|
 |
Bangkok - Warsaw | ||
Share with:
| ||
 Language Learning Forum : Language Learning Log |  | |
| doubleUelle Bilingual Tetraglot Groupie United States Joined 4036 days ago 67 posts - 95 votes  Speaks: English*, Russian*, French, Japanese Studies: Spanish, Thai | Message 25 of 69 13 January 2014 at 1:51am | IP Logged | ||
|
Just out of curiosity, Bakunin, what got you interested in Polish?
I'm vaguely interested in it (as a potential future language) because of its similarity to Russian. 1 person has voted this message useful | |||
| Bakunin Diglot Senior Member Switzerland outerkhmer.blogspot. Joined 5131 days ago 531 posts - 1126 votes Speaks: German*, Thai Studies: Khmer | Message 26 of 69 15 January 2014 at 8:46pm | IP Logged | ||
|
@Vos: thanks for the link, looks great! I'll check it out in more detail once I'm at a stage where I can read real books with ease.
@doubleUelle: well, basically because I like learning languages and we're neighbors (I'm German), sharing 1000 years of history. Since I'm currently doing an extensive reading experiment with Polish, I've been thinking about that technique a bit. The only thing I measure in my experiment is the initial coverage ratio (ratio of known words to total) and how it improves upon completion of the book (and the whole series in a second step if I'm reading a series). There are more variables to measure, like potential improvements in reading speed, listening comprehension, oral and written production etc., but I'll keep it simple (and wouldn't even know how to measure some of the other stuff). Research by Paul Nation and others shows that the optimal coverage ratio is 98% and higher. I've been reading a few books at 95%, but that measure has been inching up lately (due to a better selection of books). While I can comfortably read books of the 95% range, I suspect that the vocabulary uptake is less than with books at 98% or higher. Let's say, I've got a text at 99%, then that's on average about 2 unknown words per page. Those 2 words will stand out, and they will be surrounded by a lot of context I understand well, making it easy to infer them. Well, I'd like to digress for a moment… I see something like three types of unknown words: (1) unknown, but sufficiently clear from context, (2) unknown and irrelevant, and (3) unknown, but both relevant and not clear from context. (1) is fine, those words will be acquired after a number of occurrences of that kind; (2) - I don't care… if the word is frequent, it will reappear in a hopefully more meaningful context and I will have another chance; (3) - those are the words that one might want to look up, but I prefer to wait for a few more instances of the same word before I do that, to confirm both relevance and (relative) obscurity. From my experience, (3) is very rare, (1) and (2) are very frequent at the 98%+ level. Learning happens with words of type (1), obviously. Now when I read a text at 95%, I've got maybe 10 unknown words per page. Unfortunately, unknown words often cluster together. This is because if there is one unknown word somewhere, it tends to destroy context in the vicinity of this word, and then other words which might have been clear from context if they appeared as isolated unknown words, are now unclear as well. At the 95% level, many more words end up being of type (3): they are important for the story, but can't be inferred from context. Comprehension suffers, and with it some of the fun. Relatively fewer words are of type (1), but still a lot. However, even those words won't be acquired at the same rate than when reading a text at 98%+, because there are just too many of them and they get buried by the onslaught of those other unknown, but somehow guessable, words that follow in quick succession. The brain just doesn't have time to pay enough attention to all of them, it's overwhelmed. I don't know if this is correct or makes sense at all, but it's something I believe having observed in my own reading. The consequence for me is clear: I need to make an effort to find reading material at my level, at the 98% coverage ratio level. Books at 95% are possible, but they're less enjoyable and it's a less efficient use of my time. Books below 95% are like studying, and that's not what I want when I sit down to read. Edited by Bakunin on 16 January 2014 at 9:49am 5 persons have voted this message useful | |||
| Bakunin Diglot Senior Member Switzerland outerkhmer.blogspot. Joined 5131 days ago 531 posts - 1126 votes Speaks: German*, Thai Studies: Khmer | Message 27 of 69 16 January 2014 at 9:51am | IP Logged | ||
|
I forgot to discuss interest in my previous post. While I still think it's true that texts at the 95% level and below are less enjoyable to read than those at 98%+ - all other things being equal (which they never are) -, being interested in a topic impacts the pleasure of reading even more. If I'm really interested in a topic, I might even fight my way through a text at 90% coverage, and texts at 95% will be fine / enjoyable. By the same token, a text at 98% which is boring and badly written won't be enjoyable at all.
Another thing that came to my mind is impact on reading speed. Let's assume I read at 150 wpm and pause 2 seconds for every unknown word (no sight recognition, re-reading, guessing at the meaning etc.), then reading speed drops to 120 wpm at 95% coverage ratio and to 143 at 99%. Also, at 95% coverage ratio, I then encounter 6 unknown words per minute, that's one every ten second, whereas at 99% coverage ratio, it's only 1.5 unknown words per minute, one every forty seconds. Edited by Bakunin on 16 January 2014 at 9:52am 3 persons have voted this message useful | |||
| Bakunin Diglot Senior Member Switzerland outerkhmer.blogspot. Joined 5131 days ago 531 posts - 1126 votes Speaks: German*, Thai Studies: Khmer | Message 28 of 69 18 January 2014 at 2:33pm | IP Logged | ||
|
French is on the back burner, but still simmering. I've just been listening to a fascinating travel report from North Korea on Voyagecast, so I thought I could at least share the current list of podcasts I'm subscribed to - to justify the 'Paris' in the title of my log. I love listening to French podcasts during long walks or when doing my shopping. If past years are anything to go by, then this list will look quite different 12 months from now.
France Culture: Affaires étrangères France Culture: Concordance des temps France Culture: Place de la toile France Culture: Planète terre France Culture: Les enjeux internationaux RTS: InterCités RFI: Accents d'Europe RFI: Grand reportage RFI: Géopolitique, le débat Voyagecast 2 persons have voted this message useful | |||
| Bakunin Diglot Senior Member Switzerland outerkhmer.blogspot. Joined 5131 days ago 531 posts - 1126 votes Speaks: German*, Thai Studies: Khmer | Message 29 of 69 19 January 2014 at 8:06pm | IP Logged | ||
I've been thinking about extensive reading a lot recently, mostly in the context of my little experiment (more about that in a later post when I have some results). Word acquisition is a complex process, and there are many shades of 'knowing a word'. I've tried to get a bit more clarity on what I measure when I measure the coverage ratio of books I read. Here's a mind map I've produced this afternoon, showing various stages of the acquisition process. I may very well overlook something important; this is not meant to be a developed theory, it should only help me structure my thoughts.
Acquisition usually goes from 'unknown and context sensitive' to 'well known and not context sensitive', passing through a 'known but context sensitive'-stage. This can be quite a long process, and it is rarely linear. If I don't encounter new words again within a certain time, they tend to slip back to the upper right corner. Here's an example for A: The man took the --- and threw it out of the window of the moving car. I don't really know what --- is, but it's clearly an object that fits into a car, can be removed in case it is actually part of the car, can be grasped and can be thrown out. It could also be an animal small enough to be thrown out of a moving car. Here's an example for B: I punch the 4 digit code but instead of opening up the garage door, the keypad light --- rapidly. My guess would be 'blinks'. There are other possibilities, but 'blinks' is a pretty solid guess, I'd say. The dashed line separates known from unknown for the purpose of my extensive reading experiment. I count words that I have a plausible guess for as known, and words of which I only (believe to) understand the broad area of meaning as unknown. It's a bit arbitrary, but I have to draw the line somewhere. As is clear from the picture above, I could be reading a text at 100% coverage ratio and still learn and improve a lot. All those words that are only clear in a particular context still have a long journey before them until they arrive in the lower left corner. The same words in a different text, without sufficient context, might be on the other side of the dashed line. 100% coverage ratio doesn't mean I understand 100% of the words in isolation, it only means that I understand (or have a plausible guess for) 100% of the words in this particular text. 100% coverage ratio means that I can fully enjoy the text, but it doesn't mean at all that I'm not learning anything, usually quite to the contrary. Edited by Bakunin on 19 January 2014 at 10:41pm 5 persons have voted this message useful | |||
| Bakunin Diglot Senior Member Switzerland outerkhmer.blogspot. Joined 5131 days ago 531 posts - 1126 votes Speaks: German*, Thai Studies: Khmer | Message 30 of 69 22 January 2014 at 10:32pm | IP Logged | ||
I've just finished a short-term extensive reading experiment, using a series of books for children starring a pony and his friends, translated from Swedish into Polish. The stories, 17 in total, were actually quite cute and entertaining, and the language was very repetitive, especially towards the end. I read an estimated total of 160'000 words which equals about 2 normal books, over a time period of 12 days. For every book, I counted unknown words within a 500 word section starting on page 20. I counted unknown words while reading (blue), after I'd finished the book (orange), and after I'd finished the series (brown). Here's how the coverage ratio developed over time:
It's striking how quickly the coverage ratio improved from around 95% to reliably above 98% towards the end. This is well in line with my perception: reading became substantially easier and faster after the first 5-7 books. I was also surprised at how many words I would understand after having finished the current book (orange dots and line). Gains of that kind were substantially larger in the beginning than towards the end; this corresponds to words moving from bubble A to bubble B in the chart in my previous post; unfortunately, I can't measure how words move down from bubble B to context-independent understanding which is where most of the learning is taking place. Why there is a clear upwards trend in the coverage ratio after having finished the series (brown dots and line) is unclear to me; maybe the vocabulary was actually objectively more difficult in the first few books than in the last few ones. The experiment was pretty controlled in the sense that I strictly didn't look anything up, with the exception of animal names (which I never counted as unknown anyway), and I didn't do any other studies on the side apart from listening to podcasts. The vocabulary in those podcasts, however, was very different (politics, economics) to what I encountered in the stories, so I don't think the listening had any notable impact on vocabulary acquisition through reading. It is maybe a bit silly :) but it was actually good fun to watch the stats develop over time, and I was extremely motivated to read a lot as quickly as possible… which is exactly the mindset I need to work down that intimidating pile of books I've amassed! :) 4 persons have voted this message useful | |||
| pesahson Diglot Senior Member Poland Joined 5729 days ago 448 posts - 840 votes Speaks: Polish*, English Studies: French, Portuguese, Norwegian | Message 31 of 69 24 January 2014 at 5:34pm | IP Logged | ||
|
Thanks for the podcasts list. Interesting shows are hard to find, I get bored too easily and now I have some new stuff to check out. I've listened to an interesting discussion about wine today curtesy of Planet Terre. Edited by pesahson on 24 January 2014 at 5:34pm 1 person has voted this message useful | |||
| Bakunin Diglot Senior Member Switzerland outerkhmer.blogspot. Joined 5131 days ago 531 posts - 1126 votes Speaks: German*, Thai Studies: Khmer | Message 32 of 69 26 January 2014 at 12:40pm | IP Logged | ||
You're welcome. I'm not sure if you've seen this thread, there are more recommendations for interesting French podcasts. Thai: I've graduated from ป.2 and have started working on 'Society, religion and culture' ป.3. As could be expected, ป.3 recycles previously introduced topics, upping them a notch or two in terms of complexity and depth, but also adds a few new ones, e.g., cultural diversity, taxes and competition, managing resources. I've also been working through some of the news articles on selfstudythai with the main intention to add them to my corpus of curated Thai texts. I've privately committed to one article per day and will log my progress in Serpent's consistency thread. Let's see how it goes. Polish: Still reading like crazy. It's getting easier and easier, but there's oh-so-much to learn. 1 person has 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. |
