This is the second in a series of critiques of the article "Print and Braille: Evolving Codes to Meet the Needs of a Changing World" by Jennifer Dunnam which was just published in the May 2012 Issue of the Braille Monitor and is available online. This second installment of the overall critique focuses on the statements about the Nemeth Uniform Braille System (NUBS) that were made in Dunnam's article
Dunnam's article is a response to information from BANA on the desirability of adopting a unified braille system, such as UEB or NUBS, to address problems with our current braille codes. She offers her perspective on certain problems encountered in our current literary braille system and concludes that UEB is better equipped to address these problems than NUBS would be because NUBS "introduces some possible difficulties into material Brailled for everyday use."
My original approach to this critique was to analyze Dunnam's statements about NUBS and to compare them with related aspects of UEB. However, the more I learned about the details, the more I became confused. For example, why is Dunnam more concerned about the NUBS use of a special braille symbol for the period in a mathematical context than about the UEB requirement to repeat a Numeric Indicator after the colon in a time of day? Why is there no discussion that certain features of NUBS are what make it possible for NUBS to avoid UEB's required elimination of sequencing and of ten contractions from EBAE? However, once I began to appreciate that two assumptions, one about the possibilities for better automated backtranslation tools and one about teaching tactics, have a big impact on the evaluation of the relative importance of the details, I was able to fit the details into a framework. I hope this analysis will be helpful to you in understanding the issues.
The next two paragraphs quote what Dunnam writes about NUBS. This section is followed by seven numbered sections where extracts from these paragraphs are analyzed in detail.
When the work was basically completed on the Nemeth Uniform Braille System two years ago, I eagerly studied the sampler and the rule book, hoping that the Nemeth Uniform Braille System would be the answer we had been seeking—a way to minimize the difficulties with the current literary code while preserving our tried-and-true system for working with math and science material. The basic features of NUBS are also discussed in the BANA article, and I will not repeat all of them here. NUBS is not simply our current literary Braille with lower numbers. It is my view that, although NUBS takes a systematic approach to addressing the problems and would offer some real benefits for working with technical material, it introduces some possible difficulties into material Brailled for everyday use.
As discussed in the BANA article, most Braille codes, including our current ones, use "modes,” in which a given Braille character has different meanings depending on which mode is in use. Mostly the everyday Braille reader need not pay much mind to modes because their application is quite intuitive. NUBS has two modes, narrative for normal literary material, and notational for numeric and technical material. Punctuation is different in the two modes—the comma, period, and colon are completely different, and the other marks of punctuation require an extra indicator in notational mode. The Braille learner must grasp the concept of these two modes right from the very beginning, because notational mode is not used just for math and technical material. It is used anytime a number is present—in a numbered list of spelling words, in references to time of day or money, in the mention of a year. It is also used in cases where no numbers are present—email addresses and middle initials, among others. The Braille user must be mindful of the two modes when writing, or backtranslation errors will creep in. To have two sets of punctuation throughout all texts creates more complexity for everyone. Additionally, NUBS uses a method for indicating accented letters that requires a special symbols page in order to give the reader information about the accents while avoiding a great deal of clutter in the text. The code therefore misses an opportunity to improve the experience of students learning foreign languages as well as providing the general reader better information about the accented letters used in English.
There is only one statement in the above, the one related to accented letters, that might be considered incorrect. However, the significance of all of the statements is difficult to evaluate without digging considerably deeper into NUBS, UEB, and more general braille-related issues than many readers might have the time for. This section thus analyzes Dunnam's statements about NUBS from that perspective.
NUBS has two modes, narrative for normal literary material, and notational for numeric and technical material.
I don't think that what is going on here is simply semantics but I do want to point out that Nemeth gives a special name, "narrative mode," to its default mode whereas a braille system's default mode doesn't always have a special name. For example, UEB doesn't seem to have a special name for its normal or literary mode which is contracted braille. In any case, I would not expect an experienced teacher to use technical terms such as "narrative mode" and "notational mode" when teaching young braille learners. See Section 4 for more discussion of the effect of modes on braille learners.
UEB has at least five named modes in addition to its default mode: grade 1 numeric mode, horizontal line mode, grade 1 mode (which is not the same as either grade 1 numeric mode or grade 1 braille), arrow mode, and shape mode. So my view is that the fact that NUBS has only two modes is a positive, not a negative when you compare NUBS with UEB.
Punctuation is different in the two [NUBS] modes—the comma, period, and colon are completely different, and the other marks of punctuation require an extra indicator in notational mode. ... To have two sets of punctuation throughout all texts creates more complexity for everyone.
This is true as far as it goes—except that NUBS does not use a different symbol for the colon—and I'll discuss it more in the next Section but I'm not sure why it is such a big deal. There was a BANA-supported study that caused concern about the Nemeth Code because of the frequency of cases where a Punctuation Indicator would be required before a period and that may be one reason why NUBS adopted a separate notational period.
EBAE also uses different braille symbols for characters that are the same in print. EBAE uses different symbols for a period and for a decimal point even though they are the same character in print; this does not seem to have caused much of a problem. On the other hand, EBAE's use of two different symbols for an apostrophe and for a right single quotation mark even though they are the same character in print continually causes translation errors in EBAE that have not been addressed in UEB.
The phrasing "two sets of punctuation throughout all texts creates more complexity for everyone" is unrealistically negative. In the first place, as the previous paragraph shows, the use of two different symbols for the same character is not new to braille and does not necessarily cause problems. Moreover, as discussed in Section 4 in more detail, any complexity can be avoided by using different terminology for the characters in the two sets as UEB does for quotation marks. As many are aware, different locales use different sets of characters for quotation marks. For example, some English-speaking countries in ICEB use single quotation marks rather than the double ones for outer quotes. UEB addresses this issue by defining special braille quotation marks called "nonspecific quotation marks" which don't correspond to any print characters at all but, per UEB Rule 7.6.2, are to be used to represent the "predominant quotation marks in the text" unless the specific form has special significance in which case a differently-named set appropriate to the locale is to be used.
Another case where UEB introduces multiple symbols is the case of the hyphen and minus characters. Most electronic source files, most people, and all programming languages I know of use the keyboard character properly called the "hyphen-minus" to represent both a hyphen and a minus sign. UEB uses dots 3-6 for a hyphen and allows either dots 3-6 or dots 5, 3-6 to be used for a minus sign. Unicode, which is the international standard for characters, supplies two additional characters, one called "hyphen" and one called "minus sign", that are permitted to be used to replace the "hyphen-minus" when there is some need to make a distinction whereas UEB only has two different characters, not three.
A related issue is that given that one goal of UEB is to represent print characters accurately, it seems that UEB should follow the electronic source file and not use the two-cell minus sign unless the Unicode minus sign character is actually used in the source file. (This of course creates a problem when representing numerical subtraction linearly since the one-cell minus sign requires a Numeric Indicator after it.) Although the Rule 3.3 in the UEB Technical Guidelines states that "most maths and science texts show the minus sign as slightly longer than a hyphen in print so the dot 5 form is best used throughout" it is not true that the visual appearance of a character, i.e. its glyph, can always be used to determine which character has been used.
Given that Unicode specifies more than one hundred thousand different characters it would seem useful to base any further discussion related to different braille symbols in different modes on the Unicode standards for terminology and for character identification.
... the other marks of punctuation require an extra indicator in notational mode.
It is true that when literary punctuation marks occur in NUBS notational mode (or in Nemeth mathematics) they require an extra indicator to avoid ambiguity. However, the study mentioned in Section 2 found that in the literary documents that were analyzed there would be only about one case in ten pages of braille where an extra indicator would be required because a number was followed by a semi-colon, colon or a question mark.
Of course, in order to make a more fair comparison, the extra indicators with punctuation marks in NUBS should be balanced by the extra numeric indicators in UEB which are required by its use of upper numbers and the use of two-cell rather than one-cell symbols for common print characters including the plus sign. Also, according to UEB Rule 6.3, both a colon and a hyphen (as well as many other symbols) terminate numeric mode so the UEB numeric indicator needs to be repeated after these symbols such as when representing the time of day. Finally, for completeness, it might be necessary to take into account that the UEB numeric mode automatically sets grade 1 mode. What this means is that the UEB equivalent of a letter sign has to be inserted after an embedded number to cancel grade 1 mode if any contractions are to be used after the number.
The Braille learner must grasp the concept of these two modes right from the very beginning, because notational mode is not used just for math and technical material. It is used anytime a number is present—in a numbered list of spelling words, in references to time of day or money, in the mention of a year.
The statement that the "Braille learner must grasp the concept of these two modes right from the very beginning" gives the impression that something much more complex is occurring than it seems to me is necessarily the case in practice. As I wrote earlier, I would not expect an experienced teacher to use technical terms such as "narrative mode" and "notational mode" when teaching young braille learners.
Of course, no matter what they are called, braille learners would have to somehow learn to use at least two UEB modes right from the beginning as well: the default mode and the "numeric grade 1 mode." The latter mode might be even more difficult to grasp than the NUBS notational mode. NUBS always uses the one-cell dots 3-4-5-6 symbol as a number sign or "notational indicator" before a number and its effect is always terminated by either a space or the specific indicator used for that purpose. UEB, by contrast, uses one of fifteen different two-cell symbols as a numeric grade 1 indicator and its effect is terminated by any symbol not in the following list: the ten digits, the ten numeric space-digit symbols, the period, the comma, the simple numeric fraction line, and the two line continuation indicators. That is, both a colon and a hyphen are among the symbols that terminate the numeric grade 1 indicators and failure to repeat the UEB numeric indicator after a colon in the time of day would result in a braille error.
(UEB has to use two-cell symbols for its numeric grade 1 indicator it uses dots 3-4-5-6 in the construction of unrelated indicators. Per UEB Rule 6. the numeric grade 1 indicators are constructed from dots 3-4-5-6 followed by one of "the ten digits and the two symbols which are used as decimal signs," by a space, or by the second cell of the numeric passage indicator or or numeric passage terminator.)
It is correct that, as Dunnam puts it, the NUBS "notational mode is not used just for math and technical material." For example, NUBS uses dots 1-2-4-5-6, not dots 2-5-6, to terminate a number such as when numbers are used in a numbered list of spelling words. But it seems to me the only way this would cause difficulty for a student would be if the teacher made a big deal of it and told braille learners they had to remember two different modes and remember which period to use in which mode. This would not be good pedagogy. One easy solution would be to tell braille learners that the NUBS dots 1-2-4-5-6 number terminator is a "number stop" or something similar. It's certainly not a difficult cell to recognize in that context. I can't think of any reason why one would need to explain NUBS modes and refer to a number terminator as a "notational mode period" when teaching NUBS to braille learners.
In summary, it seems that the complexity of the rules for the UEB numeric grade 1 mode could easily nullify any concerns about simpler NUBS notational (numeric) mode.
[NUBS Notational mode] is also used in cases where no numbers are present—email addresses and middle initials, among others.
There is no way to avoid the same braille cell's having different interpretations in different contexts, especially in contracted braille. NUBS has two ways to address this problem. In most cases NUBS places its dots 5-6 or dots 3-4-5-6 notational mode indicator before a word where one or more symbols has a different interpretation from what it would have in contracted braille. In a few cases NUBS uses special symbols as explained in the previous section. This seems to me to be an amazingly simple solution to a difficult problem. These tactics are the main reason that NUBS does not require changes to the EBAE rules for contracted braille.
UEB also has several ways to address the same problem as can be seen from this extract from the UEB Rulebook.
2.4.2 Many braille signs have more than one meaning. ... 2.4.3 The reader determines the meaning of a braille sign in several ways:
2.4.4 Use an indicator to establish the mode which determines the meaning of the braille signs which follow. ... [See five modes listed in Section 1.]
- by its spacing (e.g. the vertical solid line segment)
- by applying the Standing Alone rule (e.g. alphabetic wordsigns)
- by its position in relation to other signs (e.g. opening nonspecific quotation mark, line indicator, final-letter groupsigns)
- by the mode in effect (e.g. digits, arrow indicator)
Dunnam mentions email addresses which are a special case where the embedded period is usually called a "dot." Here both NUBS and UEB can get into trouble because if you want to enter a "dot" directly from a braille display without going through backtranslation, you typically need to enter dots 4-6. It's probably best to leave any further discussion of computer items to a separate article.
I'm not quite sure what is meant by "middle initials." According to EBAE Rule VI. 27.c. a letter sign is not needed for personal initials whether spaced or unspaced. According to UEB Rule 5.2 spaced initials followed by a period require a leading dots 5-6. UEB's solution to the problem of embedded period characters as occur in unspaced initials is the elimination of the contraction for "dd." However, since UEB does not eliminate the contraction for "dis" it must use its grade 1 mode to represent a leading period such as occurs in a file extension, e.g. .doc. (This latter is a case where the meaning of a cell depends on the UEB mode in one relative position but not in other relative positions.)
The Braille user must be mindful of the two [NUBS] modes when writing, or backtranslation errors will creep in.
Braille is designed to be efficient and pleasant for tactile readers. This is one of the reasons why braille is intentionally compact. Anyone who knows braille knows that sometimes just a single omitted or extra dot let alone a missing braille cell can entirely change the meaning of a piece of text. This is probably one of the reasons why Dunnam emphasizes the importance of avoiding translation errors and states earlier in her article that "Frequent exposure to such errors can undermine the process of learning correct Braille and of learning the material being read in Braille."
The downside of braille's compactness is that the same issue happens in writing as in reading; if braille is not perfect it can easily be misinterpreted. Changing a braille system is as unlikely to reduce the number of errors in directly-entered braille as changing print spelling is as unlikely to reduce the number of typographical errors. As long as automated braille backtranslation relies on perfect braille, there are going to be backtranslation errors.
However, it is not necessary for automated backtranslation tools to rely on perfect braille or even entirely unambiguous braille rules. After all, as anyone familiar with computer languages knows, compilers are able to detect programming errors automatically and are usually able to correctly identify and state the cause of the error. Compilers accomplish this with standard tools known as parsers which can also be used for braille backtranslation. And as anyone familiar with word processors knows, print spelling checkers are able to detect likely spelling errors and to query the user as to whether what they really meant is what the spelling checker has guessed or whether they meant something else. Spelling checkers accomplish this by maintaining a large database of typical errors and by various levels of artificial intelligence. Something similar could solve Dunnam's will.i.am example. An automated process could determine that interpreting dots-256 as "dd" is very likely incorrect since willddiddam is not in a standard list of English words. On the other hand, interpreting dots-256 as a separator is very likely correct since it yields three words that are in the list.
Improving the accuracy of braille backtranslation is clearly a situation where an investment in better software is likely to yield the desired results at less cost than adopting a new braille system.
I understand that Dunnam's point here is her speculation that the design of NUBS would be more likely to cause brailling errors than the design of UEB. However, I believe that research on this issue would be very difficult, especially in the case of early learners where the effect of the teacher is likely to play a significant role. Studies of the effect of a teacher's knowledge of mathematics on sighted students' early understanding of arithmetic has shown that the effect of a student's first grade teacher's knowledge is the dominant factor in understanding and that this effect persists at least until third grade. So improving teacher training could well be another situation where an investment is likely to yield the desired results at less cost than adopting a new braille system.
Additionally, NUBS uses a method for indicating accented letters that requires a special symbols page in order to give the reader information about the accents while avoiding a great deal of clutter in the text. The code therefore misses an opportunity to improve the experience of students learning foreign languages as well as providing the general reader better information about the accented letters used in English.
It is not correct that the NUBS method for indicating accented letters "requires a special symbols page." The use of a special symbols page is not required; it is just the recommended option for leisure reading to avoid "clutter in the text." It is consistent with the design of NUBS that its more informative symbols for accented letters, or what UEB calls modified letters, can also be used directly in the text. This is in contrast to UEB where its design prohibits this option so its multi-cell symbols for accented letters must be used for all occurrences. However, it turns out that there is a different reason why NUBS and also UEB miss "an opportunity to improve the experience of students learning foreign languages."
As for NUBS' missing an opportunity, BANA's Braille Formats provides unique symbols for accented letters which are currently used in textbooks for teaching a number of different foreign languages and I don't know of any reason why NUBS would require this approach to be changed. On the other hand, adopting UEB would not "improve the experience of students learning foreign languages" because the UEB Rulebook does not advocate using its accented letters in foreign languages. Its Rule 4.2.7 states, "Where a significant knowledge of a foreign language is presupposed or is being taught, use signs from the indigenous foreign language braille code." It thus seems possible that if the United States were to adopt UEB, Rule 4.2.7 would require significant changes to our current braille rules for teaching foreign languages.
Although I wrote about this in another article, I'll repeat the idea here for convenience. UEB specifies twelve two-cell modifiers for representing twelve different marks. The correct one of these modifiers must be placed before any letter with that mark. While it is true that these modifiers supply more information than the EBAE dot 4 used for all marks of any type, these modifiers only provide "the general reader better information" if the reader knows what they mean. For example, the modifier for "circumflex above" is dots 4-5, 1-4-6 and for "cedilla below" is dots 4-5, 1-2-3-4-6. It seems it could be difficult to remember what all of these modifiers mean but again that's just my opinion.
I recommend that everyone take a closer look at supposed NUBS issues in conjunction with related issues in UEB. It seems difficult to discover any cases where NUBS "introduces some possible difficulties into material Brailled for everyday use" that aren't nullified by some corresponding difficulties with UEB.
May 22, 2012 Draft.