4

In the article, "How Many Words Do You Need to Know in Spanish (or any other foreign language)? And WHICH Words Should You Be Learning?" I came upon the following:

“Assume that a language learner is aiming for 90% coverage in each of the four parts of speech that represent open classes — nouns, verbs, adjectives, and adverbs. This 90% figure will be obtained by knowing about 2600 nouns, 230 verbs, 980 adjectives, and 50 adverbs, or a total of about 3800 total forms.” [refer to page 110 of the study for a detailed table that breaks down these four word types in much greater detail]

The passage above is referring to a study conducted by Mark Davies, a professor of linguistics and author of numerous publications including, “Making Google Books n-grams useful for a wide range of research on language change" and "A Frequency Dictionary of Spanish: Core Vocabulary for Learners, among several others.

Sometime after reading this article, I decided to conduct my own study with various subsets of corpora and one of the things I'd like to add to this study is a comparison of parts of speech. It may sound simple enough, but, there's actually a number of factors to consider that aren't so easy to iron out. I'm willing to live with the fact that the survey I am in the process of conducting may not be a perfect scientific study using rigid and rigorous methods of determining how and which parts of speech should be assigned to each word across my categories of corpora. However, one aspect of my study that I am grappling with how to account for relates to gerunds. Though my earliest understanding of a gerund was its use as a noun, often times they are used in a verbal phrase. In a sentence such as

"The teaching of a language doesn't have to be boring,"

it is easy to label the word "teaching" as a noun; "boring," on the other hand, is being used as an adjective.

Or what about a sentence like this:

"The police are asking for help in identifying the perpetrators."

The word "identifying" clearly seems to be something that could be considered a noun, but the word "asking" is clearly part of a verbal phrase.

My dilemma is this:

When looking at a single word such as "teaching," "boring," "asking," et cetera, you have no way of knowing which role such a word is fulfilling. This is true for many words, so my solution around this is to simply assign the part of speech associated with the most common meaning for any particular word. The size of my study is large enough that this should work. And if I had no interest in comparing it to other studies, no problem would exist, but I do want to compare it to other studies and not necessarily the one I quote above. So, my question is:

In studies that analyze the parts of speech in a set of corpora, are there standard linguistic practices that are adhered to in the assignation process, or does it really just depend on who is conducting the study, and therefore a careful reading of the researcher's methodology is required in order to know how such categories were assigned?

To put this more precisely, I'll use an excerpt from the quote I included at the beginning:

This 90% figure will be obtained by knowing about 2600 nouns, 230 verbs, 980 adjectives, and 50 adverbs, or a total of about 3800 total forms.”

Do these 2600 nouns include gerunds? I've actually searched the study for any mention of how they are treated, but it does not appear to address it.

My initial thought was to eliminate them, but not even halfway through the collection phase I see that they are quite prevalent. I've already made some judgment calls on certain words to exclude, but to omit gerunds simply because I'm not sure how to classify them seems like a step too far.

Some of you might suggest I contact Professor Davies directly, and I may just do that, but in the meantime, I wanted to know if anybody in this community is aware of standard practices with regard to this topic.

Thanks in advance for any insight or references you can provide.

  • 2
    Something to remember: not every linguist has gerunds in their theory of syntax. – Alex B. Oct 22 '16 at 2:36
  • 2
    By the way, "corpora" is the plural of "corpus". You cannot say "a corpora". – fdb Oct 22 '16 at 10:55
  • @fdb You learn something new everyday. Thank you for the correction. I will correct that mistake now. – LISA Oct 25 '16 at 19:09
  • 1
    I am not sure of how many Castillian words you need to know, but I am pretty sure that without some notion of Castillian grammar you won't be able to understand Castillian, nevermind how many lexical entries you memorise. (continued) – Luís Henrique Oct 26 '16 at 11:54
  • (continuing) In the case in point, English gerunds are syntactically different from Castillian gerunds; they perform tasks that Castillian gerunds do not. For instance, in your example "The learning", the English gerund stands for a substantive nominalisation of the verb - 'the act of learning'. Castillian gerunds cannot do this (the infinitive is used instead, "El estudiar de la lengua" - literally, 'the to study of the language'). If you do not know this syntactical feature, you won't be able to use "estudiando" properly. – Luís Henrique Oct 26 '16 at 11:54
3

Present and past participles can function like nouns, adjectives or verbs. There are clearly some examples where the non-lemma form has taken on a life of its own. There are some which have currency and their own dictionary entry but are still essentially semantically logically derived from their theoretical lemma. And there are those which do not have much currency but can be derived productively.

The Universal Dependencies VERB doc considers it arbitrary:

Note that participles are word forms that may share properties and usage of adjectives and verbs. Depending on language and context, they may be classified as either VERB or ADJ.

They forgot NOUN.

See also:

Stolen, part of speech

English words which are both verbs and adjectives

Why do "determining" and "determined" have different lemmas?

| improve this answer | |
  • Thank you for your answer. I look forward to exploring the links you've provided! – LISA Sep 6 '19 at 0:04
  • BTW, I've decided to award you with the green check mark. Both your answer and the one from jknappen are very good, but the first link you directed me to was so good and so exactly what I was looking for that I felt immediately compelled to give you the award. Thank you for taking the time to answer this question. – LISA Sep 6 '19 at 0:35
  • 1
    Thanks. UD is the standard for people actually building stuff, so it's made to be readable and actionable, and evolves. – Adam Bittlingmayer Sep 7 '19 at 5:28
2

Generally speaking, gerunds are lemmatised and counted as verbs (and typically, there is just one tag like VING for the ing-form of the verb, no matter how it is used in a sentence) in corpus linguistics. From the examples given, I think many taggers and human annotators will annotate boring as an adjective despite the existence of a verb to bore, and chances are good that teaching in the example will be tagged as a noun. Both asking and identifying will very probably tagged as verbs.

The number of nouns only contains a few "gerunds" that are far enough removed from verbal use.

| improve this answer | |
  • 1
    ADJ is also a possibility. And the tags for boring and teaching will depend on the context. – Adam Bittlingmayer Aug 2 '18 at 8:30
  • @A.M.Bittlingmayer: I am referring to the examples in the original post here. – jk - Reinstate Monica Aug 2 '18 at 16:10
  • Thank you for your answer! Very interesting. But my curiosity is piqued even further now. Who decides what gets tagged as a noun and what gets tagged as a verb? What criteria is employed in determining whether or not a gerund is far enough removed from verbal use? I suppose you could take a random sample of collocations and see which -ing words were used as a noun and which were part of a verbal phrase and then extrapolate from there. But that's just how I would go about it. How do legitimate/credentialed/full-fledged linguists go about making the determination? – LISA Sep 6 '19 at 0:11
  • 1
    @LISA They are writing a tagging handbook and document difficult decisions. In the beginning, the tagging guidelines are just a few pages long, but when doing large corpora and maybe non-standard language varieties, the handbook grows. It is also good practice to have more than one annotator at least for a representative sample of texts and to calculate inter-annotator agreement. – jk - Reinstate Monica Sep 6 '19 at 9:31
  • @jknappen Wow ... a book even. I'm impressed. Here's to the linguists behind that! I can't wait until it's published!! – LISA Sep 19 '19 at 7:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.