Tuesday, October 13, 2009

difference between English and Chinese dyslexia


In contrast, the new findings show that developmental dyslexia in Chinese is really two disorders: a visuospatial deficit and a phonological disorder combined.

Siok and her colleague Li Hai Tan say the difference can be traced to the characteristics of the two languages. "In English, the alphabetic letters that form visual words are pronounceable, so access to the pronunciation of English words is made possible by using letter-to-sound conversion rules," Siok said. "Written Chinese maps graphic forms—i.e., characters—onto meanings; Chinese characters possess a number of intricate strokes packed into a square configuration, and their pronunciations must be memorized by rote. This characteristic suggests that a fine-grained visuospatial analysis must be performed by the visual system in order to activate the characters' phonological and semantic information.

With respect to HIOXian:

HIOXian lacks "intricate strokes" and "fine-grained visuospatial" detail.
Any particular phoneme/phonete symbol is roughly comparable to the visual complexity of the Roman Alphabet.

I suppose the proposed scheme to colour-code stacked consonant clusters and vowels could encounter learning disabilities. However, that system is strictly optional.
Hmm, guess I hafta pick the colours for that with colour blindness in mind.
Enough males have it that it needs to be treated as the same seriousness as dyslexia.

To reiterate, HIOXian lacks any visual component for meaning smaller than a dash: -.
This means it can be read by ageing folk with visual problems, it can scan and copy easily without too much degradation, it is forgiving of errors.

I was looking at Omniglot.com. Visual English by Bell has many very fine swirls and flourishes that convey essential meaning. They would not display well on a computer monitor. They also are not easy to write quickly and neatly.
From the point of view of a typesetter, Visual English is handy, since like the Cree syllabary, the symbols are recycled by rotating by increments of 90 degrees.
This is not a concern of mine. I am concerned with computer typing on one hand, and cursive writing on the other.
Visual English also illustrates the struggle between incorporating elements into the core symbols versus using various after-the-fact diacritics, accents and even specialized additional characters. IPA also struggled with this design issue.

I'd like to point out that even without the up/down spacing of consonants/vowels, the symbols would still be clear to a literate end user. Don't quote me just yet, but I don't think any vowel symbol will correspond with a consonant symbol. In fact, I mean to plan ahead enough that there is no such confusion.
This makes the up/down spacing simply an additional redundant way to convey information.
For English, it would serve to faciliate learning HIOXian as well as to increase reading speed.
English has enough phonemes that they will not map onto HIOXian without use of accents to convey additional information. This should not be an issue for Decimese.
To avoid the need for diacritics for Decimese, I will need to incorporate such features as plosives, fricatives, glides and semi-liquids into the core character.
I think I could live with LRWY and H requiring diacritics.

Wednesday, October 7, 2009

thoughts on Hioxian

I have a coupla weeks off here, so have revisited the Hioxian letter system.

I had hoped to avoid use of diacritics as much as possible.
But the consonants likely require it.
Not for Decimese, with its very limited selection of phonemes.
But for English and others, yes.

I have a Mac right now and don't know how to save pics.

I have been playing with cursive writing with the Hioxian system.
I'd like to take my notes in it next fall, when I return to university for my thesis.

First of all, English requires that I flesh out the consonant diacritic system for the 40+ sounds in English, subject to particular dialect.
The vowel is the core of any syllable.
The diacritic for English vowels ought to show syllable stress.
This includes a combination of volume and duration in English.
However, it does not necessarily in other languages.

Plus for VERSE English, I wanted to also denote tone pitch. For Chinese too.

So we have an overlap of standards.
The colour-code system could work for a computer.
I.e. RGB - red green blue. If denoting both, say, duration and pitch on a vowel, perhaps it would include overlapped red and green on a bar segment. So it'd be a third colour derived from those two. Alternatively, a blue-yellow-red basis.

In computer font, the spacing is perfectly regular, like courier font.
This is important to indicate edge boundaries for characters.
I had pondered a more abstract system that always possesses the central lower 'trunk' or root bar segment, and assumed a front head facing for meaning.
But it resulted in some very arbitrary left/right meaning segments.
And so it ceased to be particularly intuitive.

It is generally an easy task to determine which way to read the hieroglyphs even if the meaning is not understood. Hieroglyphs with a definite front and back (for example, a human or animal) will generally face the beginning of the sentence, as well as being oriented in the same direction as any large human or divine figure in the associated art work. (However, in some instances, they will be reoriented out of respect to face any important personage, such as a king or deity.)

As an example, if a tableau contains a picture of a man seated and facing right, then all hieroglyphs written in text above or behind the man, and having a definite front and back, would be oriented to the right as well. The actual hieroglyphs would be read from right-to-left because these images almost always face the beginning of the sentence. (Text written in front of the man might very well be oriented to the left, facing the man out of respect.)

In cursive HIOXian, the rounded slant of characters could help show boundaries.
Since the actual shapes of phonemes are fairly limited, seeing boundaries should be fairly clear with this aid.
What I'd REALLY like to do is portray anatomical speech mechanisms in such a way that the character is always whole and continuous, without breaks.
This may be possible, but will take much WORK.

I cannot believe it but geocities is shutting down.
For that reason, I'll just cut 'n paste this website here.


Hi. Many fine attempts at a more phonetic alphabet have been made. For typesetting on obsolete printing presses, Bell's Visible Speech is excellent. It uses Cree syllabary-style of figures rotating 90, 180 and 270 degrees. It shows symmetrical pairs, like voiced/unvoiced very well.
Ygyde's alphabet bears a similarity to various short-hand writing schemes. It is fairly intuitive, and is well suited to handwriting.

However, I wished to make a system even more transparent, as well one optimized for computer displays. After a coupla years of fooling around with ideas, I have tentatively settled on Hioxian. The name is a reference to "HIOX", or the 16-segment alphanumeric font of calculator and early display fame.
A computer font requires particular qualities to be optimal. Thin lines and decorative flourishes are to be avoided. Bars should be medium thick for visual clarity. The only angles should be 45 degrees, so the pixels appear smooth. Leaning and curved characters are to be avoided. These requirements made me select the HIOX character, but not angled, and square v.s. rectangular. The remaining space thus freed up is then reserved for an optional diacritic of similar motif. In a pinch, the use of an old calculator display with the decimal indicating vowels as opposed to consonants can work.

Reading speed is increased by methodically using a mechanism that the Roman alphabet uses only sporadically and not at all methodically. Some letters are higher such as klh. Some are lower, such as gjp. Most are centered like aeiou. Wouldn't a system where character height always means something be nice? This would both increase comprehension and reading speed. In my case, vowels occupy the bottom 2/3 square of a rectangular space, and consonants the top 2/3.
Small particles resembling periods and commas are to be avoided, since they are difficult to read in small font, for those with vision problems, and do not scan/copy legibly.

Making a system more intuitive than Visible Speech was not easy. I fooled around with various systems. For the longest time I planned to use the 4 central segments that form a plus-sign shape as abstract phoneme-type indicators. However, I now wish use a stylized pictographic system, an ideographic one since it is highly stylized. Picture the human head from the side, facing left. For consonants, the top/bottom lips and teeth are left and center respectively. The nasal cavity is along the top. Air flow through the mouth would horizontal and bottom. Voice via the voice box would be bottom/right vertical. Vowels and consonants can indicate points of contact and tongue position via various diagonals.
A supplemental system for punctuation and numbers uses a similar motif. The punctuation characters are indicated with the left and right halves of letter and diacritic. The numbers and related concepts are indicated by using the top and bottom 1/3 diacritics sans a middle section. I may manage a 'visual arithmetic' system like Octomatics. A diacritic contains 5 (fairly) vertical and 2 horizontal segments apiece. Marking left to right for 1-5 and right to left for 6-10, we can indicate 1-10 and 1-10 in a single character space. Varying the horizontal segments could denote such concepts as whole number/fraction, square/root (powers), and metric prefixes for large and small.

Interestingly, for my spinoff project from Deafese called Mathese, or Decimese, I could map simple C-V form syllables onto the number system. This requires a limit of 10 possible consonants or vowels in any particular position. Or one of 20 consonant pairs, since only 1 of the pair is appropriate with Decimese's rules. Use of a binary counting system resembling Octomatics could possibly condense data even more.

We are left with a diacritic above vowels and below consonants respectively. I had intended to reserve the vowel diacritic tone in the pitch register system of VERSE. In Decimese, consonant clusters are severely restricted to C _+ LRW or Y + vowel. Recruiting the diacritic to indicate a LRWY consonant cluster leaves space to spare. English allows S + consonant plus LRWY in some cases. I had enough space left for, say, STR-. The word "strengths" is the ultimate test of capturing English consonant clusters. Notably, NG and TH are effectively digraphs. Conversely X and Q both represent compound consonants, in these case KS and KW respectively. A similar system could be used for adjacent vowels and diphthongs, since English lacks a tone system other than intonation in general.

I hope to explore using rainbow-style colour-coding to indicate multiple consonants or vowels stacked on one another for compactedness. For example, red could denote the first consonant. Blue could denote the second, but this could be purple where the 2 figures overlap in certain segments. And so on. In theory, the word "strengths" could be denoted as follows:
1) English: CCCVCCCCC
2) English/phonetically: CCCVCCC
3) using stacking (or my diacritic system): CVC. [=

Obviously, I need to hash out the entire system phoneme by phoneme. I then need to produce a complete downloadable font. I tested some sample Hioxian characters on my friends. They were guessing the sound before I could finish explaining it! <:
If things go well, perhaps I can design a system like IPA, but one that is easily understandable by the majority of normal people who are not linguists!

March '08.
Haha, hard to believe I was ever that ambitious.
Now I'd just be happy to successfully map the Decimese phonemes onto HIOXian!

Aside: I rewatched the Movie "Contact".
I remain intrigued by the potential offered with polarized displays to show more information embedded in otherwise flat 2D letters.
I suppose we could suggest syllable stress in a true 2D display simply by varying font size too.
I like the idea of additional information being optionally invisible in 2D.

My ultimate goal with HIOXian remains to supplant the Roman alphabet.
Aim high!
However, if it only serves as a complement to the Decimese language with its very limited phonemes and syllable choices then I will be content.