[b-hebrew] Karl's lexicon

JAMES CHRISTIAN READ JCR128 at student.anglia.ac.uk
Tue Aug 28 04:37:49 EDT 2007


Hi Karl,

KWR: Strip off all waws from the beginning of words with very few
exceptions (there are so few exceptions that they are easy to
memorize)

JCR: These need to be specified so that the program can 
check for them before making the generic step of 
stripping the waw.

KWR: Strip off but make note of hehs at the beginning of words. Again words
that start with a heh are rare and easy to memorize.

JCR: Ditto.

KWR: Strip off a following yod mem as that almost always is a sign of a
masculine plural. Surprisingly, there are relatively few verbs that
end with a mem that conjugate to a yod mem in the hiphil, and those
are usually easy to recognize.

JCR: Double ditto.

KWR: Were I to come across a sentence where I didn't know a word, the first
thing I would do is to look over the sentence for obvious grammatical
clues—which is the verb, which is a noun, is there the direct object
sign,

JCR: Not a good idea to start including clause grammar 
into the decision process of an automated program. Too 
many opportunities for bugs which cause incorrect 
decisions. What we need is simple generic rules that 
can be hard programmed. Better to concentrate on the 
form of the word and what we can get from that alone.

KWR: what combination of prefixes, suffixes, infixes can I
recognize that tells me what sort of words to look up

JCR: This is something we can work on when those 
affixes form part of the word. But we need to come up 
with generic rules that can be applied taking good care 
to consider exceptions and to test for them before 
applying any generic rules.

KWR: Right now I'm torn between helping students, and the best way to do
so. If I do too much hand holding, will I interfere with basic
language learning? In other words, if they come to depend on me to
supply all the forms, then if they come across a regular form that is
not found in Tanakh, e.g. in the DSS, and therefore I didn't include
it, would they not be lost? If I do no hand holding, will it not
discourage students when they can't find words? What is the best
middle ground? What would you suggest?

JCR: There is no need to *teach* grammar. The human 
brain is hardwired to figure it out for itself. In 
fact, being taught grammar is the most unnatural and 
way of learning grammar that can possibly be conceived.

The natural mechanism for learning grammar works like 
this. The human brain receives as input a series of 
hard examples and automatically looks for patterns so 
that it can make generalisations and therefore save on 
memory usage. Let me give you a hard example to work 
from.

The English brain hears examples of present and past 
verbs like (put,put), (go, went) and (come, came) and 
as there is no pattern it has to learn and store each 
example as hard wired neurocircuits. But when it hears 
examples like (start, started), (want, wanted) and 
(hunt, hunted) it can make the generalisation that 
actions ending in 't' (except for list of known 
exceptions) have the phonetic suffix 'id' in spoken 
English.

I would therefore suggest that we hand-hold for the 
core vocabulary which would enable the student to 
read large amounts of natural Hebrew text to the point 
where the meaning of unknown words can be guessed from 
context (research shows that if a reader knows 9 out of 
10 words in a sentence they have optimal chances of 
guessing the meaning of the 10th word) in the hopes 
that this provides enough irregular and regular 
for them to figure out their own mental grammar by 
natural process of generalisation. To illustrate the 
point here are some examples of made up language:

thrrf  - he went
ghtrf  - he came
sdfrf  - he gave
wsdrf  - he took

By process of generalisation can you figure out which 
affix indicates the 3rd masculine singular in the 
verbs above?

-------------------------------------------------------
-------------------------------------------------------

James Christian Read - BSc Computer Science 
http://www.lamie.org/hebrew       -  thesis1: concept driven machine translation using the Aleppo codex 
http://www.lamie.org/lad-sim.doc  -  thesis2: language acquisition simulation
http://www.lamie.org/ngelta

-------------------------------------------------------
-------------------------------------------------------















































































































































































































































































































































































More information about the b-hebrew mailing list