|
 |
Experimenting with French word frequency | ||
Tags: Word Frequency | French | ||
Share with:
| ||
 Language Learning Forum : Specific Languages |  | |
| Serpent Octoglot Senior Member Russian Federation serpent-849.livejour Joined 6395 days ago 9753 posts - 15779 votes  4 sounds Speaks: Russian*, English, FinnishC1, Latin, German, Italian, Spanish, Portuguese Studies: Danish, Romanian, Polish, Belarusian, Ukrainian, Croatian, Slovenian, Catalan, Czech, Galician, Dutch, Swedish | Message 25 of 55 05 September 2014 at 2:06pm | IP Logged | ||||||||||||||
If we speak of C1/C1 and job interviews, I agree. But in real life grammatical accuracy is surprisingly useless, I'd say. Apart from certain specific "trouble points", you can really screw up a lot and still be understood if you know the needed vocabulary. (if you use circumlocutions with poor grammar, hell breaks loose, true) I personally work on my accuracy mostly because I don't want my own speech to make me cringe, and because many of my friends are language geeks, and I want to have even more friends among them. Maybe it's time for a new "what impresses you?" thread. I remember how the administrator started one, and it was very common to value a native-like pronunciation most of all. s_allard's pet topic of mastering small talk is certainly a legitimate criterion, but it's just one of many possibilities. Edited by Serpent on 05 September 2014 at 2:11pm 1 person has voted this message useful | |||||||||||||||
| rdearman Senior Member United Kingdom rdearman.orgRegistered users can see my Skype Name Joined 5034 days ago 881 posts - 1812 votes Speaks: English* Studies: Italian, French, Mandarin | Message 26 of 55 05 September 2014 at 3:26pm | IP Logged | ||||||||||||||
I have to agree with this, I think a large vocabulary is more important than grammar if you are just trying to be understood. For example: Me you help find where train station? will get a better result for a beginning English speaker than. Could you please tell me where the ..... can't remember the word.... the .... errrr.... 4 persons have voted this message useful | |||||||||||||||
| s_allard Triglot Senior Member Canada Joined 5228 days ago 2704 posts - 5425 votes Speaks: French*, English, Spanish Studies: Polish | Message 27 of 55 05 September 2014 at 3:45pm | IP Logged | ||||||||||||||
Saying that I mix up fluency and proficiency is fighting words! (Just kidding). When I wrote "articulate and fluent", I used fluent exactly as I've always used it i.e the absence of undue hesitations in the speaking voice. I don't know how many times I have to say that more vocabulary is better than less. Do you need more vocabulary to talk about the weather in a detailed manner than to just talk about the weather? Yes, the question is how much more. But the essential argument here is that in this kind of interview situations the examiner is not counting the vocabulary you are using, The examiner sees the overall ability to manipulate the language, and particularly the ability to put the words in the right order, in the right form and smoothly. Do you talk about the weather by listing all the technical words you know? If the interviewer asks you what the weather is like do you answer "fog, low visibility, light rain, overcast, 16 degrees celsius, 80% humidity" with the idea of impressing the person by your technical knowledge? Or maybe something like: "The weather right now is not great; it's a bit on the foggy side, typical for this time of the year. I haven't been out yet and I'm not really looking forward to it. I personally prefer more sun. What's the weather like where you are?" Which answer carries the day? 1 person has voted this message useful | |||||||||||||||
 emk Diglot Moderator United States Joined 5330 days ago 2615 posts - 8806 votes Speaks: English*, FrenchB2 Studies: Spanish, Ancient Egyptian Personal Language Map | Message 28 of 55 05 September 2014 at 4:19pm | IP Logged | ||||||||||||||
There's another interesting wrinkle here: If you only know a grand total 300 words, then perfect pronunciation and fluid speech can get you into a lot of trouble. Why? Well, if somebody walks into a shop and says, with an atrocious accent: "I... wants?... <pointing> the baguette." …then the person behind the counter will speak slowly and use small words. But if somebody walks in and says the following with perfect intonation and pronunciation: "C'n I please h'v a baguette?" …then the response will be at full native speed. If don't understand words like coûter "cost" and celle-là "that one" (both of which could easily be left off a 300 word list), you probably don't want to be subjected to full native speed responses. When you actually do start producing excellent small talk, many natives will become a lot less patient. Somebody who speaks deliberately and with a heavy accent will generally be treated as a foreigner who's making a massive effort. Somebody who handles small talk perfectly but who gets temporarily stuck discussing the latest movies will occasionally be treated like an idiot. Re-examining Milton's vocabulary size estimates Anyway, all this is a bit of a diversion. :-) What I want to do next is use the Lexique coverage data to look at Milton's vocabulary estimates: 
Milton's metholodogy was a bit odd, because he based it on the XLex tests. These don't actually estimate vocabulary size. Instead, they estimate how many of the most frequent 5,000 words you know. No matter how huge your vocabulary, you can never score 5,001 on these tests. Now let's take another look at coverage chart I generated from Lexique:
Let's break this down by CEFR level, omitting A1. A2. The students in Milton's study have average passive vocabularies of 1,700 to 2,237 words. This seems reasonable: It's enough to cover a wide range of basic conversations (including the responses), and it gives 88% text coverage of native text. There's enough words to have all the common verbs, plenty of nouns, a nice selection of adjectives, and a full set of pronouns and connecting particles. B1. Milton shows vocabularies of 2,194 to 3,338 words. This gets students to roughly 90–95% coverage of books and films. This still feels fairly plausible to me, and it's a nice step up from A2. B2. Here's where I start to disagree with Milton, slightly. He gives 2450 to 4012 words. At the upper end of this range, we have only 93.42% coverage of text according to Lexique, or an average of 23 unknown words per 350-word page. If we compare this to the CEFR self-assessment checklist, we get:
I think it would be challenging to perform at this level with an average of 23 unknown words per 350-word page. For comparison purposes, shortly after I passed the DELF B2 with a solid reading score (20+), I sampled a frequency dictionary and estimated my vocabulary to be very roughly 7,000 words, give or take at least a thousand. C1. And here's where I stop believing Milton's numbers completely. He gives 2,675 to 4,300 words for C1. This is scarcely larger than the 2450 to 4012 words he gives for B2. I think there's actually three things going on here: - Milton's running into the 5,000 word ceiling on XLex. - The Spanish-speaking French students, with their 2,000-to-3,300 word vocabularies, may be getting a Romance-language discount. - There's some pretty bad CEFR inflation going on here. C2. And now we've completely escape any notion of probability. Milton measures a vocabulary size of 3,525 to 4,068 for C2 students. First of all the highest-scoring group of B2 students (Greek learners in Greece) knew a mean 4,012 words, and the highest scoring group of C2 students (English learners in Greece) knew a mean of 4069 words. Either way, we're still look at an average of about 23 unknown words per 350-word page (for French, so this is a fuzzy comparison). Let's compare this dismal performance with the actual C2 standard:
So looking at the Lexique graph and table above, here are my personal guesstimates for passive vocabulary, focusing on the formal written register, and allowing minor weaknesses with very casual speech: A2: About 2,000 words, as per Milton. B1: About 3,000 words or a bit more, slightly higher than Milton. B2: 4,000 to 6,000 words. C1: At least 8,000 words, giving 98.5% coverage of arbitrary text. C2: Roughly 16,000 words or more, giving >99.5% coverage of arbitrary text. Of course, any of these numbers could be off by 25%, given the inherent vagaries of word counting. And in no case do I wish to discourage people: it's easy to pick up a couple thousand words using Assimil and Anki (or your preferred alternatives), and you can creatively finesse extensive learning from there until you cross the 98% threshold. If you can find a way to enjoy semi-extensive reading with 90% vocabulary coverage, you can do most of the "work" while reading trashy fun novels and watching TV. Don't think of all this vocabulary as a horrible slog; think of it as an excuse to goof off shamelessly in a new language. 6 persons have voted this message useful | |||||||||||||||
| s_allard Triglot Senior Member Canada Joined 5228 days ago 2704 posts - 5425 votes Speaks: French*, English, Spanish Studies: Polish | Message 29 of 55 05 September 2014 at 5:17pm | IP Logged | ||||||||||||||
I don't think that I can convince emk that 300 words will take somebody a long way because we approach the whole question of minimum vocabulary size from two radically different positions. His position, and I hope I'm getting it right, is that we have to compare a suggested minimum number of words to a range of sample conversations and see what kind of word coverage we get. So let's say we take a given 300-word vocabulary and run it past a set of transcriptions of 10 typical A2 conversations and we get a 50% word coverage. Well, you can't do anything with that. Or take the first 300 words of a French frequency list and run that past a database of French movie subtitles. Again, you won't be able to do anything with 300 words. Here is my approach. 300 words is a threshold for a learner to start developing meaningful statements. I'm not sure what emk means by "take you long way". I prefer the term threshold. Why 300 words and which 300 words? 300 is an arbitrary figure of course that teachers of French use, but I personally base this figure on observations of real conversations and not by looking at large frequency lists. In essence, I'm looking from the ground up. I've given two examples of real conversations where the people use far less than 300 different words. In one conversation there are 224 different words. If we run those words against some A2 conversation dataset, what kind of coverage would we get? I have no idea. Let's pick an arbitrary figure, 60%. We conclude that this level of coverage is insufficient for good understanding of A2 level French. Does this make sense? Statistically, yes. If you do the same comparison with a dataset of C2-level French, you would probably get similar results because this particular conversation contains a fair number of quite specific terms. Do we conclude that these people are not speaking C2-level French and could not pass a C2 spoken French test? To be more specific, let's look at a key component of speaking proficiency: verb usage. The two conversations I looked at used 33 and 34 verbs respectively, of which only 8 were shared. Considering that it seems you need around 1,000 verbs to get 97% word coverage of all the verbs used in a dataset of French movie titles, what can we conclude about these native speakers of French? Maybe they don't really understand each other? Are they A1- level speakers? But we do see two French speaker having a great conversation with just 33 or 34 verbs. And if we combine the two conversations, 59 verbs will do. Now the big question: how many verbs do you need to participate in a conversation in French? You have a choice. If you want to speak like all the actors in French movies, you need 1,000 verbs. I agree. If you want to have a two-minute conversation with native French speaker at a given moment, I say you need around 34. You choose. Which 34? Frankly, outside the 10 most common verbs in French, the actual verbs used will depend on the subject of the conversation. The verbs used will vary from conversation to conversation. If we look at pronouns, we see that in the two conversations, only 10 pronouns were used in both conversations. French has more than 10 pronouns of course. For adverbs and adjectives, we see a similar story. Small numbers and some shared elements. Nouns are where there are the biggest differences. This is not surprising because this where most of the subject content is expressed. How does this all come together in the idea of a 300-word threshold? First of all, we look at the breakdown by category in real conversations, Then we throw in some numbers for padding. This is why I suggested 80 verbs. Considering that 33 and 34 were enough for two conversations, 80 should give us room to play with. It's the same reasoning with all the various categories. The key point here is that there cannot be a standard 300-word list for everybody. Couldn't you just take the first 300 most frequent words from a French frequency list? Why not just take the 80 most common French verbs? Why not use the 150 most common nouns from French movies? That won't work because they may not be right for you because of your specific interests. The big question is which words. This is where the learner has to do some leg work and look at their interests and their situation. If you are retired and studying French to go to spend time in France, your vocabulary needs will be different from those of a person with young children who has just moved to Montreal. But in both cases, I maintain that somewhere in the 300-word region learners will have enough material to work with and start speaking the language for real. 1 person has voted this message useful | |||||||||||||||
| s_allard Triglot Senior Member Canada Joined 5228 days ago 2704 posts - 5425 votes Speaks: French*, English, Spanish Studies: Polish | Message 30 of 55 05 September 2014 at 6:13pm | IP Logged | ||||||||||||||
|
I have always been intrigued by James Milton's article on vocabulary size for C2 candidates because of the seemingly
low numbers for receptive vocabulary on the French tests. Milton is a well-known researcher in the field of vocabulary size and presumably knows what he is doing. I should point the absolute dearth of vocabulary sizes studies of actual C2 test results. We just do not know what productive vocabulary candidates use, and at best we can estimate what the receptive vocabulary is. I feel that in the CEFR world, vocabulary size is not on the radar. There are absolutely no guidelines or numbers related to recommended vocabulary size. I take no position on recommended receptive vocabulary size for the various CEFR levels because I believe it is hard to measure this kind of vocabulary. Furthermore, my own interests are more in productive vocabulary. 1 person has voted this message useful | |||||||||||||||
| Serpent Octoglot Senior Member Russian Federation serpent-849.livejour Joined 6395 days ago 9753 posts - 15779 votes 4 sounds Speaks: Russian*, English, FinnishC1, Latin, German, Italian, Spanish, Portuguese Studies: Danish, Romanian, Polish, Belarusian, Ukrainian, Croatian, Slovenian, Catalan, Czech, Galician, Dutch, Swedish | Message 31 of 55 05 September 2014 at 6:28pm | IP Logged | ||||||||||||||
|
Sure, everyone has different interests. Just please acknowledge that those 300 active words should be accompanied by many more known passively. Maybe most Canadians pick them up at school and in general due to living in a bilingual country. But by default it's not a given, so it sounds like you claim 300 words are enough for the passive vocabulary too.
1 person has voted this message useful | |||||||||||||||
| emk Diglot Moderator United States Joined 5330 days ago 2615 posts - 8806 votes Speaks: English*, FrenchB2 Studies: Spanish, Ancient Egyptian Personal Language Map | Message 32 of 55 05 September 2014 at 8:28pm | IP Logged | ||||||||||||||
This is basically correct, with a few caveats: 1. The word list doesn't have to be based on frequency dictionaries. It can be carefully tuned for A2 tasks, in much same way Basic English was tuned.[1] 2. Nobody expects A2 students to perform B1 or C1 tasks. They need to make polite small talk, buy basic living supplies, order dinner, find the bathroom, give and receive directions, and successfully use subways and trains. But within this limited problem domain, they should satisfy all the A2 criteria. 3. Students may use pantomime and ask for new vocabulary words. Personally, I think you could carefully construct a 1,000-word vocabulary that could be creatively used to handle most of these situations. But most students will need a total vocabulary of at least 1,500–2,000 words if they want to go out in public and function at an A2 level in most situations. Not all of this vocabulary would need to be active. I base these numbers on Milton's survey and on my own experience. [1] Basic English assumes 850 general words, and 150 extra words to cover individual interests. And Basic English can tricky even for native speakers. Vernor Vinge, a writer and math professor, spoke about a science fiction story he wrote: "The word-hacker in me was also intrigued by the Basic English vocabulary the aliens used. (It turned out to be surprisingly difficult to write in that vocabulary. Once I saw the Gettysburg address redone in Basic English; it seemed about as eloquent as the original. I didn't realize until I was writing this story was a feat that was.)" You can find a version of the Gettysburg address in Basic English online. Note the use of phrasal verbs and idiomatic compounds. Many of these turns of phrase would be unknown to actual beginners.
This is the basic limit of using a small conversational corpus: each new conversation you add requires a fair bit of new vocabulary. Here's what I'd argue: What you can do with only 300 words - Have a controlled conversation on a known topic, either with a tutor or in class. - Switch to a new topic by introducing a few dozen words. - Learn (and even internalize) a large amount of French grammar. - Make brief small talk with bilingual neighbors. What you can't do with only 300 words (or so I claim) - Perform the full variety of survival tasks which are normally expected of an A2 student. - Pass a well-designed A2 exam (without the gross academic dishonesty of being given the test in advance).
I'm convinced that there's something awfully suspicious about Milton's paper. If you take his numbers seriously, it means that you can be C2 with less than 95% text coverage, or about 20 unknown words per page in the typical paperback. This is incompatible with the kind of nuanced comprehension of complex materials demanded by the C2 criteria. If we go by this standard, my passive skills would be far above the C2 threshold. Either (a) Milton's use of a 5,000 word XLex test distorted his results drastically, or (b) there's a lot of people out there calling themselves C2 (with the enthusiastic encouragement of their teachers) who can't read an ordinary paperback without a dictionary and a whole lot of guessing. Milton's numbers are also inconsistent with those of several other reputable researchers, as seen in this figure by Françoise Kusseling and Wilfried Decoo: 
My proposed numbers are much more in line with those from Schmitt, who used roughly the same methodology: He looked at the C2 level descriptors, figured out how much vocabulary would be needed to reach those rather demanding standards, and then used coverage data to estimate vocabulary size. In my case, I could also calibrate my numbers against a vocabulary size estimate I made around the time I passed a B2 exam. Milton's numbers are based on a 5,000 word XLex exam given to actual students. The students' CEFR levels were determined by teachers who had assigned them to "streams for study at each of the CEFR levels." To me, this seems lack rigor, and encourage CEFR level inflation. A better strategy would be to assess student levels using actual CEFR assessment exams such as the TCF. 3 persons have voted this message useful |
|  | 
Printable version
|
| Copyright 2024 FX Micheloud - All rights reserved No part of this website may be copied by any means without my written authorization. |
