X-Sampa / CXS

Voiceless consonants on the left, voiced ones on the right.

Note 1: I am not really sure what epiglottal consonants are pronouncible.

Note 2: P has a variant transcription v\.

Note 3: The symbol for a retroflex click is my invention.

Note 4: If there wasn't an inconsistency about r`, the retroflex columns would be copies of the alveolar columns with a retroflex/rhoticity marker ` appended to each entry.

Unrounded vowels on the left, rounded ones on the right.

Note 1: Instead of 1, the original X-Sampa, i\ is in common use on Conlang.

Note 2: The schwa, @, does not seem to be defined to be rounded or unrounded in IPA as the chart gives the letter in the very center. We need a conlang that distinguishes these two to force IPA to include both variants (you can write e_x and 2_x to indicate the two variants, though).

Note 3: I\ and U\ are not official IPA.

Note 4: X-Sampa was modified for the following vowels:

Mnemonics

Affricates and double articulations can be represented by two symbols joined with a tie bar if necessary. CXS provides two ways of doing so.

Note 1: For l_e, there is a special symbol 5 in CXS.

Note 2: The X-Sampa palatalisation ' is not in use in CXS (collides with stress marker). Instead, only _j is used. (X-Sampa allows both variants.) However, a CXS specific diacritic ; for palatalisation is also in use.

Note 1: The stress markers were modified to be more similar to IPA compared to X-Sampa. Both versions are in use and CXS has two alternatives.

The following are machine-readable versions of the above IPA to CXS mapping. You may download single files or the whole package.

"charmap-cxs.lsp": Table for Lisp

This is the primary source code.

"charmap-cxs-class.lsp": Supplement table for Lisp and the corresponding "cxs.lsp" Module for Lisp

(require 'cxs) (ipa->cxs (vector #x0262 #x0250)) (cxs->ipa "G\\7X")

This Lisp module is used on this site by the documents about my conlangs. Currently, it cannot handle the linking mark ) correctly -- only a few precomposed cases it converts it correctly.

Module for Perl, version 1.07

use CXS; my $cxs= ipa2cxs("\x{0262}\x{0250}"); my $ipa= cxs2ipa('G\7X');

This can be used in scripts like the one that generates this page: G\6 and ɢɤχ. It's the most complete conversion module.

Features
-	A function to conversion to HTML is included: cxs_encode_html().
-	Correct rearrangment of CXS linking mark ) to an infixed diacritic in IPA (and vice versa): tz_0) and t͡z̥.
-	_0 (and optionally =) is placed according to glyph. This is basic, but better than nothing: n_0 n̥ N_0 ŋ̊
-	Translation tables can be edited/customised.
-	Works with older Perls, too (those without pack('U',...)).

Tables for C/C++ and the corresponding header File

This only contains the mere tables, no conversion routines.

Modules for Python (>= 2.2) and the corresponding data table generated from the Lisp data.

I haven't had a chance to test this myself as my Python's too old -- the module is a kind addition to the gallery by Taliesin the Storyteller.

Keyboard Plugin for X

To be able to type CXS in X, I use the SCIM method with this map. Look into the .mim file to read how to install -- it's easy.

These tables are distributed as Free Open Source Software.

Download Package

CXS^HTML is a form of CXS for presentation in HTML pages when for some reason real IPA is not suitable. It is more readable than plain CXS by using markup. This was proposed by Jeffrey Jones on the Conlang Mailing List.

The mapping is simple: all modifiers starting with _ are enclosed in <sup> in HTML, e.g., n_0 will become n⁰.

Furthermore, the ) tiebar can be represented as underlining using the <u> HTML tag, e.g., aI) will become aI.

<style type="text/css"><!--
    u {
        text-decoration: none;
        border-bottom:   1px solid black;
    }
--></style>

• Current Development Version

  • Added a .mim file for SCIM (and maybe other input methods in X).

• Version 1.4pre

  • Improved CXS^HTML rendering
  • CXS^HTML moved to ipa2cxs to handle _ more easily

• Version 1.3 (2007-11-20)

  • Found bug: _0 would be output as _0_0 in some cases.
  • Features: an HTML version of the CXS string (Perl script)

• Version 1.2 (only pre-released)

  • Added option in web-interface to select affricate ligatures.
  • Found bug: I\ and U\ don't translate back from IPA to CXS correctly. Needs fix.
  • More bugs: IPA->CXS does not decompose combined chars like â, so these are not converted correctly. Needs fix.

• Version 1.1

  • Perl module v1.06
    • Still Missing: Module uses pack("U",...) for Perl versions that support it and the manually implemented pack_U(...) for those that don't (e.g. Perl 5.0).
    • A function to conversion to HTML is included: cxs_encode_html().

• Version 1.0

  • Addition: Python module added
  • Feature: Tables for the Python modules are created from the original Lisp source now to keep them up-to-date.
  • Meta Feature: A version number was created for the whole selection
  • Between version 1.00 and 1.05, the interface of cxs2ipa was extended to return the error position in array context. Use ''.cxs2ipa(...) to force scalar context when porting older programs.