Roshan Lal Dewangan, you can use a data reduction technique (factor/pca) in SPSS.

If it's just a single phenomenon that participants provided adjectives for you might just keep it simple with a frequency distribution. This webpage looks like it might be helpful: https://libguides.library.kent.edu/SPSS/FrequenciesCategorical.

Hello Roshan,

Given the way you apparently collected your set of adjectives (e.g., each participant was asked to generate his/her own set of up to 3), I agree with Caleb A. Aldridge 's suggestion that your best bet (at the moment) is to sort the adjective list by relative frequency.

Unless you have some external criterion to use, collect some additional data in conjunction with the adjectives, or collect different information about the adjectives (such as, asking participants to rate the pertinence of each adjective in your full set with respect to the target phenomenon), I don't see much utility in other kinds of analysis.

Good luck with your work.

Let me see if I understand. 30 participants were shown something, and each provide 0, 1, 2, or 3 adjectives. So you have 30 by 50 matrix with mostly zeroes and some ones. Is that right?

The next question, is in what way do you want to cluster these together (e.g., if they co-occur with each other, if they are similarly rare, their syllable length). Without knowing that it would be foolish to guess, but you can look at https://web.stanford.edu/~hastie/ElemStatLearn/ which may be helpful. ALso, do you want dimensions or clusters?

@Daniel Wright Each participant gave atleast 1 adjective. I received total 78 adjectives, out of that 50 are unique (without overlap among participants) adjectives.

Now I want to reduce these 50 adjectives in few representative adjectives; can be dimensions or meta categories.

Is there any statistical or accepted way?