Coding issues with isolation of verbal morphology

[Harry Caulton]

I am attempting to isolate verbs by morphological type in the Brown corpus starting with the past tense. I have found that using:

freq +sm|*v - t*CHI +u *.CHA

It allows me to isolate verb stems and all morphological suffixes. However, when I change this code to:

freq +sm;*v,|-past - t*CHI +u *.CHA

I get 0 results. Which is confusing as I've added the marker for regular past tense morphology which should logically yield all regular past test verbs. I've reread the CLAN and MOR manuals but I cannot seem to find anything related to where to place the relevant information in the code.

Would anyone be able to assist me here? Many thanks in advance for your time.

[Davida Fromm]

I'm not sure this is the most elegant/efficient way to write the command, but I think this works ... if I'm understanding the output you want.  Hope it helps.

freq +sm|v:*,|v,;*,-PAST +t*CHI +u *.cha

-Davida

--
You received this message because you are subscribed to the Google Groups "chibolts" group.
To unsubscribe from this group and stop receiving emails from it, send an email to chibolts+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/chibolts/cead221b-ceca-45ea-b606-6aece35dfd79%40googlegroups.com.

[Leonid Spektor]

Harry,

You need to understand the meaning of code symbols used by +sm option first. To get a full listing look in the CLAN manual or type "freq +sm" in Commands window, press return key.

The codes you are interested in are part of speech denoted by "|" symbol and PAST that can be denoted by either "-" or "&" symbols. Those symbols come straight from the words on %mor tier. For example word "worked" has %mor form "v|work-PAST" and word "went" has %mor tier form "v|go&PAST". 

You want to find all child speaker's verbs that are also in past tense, so your command would be:

freq +sm|v*,-PAST +sm|v*,&PAST +t*CHI +u *.CHA

Leonid.

[Harry Caulton]

Thank you -it really did! I tried this code and it allowed me to see by speaker the produced stems with regular past tense morphology. I substituted -PAST for &PAST and this allowed the same for irregular past tense. Then I changed it to -t*, as I'm curious about a correlation between input on production over time and it allowed me to see the information I was after.

On Tuesday, 12 November 2019 15:45:58 UTC, Davida wrote:

I'm not sure this is the most elegant/efficient way to write the command, but I think this works ... if I'm understanding the output you want.  Hope it helps.

freq +sm|v:*,|v,;*,-PAST +t*CHI +u *.cha

-Davida

On Tue, Nov 12, 2019 at 10:26 AM 'Harry Caulton' via chibolts <chib...@googlegroups.com> wrote:

I am attempting to isolate verbs by morphological type in the Brown corpus starting with the past tense. I have found that using:

freq +sm|*v - t*CHI +u *.CHA

It allows me to isolate verb stems and all morphological suffixes. However, when I change this code to:

freq +sm;*v,|-past - t*CHI +u *.CHA

I get 0 results. Which is confusing as I've added the marker for regular past tense morphology which should logically yield all regular past test verbs. I've reread the CLAN and MOR manuals but I cannot seem to find anything related to where to place the relevant information in the code.

Would anyone be able to assist me here? Many thanks in advance for your time.

--
You received this message because you are subscribed to the Google Groups "chibolts" group.

To unsubscribe from this group and stop receiving emails from it, send an email to chib...@googlegroups.com.

[Harry Caulton]

Hello,

Thank you. Honestly, your explanation was exactly what I was looking for. I was unaware that one may use commas to combine two specifications when searching. I can see that on the list (freq +sm command +return) there are commas used to that effect, but thank you for demonstrating their function. Seeing examples of what-goes-where in coding (with clear explanations as to why) is really useful to people like myself who are unfamiliar with it, so thank you! 

Leonid.

To unsubscribe from this group and stop receiving emails from it, send an email to chib...@googlegroups.com.