korean word frequency (counting philosophy as a content analysis tool)

[yaoweiwei535]

Version number of KH Coder

3.Beta.08e

Your Qeuestion

I apologize for disturbing you. While analyzing Korean data, I noticed that many words (verbs and adjectives) in Korean are composed of "N (noun) + 하다." However, in the word frequency results, "N (noun)" and "하다" appear separately, rather than as a whole "N하다," leading to inaccuracies in the analysis. Additionally, if I specify a string to set "N하다" as a single string "TAG," it still doesn't yield the correct result due to the inflection of the word. Do you have any suggestions on how to solve this issue? Thank you for your response.

[ko-ichi-h]

Can you tell me what "하다" means?

Is there such a big difference between "N하다" and "N"?

[ko-ichi-h]

If you want to count "N" and "N하다" separately, (A1) you may use coding rule like:

*N
N and not N+하다

*N하다
N+하다

If necessary please force pick up 하다 as a word. Also please select "sentences" as the coding unit.

(A2) Or you can just see "Collocation Stats" of the "N."
https://khcoder.net/en/gallery/pages/image/imagepage4.html

You can open it by clicking "Stats" button in lower left corner of KWIC window.

(B) If you want to count or analyze other things regarding "N하다", please specify your exact goal. Maybe, we can find the solution to achieve the goal.

[yaoweiwei535]

In Korean, "하다" can function as a standalone verb, but at the same time, it is also a suffix in many words, forming compound words with many nouns. For example,“exercise”, the verb "운동하다" is a compound word made up of "운동" and "하다". However, in KH Coder's word frequency analysis, "운동" and "하다" are counted separately as two different words. This causes a discrepancy because I need the word frequency of "운동하다". However, I can't "force pick up" "운동하다" as a whole, because it may not always appear in the text in the form of "운동하다". It could appear in its past tense as "운동했". I'm not sure if my explanation is clear. Thank you for your response.

[ko-ichi-h]

I need the word frequency of "운동하다".

Then, you can use "Collocation Stats" of "운동".
https://khcoder.net/en/gallery/pages/image/imagepage4.html

1. Start KH Coder and open the project
2. Go to [PRe-Processing] -> [Select Words to Analyze], check [Other] in the bottom of "parts of speech" section, and click "OK"
3. Go to [Tools] -> [Words] -> [KWIC Concordance] and search for "운동"
4. Click "Stats" button in lower left corner of the "KWIC Concordance" window to open "Collocation Stats" window
5. If necessary, select [R1] as the "Sort:" in the bottom of the "Collocation Stats" window

Find "하다" line and "R1" number. If R1 number is 10, "운동하다" appeared 10 times.

[yaoweiwei535]

Thank you for your response; this is a very good solution. However, Korean has many compound words. For example, the noun for "publication" is "출간," and the verb forms are "출간하다" or "출간되다," which are compound words of "N+하다" and "N+되다," where "하다" and "되다" function as suffixes.

As shown in Figure 1, when I search for "출간," the results mostly include forms like "출간했" or "출간된," which are the past tense of the verb "출간하다" and the attributive form of "출간되다." These should be counted as verbs, not included in the noun frequency statistics. However, when I search for "출간하다," there are no results. This is a characteristic of Korean, where many words are structured as "N+하다," leading to analysis results where "N" and "하다" are separated. Since "하다" is a suffix, it should not be counted as a separate word in the overall frequency statistics for "하다." Due to the verb's conjugation, I cannot "force pick up" the entire verb form, which results in inaccurate statistics.

Given that Korean has many compound words, it would be very time-consuming to use this method for all of them. I would greatly appreciate it if you could upgrade the version of the Korean word frequency statistics. Thank you for your response.

[ko-ichi-h]

Thank you for the valuable feedback.

But currently, as a content analysis tool, KH Coder is developed with the following philosophy.

Counting "Publication (출간)" as a whole is the priority goal, and distinguishing between "Doing Publication (출간하다)" and just "Publication (출간)" is the next step.

It doesn't matter what conjugation it is, or how it forms a compound word. We just want to count how many times the concept "Publication (출간)" appears. That number is more important for statistically understanding the content of the text. When you break down the numbers into frequency of "Publications (출간)" and frequency of "Doing Publications (출간하다)," it may become more difficult to statistically understand the content of the text. Again, content analysis of the text is the goal of KH Coder.

If you need such a distinction between "publication (출간)" and "doing publication (출간하다)", please use Collocation Stats or coding rules mentioned above. To make such a distinction for the top 100 nouns, for example, you just need to do the same operation 100 times. Input the new word to KWIC window and hit Enter key. Collocation Stats window will update automatically. Or, just make 100 entries in the coding rule file. (If necessary, you can develop a Perl plugin that operates KH Coder automatically. If you do not want to develop it yourself, you can outsource it to a software company.)

[yaoweiwei535]

Thank you for your response. Have a good time！