them, can see how similar except and except for are in the eight genres of the COCA. Put differently, we, by analyzing the ranking of except and except for in the eight genres of the COCA, can see how similar they are. Second, we aim to measure the distance between except and except for in the eight genres of the COCA. More specifically, we, by using the Euclidean distance, see how close except and except for are. Note that the more except and except for are close, the more they exhibit a similarity. Third, we aim at comparing the collocation of except and that of except for in the COCA. By comparing the collocations of except and except for in the COCA, we can see how similar they are in the COCA. Finally, we aim to capture the degree of the www.scholink.org/ojs/index.php/elsr Education, Language and Sociology Research Vol. 4, No. 1, 2023 22 Published by SCHOLINK INC. similarity between except and except for by using the software package NetMiner. By linking the collocations of except and except for, we can calculate the degree of the similarity between except and except for. 2. The COCA and Eight Genres In what follows, we aim to inquire into the similarity between except and except for in the eight genres of the COCA. Table 1 shows the frequency of except and except for in the COCA: Table 1. Frequency of Except and Except for GENRE ALL BLOG WEB TV/M SPOK FIC MAG NEWS ACAD Except 97,441 14,256 16,003 12,686 7,757 20,688 9,740 7,971 8,310 Except for 27,234 3,440 3,580 3,467 2,139 6,857 2,778 2,455 2,518 It is probably worthwhile noting that the overall frequency of except in the COCA is 97,441 tokens, whereas that of except for is 27,234 tokens. Put differently, the overall frequency of except is three times higher than that of except for. It seems thus reasonable to hypothesize that Americans prefer using except to using except for. Perhaps it is worthwhile pointing out that except and except for rank first (20,688 tokens vs. 6,857 tokens) in the fiction genre. This in turn means that the types except and except for exhibit a similarity in the fiction genre. Simply put, they exhibit the same pattern in rank-one. It should be pointed out, however, that in the fiction genre, the use of except is by far higher (three times) than that of except for. We take this as meaning that American writers prefer using except to using except for in their novels. It is particularly noteworthy that except and except for rank second (16,003 tokens vs. 3,580 tokens) in the web genre. Again, the types except and except for show the same ranking in the web genre, thereby revealing a similarity. It must be stressed, however, that in the web genre, the use of except (16,003 tokens) is even higher (more than four times) than that of except for. This in turn implies that except (16,003 tokens) may be preferred over except for (3,580 tokens) by American web designers. It is interesting to note that except ranks third (14,256 tokens) in the blog genre, whereas except for ranks third (3,467 tokens) in the TV/movie genre. Quite interestingly, the types except and except for show a different pattern, thus exhibiting no similarity. When it comes to the blog genre, the use of except (14,256 tokens) is much higher (more than four times) than that of except for (3,440 tokens). We take this as indicating that American bloggers prefer using except to using except for. It is worth observing, on the other hand, that in the TV/movie genre, the use of except (12,686 tokens) is still higher (more than three times) than that of except for (3,467 tokens). We take this as implying that American celebs like using except. www.scholink.org/ojs/index.php/elsr Education, Language and Sociology Research Vol. 4, No. 1, 2023 23 Published by SCHOLINK INC. It is worth mentioning that except ranks fourth (12,686 tokens) in the TV/movie genre, whereas except for ranks fourth (3,440 tokens) in the blog genre. What this suggests is that the types except and except for show a mismatch between themselves with respect to their ranking, hence exhibiting no similarity. It is interesting to point out that except and except for rank fifth (9,740 tokens vs. 2,778 tokens) in the magazine genre. More interestingly, the types except and except for exhibit the same ranking in the magazine genre, thereby showing a high degree of similarity in the magazine genre. It must be pointed out, on the other hand, that in the magazine genre, the use of except (9,740 tokens) is even higher (more than three times) than that of except for (2,778 tokens). This in turn shows that American journalists are fond of using except (9,740 tokens) rather than using except for (2,778 tokens). It is worth pointing out that except and except for rank sixth (8,310 tokens vs. 2,518 tokens) in the academic genre. Again, the two types reveal the same property, thus showing a similarity again. It should be noted, however, that in the academic genre, the use of except (8,310 tokens) is by far higher (more than three times) than that except for (2,518 tokens). It can thus be inferred that teachers in America prefer using except rather than using except for. It is interesting to observe that except and except for rank seventh (7,971 tokens vs. 2,455 tokens) in the newspaper genre. Exactly the same can be said of rank-seven. That is to say, the two types show the same property in rank-seven, thus revealing the similarity between them. It is important to mention, on the other hand, that the use of except (7,971 tokens) is still higher (more than three times) than that of except for (2,455 tokens). It seems thus reasonable to assume that American journalists are keen on using except. It is worth noting that except and except for rank eighth (7,757 tokens vs. 2,139 tokens) in the spoken genre. Again, the types except and except for exhibit the same ranking in the spoken genre, hence revealing a similarity in rank-eight. It is significant to note, however, that the use of except (7,757 tokens) is much higher (more than three times) than that of except for (2,139 tokens). This in turn indicates that Americans like using except in daily conversation. To sum up, except and except for exhibit the same pattern in six genres, whereas they show a different pattern in two genres. From all of this, it seems clear that except is 75% the same as except for in the genre analysis of the COCA. Now attention is paid to the percentage of except and except for in each genre: www.scholink.org/ojs/index.php/elsr Education, Language and Sociology Research Vol. 4, No. 1, 2023 24 Published by SCHOLINK INC. Figure 1. Percentage of Except As exemplified in Figure 1, the fiction genre is the most influenced by except, followed by the web genre, the blog genre, the TV/movie genre, the magazine genre, the academic genre, the newspaper genre, and the spoken genre, in that order. Figure 2. Percentage of Except for www.scholink.org/ojs/index.php/elsr Education, Language and Sociology Research Vol. 4, No. 1, 2023 25 Published by SCHOLINK INC. As illustrated in Figure 2, the fiction genre is the most influenced by except for, followed by the web genre, the TV/movie genre, the blog genre, the magazine genre, the academic genre, the newspaper genre, and the spoken genre, in descending order. 3. The Euclidean Distance In the following, we aim at investigating the similarity between except and except for in the eight genres of the COCA. Note, to begin with, that the Euclidean distance provides the similarity index between two types (except vs. except for) in each genre. It indicates that the more the distance between two types is close, the more they exhibit a similarity. Now let us define the Euclidean distance as follows: (1) Euclidean distance Now attention is paid to the distance between except and except for in each genre: Table 2. Euclidean Distance between Except and Except for GENRE BLOG WEB TV/M SPOK FIC MAG NEWS ACAD Percentage of except 14.63 16.42 13.01 7.96 21.23 9.99 8.18 8.53 Percentage of except for 12.63 13.14 12.73 7.85 25.17 10.2 9.01 9.24 Euclidean distance 2 3.28 0.28 0.05 3.94 0.21 0.83 0.71 It is important to note that except is the furthest from except for in the fiction genre. To be more specific, the Euclidean distance between except and except for is 3.94, which is the highest among eight genres. This in turn suggests that the types except and except for reveal the lowest similarity. It is significant to note, on the other hand, that except is the nearest to except for in the spoken genre. More specifically, in the spoken genre, the Euclidean distance between except and except for is 0.05, which is the lowest among eight genres. This in turn implies that the types except and except for exhibit the highest similarity in the spoken genre. As exemplified in Table 2, the spoken genre reveals the highest similarity and the magazine genre, the TV/move genre, the academic genre, the newspaper genre, the www.scholink.org/ojs/index.php/elsr Education, Language and Sociology Research Vol. 4, No. 1, 2023 26 Published by SCHOLINK INC. blog genre, the web genre, and the fiction genre follow. Thus, it can be concluded that except and except for show the highest similarity in the spoken genre. 4. The Collocations of Except and Except for in the COCA In what follows, we aim to examine the collocations of except and except for in the COCA. Also, we compare the collocation of except and that of except for in the COCA. In both cases, our list was cut off in the top 25: Table 3. Collocation of Except in the COCA Number Collocation Frequency 1 Except people 72 2 Except death 46 3 Except Mr 41 4 Except water 34 5 Except Christmas 33 6 Except oil 27 7 Except time 27 8 Except food 24 9 Except money 22 10 Except sleep 21 11 Except play 20 12 Except wait 20 13 Except sex 19 14 Except treason 19 15 Except watch 19 16 Except work 19 17 Except men 18 18 Except talk 18 19 Except family 17 20 Except lag 16 21 Except Islam 15 22 Except thanksgiving 15 23 Except things 15 24 Except apple 14 25 Except chicken 14 www.scholink.org/ojs/index.php/elsr Education, Language and Sociology Research Vol. 4, No. 1, 2023 27 Published by SCHOLINK INC. It is important to mention that the expression except people has the highest frequency (72 tokens) in the COCA. This in turn means that the collocation people along with except is the most preferred one for Americans. It is interesting to consider the expression except death. It is worthwhile noting that the expression except death ranks second (46 tokens) in the COCA. This in turn implies that it is the second most widely used one (46 tokens). It is interesting to note that the expression except Mr ranks third (41 tokens) in the COCA. It is worth observing, on the other hand, that the expression except water ranks fourth (34 tokens) in the COCA. This in turn indicates that it is the fourth most occurred one (34 tokens). It seems thus reasonable to hypothesize that except people is the most preferable one among Americans, followed by except death, except Mr, and except water, in that order. It should also be pointed out that except time ranks sixth (27 tokens) in the COCA, whereas except money ranks ninth (22 tokens). To sum up, the expression except people is the most preferable one for Americans (72 tokens). Now attention is paid to the collocation of except for: Table 4. Collocation of Except for in the COCA Number Collocation Frequency 1 Except for people 45 2 Except for Mr 22 3 Except for cases 20 4 Except for things 18 5 Except for emergency 16 6 Except for Mrs 16 7 Except for emergencies 15 8 Except for defense 14 9 Except for work 14 10 Except for food 12 11 Except for money 12 12 Except for age 11 13 Except for children 11 14 Except for school 11 15 Except for family 10 16 Except for parts 10 17 Except for aunt 9 18 Except for college 9 19 Except for compliance 9 20 Except for holidays 9 www.scholink.org/ojs/index.php/elsr Education, Language and Sociology Research Vol. 4, No. 1, 2023 28 Published by SCHOLINK INC. 21 Except for right 9 22 Except for water 9 23 Except for gender 8 24 Except for sex 8 25 Except for provisions 8 It is significant to note that the expression except for people has the highest frequency (45 tokens). This in turn shows that except for people is the most preferred (45 tokens) by Americans. It should be noted, on the other hand, that except Mr ranks third (41 tokens) in the COCA, whereas except for Mr ranks second (22 tokens). This provides confirming evidence that except and except for exhibit a similar property with respect to the use of their collocation. It is interesting to observe the expression except for cases. The expression except for Mr is followed by the expression except for cases and the latter ranks third (20 tokens) in the COCA. It seems thus safe to assume that the expression except for cases is the third most widely used one (20 tokens) in the COCA. It should also be emphasized that except for things ranks fourth (18 tokens) in the COCA. It seems thus reasonable to hypothesize that except for people (45 tokens) is the most preferable one for Americans, followed by except for Mr, except for cases, and except for things, in that order. Additionally, it must be noted that except work ranks thirteenth (19 tokens) in the COCA, whereas except for work ranks eighth (14 tokens). To sum up, except people and except for people have the highest frequency, respectively, which in turn implies that they are the most preferable ones for Americans. Now let us turn our attention to the visualization of the collocations of except and except for. We attempt to capture the collocations of except and except for in terms of the software package NetMiner. By linking their collocations, we can see how similar except and except for are. As exemplified in Figure 3, 16 nouns are linked to except and except for, respectively. Most importantly, only 9 nouns are linked to both except and except for. Note that in both cases, our list was cut off in the top 25: www.scholink.org/ojs/index.php/elsr Education, Language and Sociology Research Vol. 4, No. 1, 2023 29 Published by SCHOLINK INC. Figure 3. Visualization of the Collocations of Except and Except for It must be emphasized that only 9 of 41 nouns are linked to both except and except for. These nouns are the collocations of both except and except for. The nine nouns linked to both except and except for are work, Mr, food, people, money, family, sex, water, and things. From all of this, it seems evident that 21.95% of 41 nouns are the collocations of both except and except for. It seems thus reasonable to conclude that except is 21.95% the same as except for in the analysis of 41 collocations. For the visualization of synonyms and keywords, see Kang (2022a, 2022b, 2022c, 2022d, 2023a, 2023b). 5. Conclusion To sum up, we have shown how similar except and except for are in the COCA. In section 2, we have argued that except and except for exhibit the same pattern in six genres, whereas they show a different pattern in two genres. It seems thus clear that except is 75% the same as except for in the genre analysis of the COCA. In section 3, we have further argued that except and except for exhibit the lowest similarity in the fiction genre, whereas they show the highest similarity in the spoken genre. In section 4, we have contended that except people is the most preferable one among Americans, followed by except death, except Mr, and except water, in that order. We have also contended that except for people (45 tokens) is the most preferable one for Americans, followed by except for Mr, except for cases, and except for things, in that order. Finally, we have shown that 21.95% of 41 nouns are the collocations of www.scholink.org/ojs/index.php/elsr Education, Language and Sociology Research Vol. 4, No. 1, 2023 30 Published by SCHOLINK INC. both except and except for. This in turn implies that except is 21.95% the same as except for in the analysis of 41 collocations. References Kang, N. (2022a). A Comparative Analysis of Search for and Look for in Four Corpora. Advances in Social Sciences Research Journal, 9(3), 168-178. Kang, N. (2022b). A Comparative Analysis of Impressed by and Impressed with in Two Corpora. Theory and Practice in Language Studies, 12(5), 819-827. Kang, N. (2022c). On Speak to and Talk to: A Corpora-based Analysis. Theory and Practice in Language Studies, 12(7), 1262-1270. Kang, N. (2022d). On Speak with and Talk with: A Corpora-based Analysis. International Journal of Social Science and Human Research, 5(8), 3354-3360. Kang, N. (2023a). K-Pop in BBC News: A Big Data Analysis. Advances in Social Sciences Research Journal, 10(2), 156-169. Kang, N. (2023b). K-Dramas in Google: A NetMiner Analysis. Transaction on Engineering and Computing Sciences, 11(1), 193-216.