Ask Your Question

Revision history [back]

click to hide/show revision 1
initial version

Hi Diana

This is a HUGE topic.

I think you have to keep at least three dimensions in mind. Let's use an example so we have something to talk about: "John loves Mary".

First, what is the frequency you are measuring? You can look at the frequency of words, phoneme sequences, stress patterns, part of speech sequences, syntactic phrases, syllable types. ANY linguistic unit you can imagine.

Second, what are you computing frequencies over? Some corpus? A dictionary? How are you tokenizing/lemmatizing? For example, does "loves" get counted separately from "loved"?

Finally, what is your frequency affect meant to explain? In the psycholinguistic literature, frequency effects have been found for production, comprehension, lexical access, etc. In other areas, frequency effects show up in historical change, all over the place in sociolinguistics, phonology, phonetics, etc.

I hope this helps. The take-home point is that there is lots and lots of work in this general area and you probably want to focus in more than on "frequency effects in language" per se.

Mike Hammond

Hi, Diana,

There are, as Dr. Hammond pointed out, lots of things in language that could be considered to be or to be caused by "frequency effects". Likewise, there are many ways of considering elements to be "frequent". Word frequency is usually measured by lemma (ie, roughly, dictionary headword, that is, a conglomerate of all morphological forms of a given "word" [and read up on the many varied definitions of the term "word"]), although there are reasons for assuming that different morphological forms of the "same" word might have different frequencies of use and therefore even be pronounced differently, say (case in point in English: first-syllable vowel reduced in 'mistake' [verb] but nonreduced in 'mistook' [past tense of the same verb]. Frequency figures for different forms of the same words are not easy to come by, for example, in the Thorndike & Lorge counts from the 30's and 40's, so if you get interested in that aspect of things, I would recommend using the Carroll et al. (1970 or 71) Word frequency book, which is both more recent and lists separately all the forms encountered in a 5M-word corpus.

I have pretty much agreed with your understanding of the term "frequency" as principally applying to word frequency; you should be aware, though, that several studies might be taken to suggest that word frequency, and especially for words like homonyms, etc., might not be the whole story (time/thyme are not pronounced quite the same; see mistake/mistook above, etc.). Grammaticalization studies likewise might be taken as suggesting that even frequent words may have different pronunciations depending on their subpart of speech (eg, well as an adverb or as an interjection/marker).

Many of these themes (and I have here concentrated on word form pronunciation, because it's what I know best) have only recently come to the fore in linguistics (eg, markers and interjections have for some centuries not really been taken seriously in linguistics, until the last few decades).

If you go into frequency, you should also be aware that psychologists tend to use a rather different scale for what they consider 'frequent' than linguists might be prone to do. For example, I would defend a frequency of over about 5 occurrences per million words of running, general text as 'frequent' (eg, astronomy [4/M, according to T&L] compared to gastronomy [1/M], and note their pronunciation), whereas my strong impression is that works in psychology would consider words among the less-frequent in that group as infrequent.

Likewise, there are a number of independent effects that can skew the familiarity of words, making some infrequent words (eg, berserk) behave as if they were rather frequent, because they are 'striking' (in meaning and pronunciation).

If you have any further questions, or if any of this piques your interest, let me know and I can give you some further references.

Good luck.


James L. Fidelholtz Graduate Program in Language Sciences Instituto de Ciencias Sociales y Humanidades Benem'erita Universidad Aut'onoma de Puebla, M'EXICO

You're writing to us from an address in Germany. The panelists are all responding using English examples. We have no idea what language your psycholinguistic experiment's materials will use.

We have only guesses at what your term "non-literal" language means (perhaps "unwritten" or perhaps "metaphoric" or "polysemous" or perhaps involving "nonsense words" - any of which could be reasonable experiments to do).

So as you can see we're here to answer questions that are answerable, but this looks more like a conversation that needs to happen. And that's why I'm 100% with Dr. Fischer's suggestion. Talk to your academic supervisor, whoever is suggesting that you need to relate your findings to frequency effects. As our other colleagues have suggested, there are several potential kinds of frequency to consider, not all of which will apply in the case of your specific study.

Much success to you!

Nancy J. Frishberg

These are questions you should be asking of whoever assigned you the topic. And why would someone assign you a topic for a dissertation? Don't you get to choose your own topic? If you're stuck with the topic, then it is your job to narrow it down to something you can achieve in a finite amount of time. I would suggest that you start reading the literature (it appears you already have read some) and then chose a sub-are that appeals to you. If you've been given a very general topic, then you have the power to construe it as you like; but heck with your advisor.

Susan D Fischer