Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

"Accent marks" for English words

5 views
Skip to first unread message

Yao Ziyuan

unread,
Mar 7, 2009, 6:33:21 AM3/7/09
to
An English word's literal form may not carry enough information for a
non-native reader to pronounce it correctly. We can supply accent
marks and/or superscripts/subscripts and/or other visual hints to
augment a word's "pronouncibility". Such additional pronunciation aids
are added when the original text is transmitted through a computer.

Yao Ziyuan

unread,
Mar 7, 2009, 6:44:26 AM3/7/09
to
We don't have to supply "sufficient" phonetic information but just
those most important, if too much visual hint makes reading
inconvenient.

Yao Ziyuan

unread,
Mar 7, 2009, 12:45:56 PM3/7/09
to
In the past I also talked about encoding (partial) semantic
information in a word's literal form. For example, put a x in a circle
and attach this x-circle icon before the word "stop". A seemingly
better approach is to implement such a semantic clue as a visually non-
disruptive/intrusive mark or stroke, such as a mark above, in or
around a vowel letter like accent marks. For example, draw "x" in the
"o" of "stop".

On Mar 7, 7:33 pm, Yao Ziyuan <yaoziy...@gmail.com> wrote:

Yao Ziyuan

unread,
Mar 7, 2009, 1:00:13 PM3/7/09
to
Further speculations include:

(1) Encoding a word's literal and/or (partial) semantic information in
its phonetic form, preferably as acoustically insignificant phonemes
or tones (so that they don't sound intrusive/disruptive against the
original speech);

(2) Encoding a word's literal and/or phonetic information in its
semantic form. But we can argue that a word's semantics itself has no
"form", unless in an artificially defined formalism, where the literal
and/or phonetic information are part of the formal definition of a
word.

Speculation 1 seems only useful when a message is read aloud by a
computer. Even so, it can still be not very useful.

Speculation 2 seems even less useful.

On Mar 7, 7:33 pm, Yao Ziyuan <yaoziy...@gmail.com> wrote:

Yao Ziyuan

unread,
Mar 18, 2009, 9:40:36 AM3/18/09
to
A three-row literal form of a foreign language is indeed useful for
non-native users to learn and retain the phonetic, literal and
semantic information of the associated foreign language element. Take
English for example, we can add marks above the letters to augment
phonetic representation, and semantic hints (or exact meanings in the
user's native language) below the original literal form to facilitate
semantics memorization and recall.

Technical implementation of such three-row (or tri-row) lines on a Web
page can be with JavaScript, HTML tables, Flash, etc.

Design considerations:

* representation of phonetic/semantic information doesn't intervene
with the original literal form (so they are placed above/below rather
than in the way of the original literal form);

* Using semantic hints rather than exact meanings in the semantic row
(the row below the original literal form) can have certain merits
since semantic hints, just like semantic radicals in Chinese
characters, can be relatively few which eases memorization of such
semantic hints as part of the new three-row literal form, and semantic
hints are a middle step toward independent recall of a foreign
language element's semantics.

* The index effect: Foreign language elements assigned with the same
semantic hint or phonetic hint are mentally categorized together in
the user's memory, which facilitates associative memorization in which
memorizing/recalling one member in such a category can subconsciously
reinforce the memory of other members in the same category.

Yao Ziyuan

unread,
Mar 18, 2009, 1:29:39 PM3/18/09
to
On Mar 8, 1:45 am, Yao Ziyuan <yaoziy...@gmail.com> wrote:
> In the past I also talked about encoding (partial) semantic
> information in a word's literal form. For example, put a x in a circle
> and attach this x-circle icon before the word "stop". A seemingly
> better approach is to implement such a semantic clue as a visually non-
> disruptive/intrusive mark or stroke, such as a mark above, in or
> around a vowel letter like accent marks. For example, draw "x" in the
> "o" of "stop".

Drawing 'x' in the 'o' of 'stop' is too fantastic because it can be
hard for the user to find out what is in a letter and what it means.

Yao Ziyuan

unread,
Mar 20, 2009, 3:50:07 PM3/20/09
to
On Mar 8, 2:00 am, Yao Ziyuan <yaoziy...@gmail.com> wrote:
> Further speculations include:
>
> (1) Encoding a word's literal and/or (partial) semantic information in
> its phonetic form, preferably as acoustically insignificant phonemes
> or tones (so that they don't sound intrusive/disruptive against the
> original speech);

This makes sense. Particularly, if we encode such artificial phonetic
differences back to a word's literal form (marks above letters), the
learner will now have two ways to recall a word's literal AND phonetic
forms.

Yao Ziyuan

unread,
Apr 23, 2009, 4:57:04 AM4/23/09
to
A good way to mark the stress position in a word is underlines.

On Mar 7, 7:33 pm, Yao Ziyuan <yaoziy...@gmail.com> wrote:

Yao Ziyuan

unread,
Apr 23, 2009, 11:15:53 AM4/23/09
to
On Apr 23, 4:57 pm, Yao Ziyuan <yaoziy...@gmail.com> wrote:
> A good way to mark the stress position in a word is underlines.

or raise the stress letter(s)' vertical position.

Yao Ziyuan

unread,
Apr 27, 2009, 5:57:13 PM4/27/09
to
On Apr 23, 11:15 pm, Yao Ziyuan <yaoziy...@gmail.com> wrote:
> On Apr 23, 4:57 pm, Yao Ziyuan <yaoziy...@gmail.com> wrote:
>
> > A good way to mark the stress position in a word is underlines.
>
> or raise the stress letter(s)' vertical position.
>

but i don't exclude other possible decorative methods such as a dot
under the stress letter.

0 new messages