          UEBC and Ambiguity
             July 2002

Q. I understand that UEBC takes a very strict
  approach to ambiguity, so that braille is just as
  clear as print in representing the symbols
  basic to reading. But is this really necessary? Can't
  ambiguities such as in our current codes be
  resolved by considering context?
A. Yes, most of the time, even with our present
  ambiguous codes, you can apply common sense
  and figure out what the meaning "must be" because the
  other possible meanings do not make sense in
  context. But that requires that you already know something
  about what you are reading. What happens when you
  are entering into a new subject, as for example
  a young child might often be? Or when there is
  simply insufficient context present?
Q. Can you give some examples?
A. First, consider a brand name appearing in
  current American English Braille as

  Daburstationature

  Knowledge of English words would suggest that the dots
  6, 1345 is a capital n, not
  "ation." But is that a slash or the letter-group
  "st" after the first r? You can't tell unless you
  already happen to know the brand.
  Second, if a reader of today's American
  math code (nemeth code) were to encounter the
  braille text

  = = .,

  he'd probably guess that the first instance of
  dots 46, 13 means the Greek letter kappa
  and the second means an equals sign (and that
  neither means the word "knowledge" in italics)
  --and he'd probably be right, provided the
  context is classical mathematics, since
  only that interpretation yields a sensible
  equation in normal algebraic form. In other
  contexts, however--e.g. if the same sequence
  were presented as a token list in a discussion about
  a computer algorithm--such assumptions would not
  necessarily be valid and could lead to misreading
  or simply confusion.
Q. Couldn't the transcriber anticipate such
  situations, and help the reader with an
  explanatory note, or grade 1
  treatment, or some other device?
A. Even if we imagine that human
  transcribers could be expected to spot all
  cases of potential ambiguity and take
  corrective action, would we want to require
  such help with our basic reading? Shouldn't
  basic symbols be directly accessible, no
  matter what the context? Furthermore, more and more
  frequently, conversion between print and braille is being
  carried out by computers, with or without the added help
  of human transcribers. Computers thereby
  foster independence as well as efficiency--but
  they are very poor at "understanding" context or
  otherwise helping with ambiguity. This is true
  not only for conversion from print to braille, but also from
  braille to print.
Q. Is all this for the sake of computers, then?
A. No. Computers have no importance beyond their
  role as servants of people. By enabling computerized
  braille translation to be more accurate with less
  help from humans, the net effect is to make
  more and better braille more readily available
  to readers. Even transcribers benefit, being
  able to concentrate more on their special skills
  rather than routine translation issues, and
  thereby can produce more braille overall.
Q. Can you give me an example of where a
  translation error due to ambiguity might
  matter?
A. Let's say you braille the brand name
  "Brailleationote" into a note-taking device.
  Notice that the capital n has the same
  form as the contraction for "ation". Now, at
  least in the case of one such device, if you
  used it to translate your braille into a print
  document to be given to a sighted colleague,
  he would find himself reading
  "Brailleationote." Depending on the
  circumstances of the communication--private or
  public, personal or professional, for
  instance--it's not hard to imagine that the error
  might not be inconsequential in at least some of
  them.
Q. Couldn't that be fixed just by updating the note-
  taker's translation tables?
A. Of course, but that would only fix that
  particular instance--not the basic problem, which
  gives rise to an indefinite number of others
  like it. UEBC fixes the basic problem.
Q. How?
A. UEBC starts by making sure that it's always
  clear where symbols begin and end. That is, there
  is never any question as to whether two braille cells
  in sequence are to be read as a single two-
  cell symbol or as two separate symbols
  --something that is not always clear in current braille
  codes. That way, even if you encounter a
  symbol that is new to you, at least you know what
  to look up in a symbol table. Secondly,
  UEBC assigns only one meaning to any one
  such symbol in any one mode, a mode being a
  condition in effect, such as grade 1 or 2.
  Finally, UEBC ensures that any modes, of
  which there are very few, are always clearly
  indicated.
Q. What method is used to clarify the extent
  of symbols?
A. UEBC does this by providing strictly-
  defined "prefix-root" rules for symbol
  formation that both establish a firm foundation for
  practically all current braille symbols and
  also assure that additional symbols as needed,
  both now and in the future, can be accommodated
  in the same system without ambiguity.
Q. What are prefixes and roots?
A. Prefixes are cells containing only right-
  hand dots, plus the number sign. Roots
  are all other dot combinations. The basic idea
  is that a complete symbol is always either a
  root, or one or more prefixes followed by a
  root. (for full details, including a
  mathematical proof that the UEBC method is
  sound, see the March 1995 report of
  UEBC Committee II, which is available
  on the www.iceb.org web site.)
Q. Is this a new idea?
A. Not really; the basic concept is implicit
  in braille as Louis Braille originally designed
  it, not only in the famous "seven line" layout
  but also in the way that the signs were used. For that
  reason, virtually all current braille codes
  exhibit the prefix-root principle to some
  extent, whether consciously or not. The only
  thing that is new is that, in UEBC, the
  principle is strictly applied.
Q. Can you give some examples of how this is
  different from current codes?
A. In today's English codes, a dot 5 could
  be the beginning of a contraction or a less-than
  sign (in literary code and Nemeth
  code respectively), or a symbol in its
  own right (e.g. the baseline indicator in
  Nemeth code, or the double quote mark in
  computer braille code). In Nemeth code,
  dots 46 before dots 13 could be the beginning of a
  two-cell symbol for an equals sign, or
  the beginning of a two-cell symbol for a Greek
  letter kappa, or a symbol in its own right
  indicating that the word "k" is in italics.
Q. Couldn't these ambiguities be fixed
  simply by changing a few symbols?
A. Possibly, if there were not many more
  examples, but there are--because the problem is in a
  fundamental aspect of code design, not
  only in symbol assignment.
Q. Does this mean that braille would be following print
  slavishly?
A. Not a chance. On the contrary, the UEBC
  approach makes braille fully equal to print
  in its ability to express the symbols
  fundamental to reading. A symbol, for
  example a "dollar sign," is after all an
  abstract entity that can be represented in various
  ways in different writing systems--print and
  braille are just two such systems. The
  UEBC philosophy is that braille can and should have
  the same clarity and expressive scope as
  any writing system. With UEBC, we can
  imagine that one day a blind scientist will invent a
  new braille symbol for some concept central
  to his or her new theory--and then users of
  print will have to come up with a corresponding print
  symbol, following braille!

















 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
