Short of a keyboard like the monster ones that used to be used in Chinese typesetting, what do you suggest as an alternative?
Technically, the industrial design of input methods are derived from the Remington TYPE-WRITER Model 1 of 1874 and the International Telegraph Alphabet No 1 of 1874.
Wheatstone, Hughes and Baudot had the idea of an input method derived from the piano keyboard which increases the number of keys by increasing the length of the input method.
Sholes had the idea of adapting the Morse key by placing the slender and space saving keys in tiers one above the other - more like an organ than like a piano, if you will.
Samuel Clemens bought a Remington in Boston for USD 125. He wrote his brother Orion about the machine, see the Clemens correspondence published by the University of California.
The Library of the British Museum considered the Remington for its card catalogue, but in 1875 decided otherwise as the machine had neither lower case Latin (: English), loose diacritics for Latin (: French, German, Spanish), ligated letters for Latin (: French, German, Nordic), nor Greek.
The first lower case national coded character sets were published for English (ANSI X3.4) and German (DIN 60 003) in 1963. The first upper case and lower case international standard character set without Latin diacritics and Latin ligations was published in 1973 (ISO 646). The first international standard with loose diacritics for Latin with serial spelling was published in 1983 (ISO 6937 and ISO 5426)) and the first international standard for Latin with single spelling was published in 1987 (ISO 8859-1).
ANSI Z39.47 for computerised cataloguing was based on loose diacritics for impact printers used in card cataloguing in 1968. The Xerox Coded Character Set for Xerox Interpress and the Xerox 100ppm laser printers was based on loose diacritics for non-impact printers to limit the RAM requirements in imaging (PostScript version 23 in the original Apple LaserWriter did something of the same). ISO 6937 and ISO 5426 were similarly designed for impact printers - nobody had any concept of the everyday enduser interacting with the character string after printing.
The business with key modifiers in the input method is the same as the business with productive/generative character coding models for impact printers and low memory non-impact printers. The reality is that the information economy of the European Union is in part running an a page description model - PostScript and its derivative PDF - introduced in 1985 and deliberated without a foundation on a coded character set in the first place. Getting Adobe and Heidelberg to whisper in public that Type 1 and PS to PDF is dead is well neigh impossible.
Getting a public discussion of the fact that the industrial design of the personal computer that was so successful for PostScript paper publishing is still more of a struggle -:).
/hh