Collocation Tool

The Collocation tool is to count the context words that occur within the specified range of the keyword. Type the word/phrase you want to search, set the span, and hit Search button.

Values in red denote the most frequent positions. As in Word Count, you can set the minimum number of occurrences to include in the results. Go to Preferences -> Others -> Minimum Frequency.

The list can be sorted by clicking the header of each column. If you want to change the span after creating the collocation list, change the span values and click Rearrange.

Just as in the Word Count, you can filter the results.

If you select n-grams, you can create a list of n-grams as context words (collocates).

When n-grams are counted, the context words at L1 start withnth words from the keyword. With 2-gram selected, items on L1 starts with 2 words to the left of the keyword, items on L2 starts with 3 words to the left of the keyword, and so on. This means the same word is counted at mostntimes as a part of context n-grams of a single keyword (this is the same as counting n-grams in Word Count).

3.3.1.1 Collocation Statistics

If you have created a collocation list and a word list with the same files/corpus/database, you can calculate collocation statistics. Go to Stats -> Collocation and select the statistic you want to calculate.

A new column will be inserted to the left of LR Total. As you can see, when context words occur very infrequently, collocation statistics are often biased. So you might want to consider setting the minimum frequency to the list.

3.3.1.2 Collocation Visualizer

Just like calculating statistics, if you have created a collocation list and a word list with the same files/corpus/database, you can visualize collocation information based on frequency information and/or collocation statistics by clickingVisualizer button. This is an experimental feature, so the details or representation of values might change in the future.

On the Visualizer window, select a statistic you want to use, and select the information to use.

You can select either a specific column or a range. Click the radio button left to the pop-up buttons to select which one to use. If you select a specific column (upper), you can use it as a starting/

ending point of the span to the left or right of the keywords. So if you select L5and check Span, the information used is between L5 to L1. Then select how many context words on the list to be used. If the number of the context words is smaller than the specified number, all the context words on the list will be used.

context words within L5 ~ R5 are used.

Other options are the following:

Ignore zero occurrence - zero frequency words will be ignored

Include Freq Info - frequency info is used (in gray) in addition to the specified statistic Convert LL val to log - Log-Likelihood values are very large, so convert values to log Use Multiple info - if checked three statistics can be combined; assign colors to them Frequency information is represented in gray, so lower frequency words appears whiter.

The above settings returns the following visualization. The size of letters represents the main statistic, Log-log. The shade of letters represents the frequency. The color represents the combination of three statistic values.

You can check the statistic values by clickingStatsbutton. Two-finger or right-click on the table to allow you copy the statistic values.

3.3.1.3 Experimental Features

To use these features, you need to check Experimental Features in Preferences -> Others ->

Experiments.

3.3.1.3.1 File Frequency

This feature aggregates the number of files each collocate appears at a certain position. The numbers will be displayed in the brackets except for the LR Total frequencies. To use this feature, you need to check Record File Frequency in Preferences -> Others -> Collocation.

3.3.1.3.2 Word Chain

This feature record the chains of collocates with frequencies. If you click a collocate at L1 or R1 on the table, a list of words that appear before (L2) or after (R2) the selected collocate will be listed.

The items on the left (L1-L3) and on the right (R1-R3) are independent, so even when an item on the L1 list is selected, the items on the R1 list are not affected by the selection. To use this feature, you need to check Record File Frequency in Preferences -> Others -> Collocation.

3.3.1.4 Other features

3.3.1.4.1 Treating Keywords as a Single Word

When you use wildcard characters or search two or more different words, collocates are tallied

for each keyword. But if you check Treat Keywords as One Word in Preferences -> Others ->

Collocation, all the keywords are treated as one word and frequency counts are combined. This is useful when you search for collocates of different spelling variations of a single word or grammatically inflected forms of a single word.

3.3.1.4.2 Copying Results

If you want to copy the results, two-finger or right-click the table. You can paste the copied results as tab-delimited text.

3.3.1.4.3 Search in Concord

When you selectSearch in Concordon the context menu, the keyword will be the Search Word and the context word will be the Context Word on Concord.

3.3.1.4.4 Exporting Results

To export the results as a tab-delimited plain text file, go to File -> Export and specify the encoding.