I've never been particularly good at learning languages, but I try. Over the years I've tackled a few in school, and quite a few on my own. In the process I've looked into formal linguistics and read a few texts in semantics and general sematics.
In addition, I have had brief exposure to Natural Language Programming (NLP), using grammar-based and statistic-based approaches.
My approach is:
My native tongue. Not that I approve. I consider myself an aboriginal pre-Celtic Briton, whose land was overrun by assorted IndoEuropeans (Romans, Angles, Saxons, Norse, etc). My people were pushed into mountainous Wales, then to North America, then to the West coast of North America.
2017-10-17: Current theory is that Celtic languages arrived in Great Britain via a cultural borrowing from Europe, not a direct "celtic invasion". What was the pre-celtic language? Given that there are genetic clues for continuity along the Atlantic coast, and specifically with Basque peoples, perhaps pre-celtic Welsh spoke something like Basque. If so, they may have found Indo-Eurpoean celtic "British" an improvement worth borrowing -- at least for trade and then for daily use.
Be that as it may, I've read fairly broadly in English. Have spent many hours browsing libraries and bookstores. Tend to read the entire works of an author in a burst (1-2 books a day).
Studied it in highschool. Read Faust and a few other odds and ends. My grandmother had some "blackletter" books which were fascinating but pretty hard to read. With apologies to Germans everywhere, neither the language nor the culture does much for me.
Studied it in college. Followed by graded readers and then collections of pieces from 1700-1800's. I never caught on to the nuances of the multiple french languages. My instructor said she knew 5 languages, and 4 were french.
Used TY . Ended with several contemporary novels and then El Cid. I liked the sound of El Cid.
Used Wheelock's Latin (c. 1963) text and the Using_Latin (c. 1961) texts. Ended with Gallic_Wars , Argicola , and assorted smaller readings. I tried to use the hard "c" sound (Tacitus as tack-i-tus), but I can hear how the soft "c" evolved.
Used Homeric_Greek and Reading_Greek . The long term goal is The Oddessey. It's been on on my bedside table for years.
Used Japanese in 10 Minutes a Day and worked on Japanese for Busy People . Also have several readers.
I stopped when I realized I don't have a subtle bone in my body, and that the Japanese culture might as well be Martian as far as I'm concerned. Learning an entirely different language family and 3 scripts was just too much. We'll have to meet halfway. Like in Hawaii. At a luau.
So I'll just stick with translations, like haiku, koans, and Shonagon's pillow book.
Awesome culture and awesome language. I started backwards -- learning 1000 kangi over a summer from Reading_and_Writing_Chinese (c. 1979) Of course I forgot most of them immediately. Then restarted with texts prepared by mainland China. These were pretty drab. I think I'm looking for a spicier dialog. Something like Suzy Wong does Berlitz. Anyway, reading Li Bo even in translation is a winner.
I can understand why the Chinese people had to revolt against the crushing burden of the Mandarin heritage. But once they've made their own place in the world, they need to go back and rescue some of the earlier culture and writings.
Used Richardson's Esperanto (c. 1988) I like cultures that are willing to use romance to coax readers along another page. Also did the on-line email course. Reading is easy. Getting the idioms right for writing is tough.
For a while I lurked on soc.culture.esperanto, but it is (was) mostly about the language. Kind of like listening to shortwave and hearing about receivers and antennas.
2016-12-20: See Geography-India-Language
2017-10-17: See Geography-Italy-Language
Using TY plus several dictionaries and vocabularies. Just getting started.
Mostly I've become too busy with other things to do languages. Here are some options:
Also known as Computational Linguistics. I've read a few texts, and have done MS CompSci-level exercises, but haven't really dug in. I'm vaguely interested in doing human-assist Machine Translation, mostly for vocabulary. I find it is easy to recognize grammar in languages you have once known, but the vocabulary is easily lost.
I'm just beginning to build OSS tools for multi-lingual text editing, such as mined and scim.
Creator: Harry George