If you want to jump straight to the music, the complete set of these pieces can be downloaded as mp3 files, or streamed at The Listening Room.

New, July 2014: a full series of matching videos, Harmonic Colour Stains, on YouTube.

Graph of notes, No.3 I never cared much for writing out notes, whether physically, using pen and ink on stave paper, or virtually, using music software (2000 Guitar Solos is a project which grapples with this personality defect). I've always preferred the illusion of working with sound as a plastic medium, generating and manipualting sound waves. This method allowed me greater ease in focussing on the overall impact of the musc, instead of getting hung up on moment-by-moment details.

When composing music in which notes are an unavoidable necessity, I prefer to generate or find a body of "raw material" which I can then edit and manipulate into new music. Stained Melodies was the first set of pieces I made using this basic method, by using chance operations on an array of MIDI keyboard works to extract selected pitch material from its original context, and play it back simultaneously with other, similarly alienated material.

Real Characters and False Analogues, a set of twelve pieces for microtonal piano, is a sequel to Stained Melodies. It takes the compositional premise of the earlier work and develops it, using the most elaborate compositional process I've ever bothered with to date.

Tuning the piano...

Taking advantage of MIDI's petty obsession with precision, the computer-controlled virtual piano plays these pieces with an exactness of rhythm and harmony impossible in conventional acoustic instruments. Following on from Stained Melodies' use of just intonation, Real Characters uses a microtonal tuning scheme built around the perfect fifth, with a full complement of minor and major just intonation intervals for the other notes in the conventional equal temperament scale (in equal temperament, each semitone is an exact 100 cents).

Tuning scheme for Real Characters and False Analogues

Each successive piece in Real Characters is based upon a tonic in the circle of fifths (traditionally, C - G - D - A - E - B - F# - C# - Ab - Eb - Bb - F - C). However, in this tuning scheme the circle of fifths would cycle through all 22 pitches, instead of the traditional twelve. I decided twelve pieces was quite enough, at least for now, and so broke off the cycle, leaving it stranded about halfway between its start and end. Because of the differently sized intervals between each note, the types of harmonies you can hear shift around from one piece to the next.

... and then breaking it

Layers of pitch material As with Stained Melodies, pitch material from randomly-selected keyboard works was chosen and overlaid to form the basis of each piece in Real Characters. Unlike Stained Melodies, each piece in this series uses 15 pitch classes, with the most frequently occuring note tuned to the tonic, and the second most to the fifth. The thirteen other notes in the scale are then selected by chance from the remaining twenty pitch classes. Which means any careful harmonic calculations of the overall tuning system are countered by the arbitrary grouping of intervals, in addition to the size of the relative intervals waxing and waning with the new tonic in each pice.

In each piece, the the material is divided into two harmonic groups, one based on the tonic, the other on the fifth. There is a chance that each harmonic group may pause up to four times during a piece, allowing the other group to be heard alone. The timing and durations of these pauses are determined by chance. Any pitch with fewer than 24 notes over the duration of the piece has the piano's sustain pedal applied to it.

This basic mass of notes is extended by allowing up to four recapitulations of material to occur in each piece; i.e. short sections of the music may be copied and randomly inserted into other passages of the same piece.

The long, tedious process...

The underlying metre for the notes in each harmonic group is altered independently, four times in each piece, using a chance assignment of metrical shifts ranging from four-fifths to one-and-a-half times the original beat. The overall tempo is also subject to change at any given time, to anything from 40 to 150 beats per minute.

Distribution of notes, No. 3 To make use of the full range of the piano keyboard, the register of all the notes used are subject to transposition to a different octave of the keyboard. The distribution of notes across the keyboard can range from just two octaves (in any combination), to the entire keyboard. The occurrence and pattern of the redistribution of notes is governed by chance, weighted by a probability that favours the middle range of the piano, creating a sort of bell curve of note densities spread across the full range of the instrument.

Durations of notes, No. 10 The durations of some notes are altered, to allow for the presence of some longer held tones. Each piece is subdivided by chance into sections of varying lengths, with each subdivision assigned a probability for each note being extended, up to a variable, maximum percentage of original length.

In a similar way, another set of subdivisions determine the music's dynamics. Chance operations set the upper and lower limits for the loudness within each subdivision, with a variable weighting to skew notes towards the lower end of the range, to a greater or lesser degree. Another level of dynamics is produced by an additional weighting, which assigns a smaller probability of any given note having a particularly strong emphasis.

Finally, randomly selected notes are transformed into chords of two to five notes. Again, the probabilities of this happening vary from one section to the next, within each piece.

... and then doing it all again

The series of twelve pieces was completed in 2004, and a few copies were privately circulated on CD-R under a different title. A year or so later I listened to them again and decided that, after all that effort, there were too many deficiencies in the music that annoyed me, so I decided to set them aside and figure out how to fix them at some later date.

It wasn't until 2009 that I looked back through all my charts and notes to figure out what extra work needed to be done. The music had seemed too strident and sludgy all at once. I analysed the pitch distributions and found more notes than I expected had ended up at the extremes of the keyboard's range: more appropriate proportions were calculated and the excess notes were shifted according to these newly skewed probabilities. I also applied a more sophisticated distribution of dynamics.

The 2004 version retuned the piano for each piece in the series to the new, respective tonic, as was done with Stained Melodies. In the 2009 version the same tuning is used throughout, so that the intervals within the scale change whenever the tonic changes.

Even more techincal stuff

The measurements and structural details used to compose Real Characters and False Analogues were calculated in a series of spreadsheets in Microsoft Excel (some examples are given in the above images). All the chance operations and probabilities used in these calculations were obtained from Andrew Culver's freeware DOS program ic, an I Ching simulator (a web-based version is now posted on Culver's website). The tuning, and how to apply it, was worked out using Manuel Op de Coul's freeware tuning program Scala.

The editing and processing of all the MIDI data was done in the obsolete Cakewalk Pro Audio software, version 9. The transformations to the data required by the composition process were executed by writing a variety of CAL scripts in Cakewalk.

In the 2009 recording, the piano you hear is a free soundfont made by Marcos Guglielmetti and Roberto Gordo Saez, using Steinway piano samples from the University of Iowa's Music Instrument Samples library. The recordings were edited in another piece of obsolete software, CoolEdit Pro 2.

The Names

The title Real Characters and False Analogues refers to John Wilkins' 1668 An Essay towards a Real Character and a Philosophical Language, which proposed the existence of a universal language, written and spoken, which could communicate experience without mediation. It was believed this language could reconstruct the order of nature that God had revealed to Adam, before confounding man's language at Babel.

Many people have claimed that music is the true universal language. (The first modern artificial language was Solresol, which can be transmitted musically as well as verbally.) Unforunately, this particular species of musical fundamentalist is most likely to insist that some types of music are more natural than others, when in reality all music is, essentially, as arbitrary as any language. My title thus alludes to this self-inflicted conundrum, as well as its being written for a digitally-sampled grand piano.

Each individual piece is named after one of the myriad artificial languages invented over the past century.

CD and MP3s

Real Characters and False Analogues CD cover At the moment, Real Characters and False Analogues is being circulated privately only, as CD-Rs. A small run of CDs may be printed in 2010. However, good quality mp3 files all twelve pieces can be downloaded below, or streamed over at The Listening Room.

1. Lojban [250.0741 Hz] (4'23", 7.15 MB, mp3)

Lojban is a logical language originally designed to test the Sapir-Whorf hypothesis. Its root words are derived by a “fascinating methodology“ from the six most widely spoken languages: Arabic, Chinese, English, Hindi, Russian and Spanish. Grammar is based on first-order predicate logic. The name Lojban is a contraction for "Logical Language" in Lojban.

2. Ebubo [375.1112 Hz] (8'14", 13.31 MB, mp3)

Ebubo is a philosophical language based on common stereotypes of modern human cultures. Most words are five letters long, coined by combining a vowel prefix with two syllables. Each syllable has a distinctive meaning. For example, combining prefix "i" with syllable "ly" (sleep) and syllable "li" (furniture) means bed.

3. ODODU [281.3334 Hz] (5'23", 8.51 MB, mp3)

ODODU is designed to evolve in a pragmatic manner, so that it will eventually become a true universal language, using words derivable from our own experience and interaction with others. All words begin and end with vowels which specify grammatical usage; meaning is determined by interior consonants and vowels. Language is structurally modeled on general relativity theory.

4. Ithkuil [422.0000 Hz] (5'51", 9.52 MB, mp3)

Ithkuil combines an a priori philosophical language with a logical language. Word-formation uses an array of principles from cognitive psychology and cognitive linguistics, including prototype theory, radial categorization, fuzzy logic and semantic complementarity. Lexico-semantic compression is accomplished by lexifying only the seminal underlying concept of a class of what are related words in other languages, then applying a wide array of regular, predictable, and universally applicable modifications at the morphological (grammatical) level instead of the lexical level, to derive words that far exceed the lexical range of most other languages. There are two parts of speech, formatives and adjuncts, with the former inflecting for 22 morphological categories. The Ithkuil script utilizes a unique "morpho-phonemic" principle to likewise "compress" the written word.

5. Yiklamu [316.5000 Hz] (10'36", 16.12 MB, mp3)

Yiklamu is intended to enable interested users to explore the possibilities of highly disambiguated verbal communication. The unusual degree of disambiguation provided by the language derives from its lexicon, which was created by associating stem forms to the over 90,000 thesaurus entries of WordNet 1.6.

6. Liva [480.1423 Hz] (7'34", 11.83 MB, mp3)

Liva’s phonology follows a weird symmetrical scheme. Phonotactics is designed to obtain unambiguous word segmentation. Morphology is originally isolating, and tends to become agglutinative at a higher level. Syntax is based on predicate logic (as in Lojban) with a special role for deictics. Stems are (or will be) generated randomly.

7. Dashul [360.1067 Hz] (4'54", 8.27 MB, mp3)

Dashul follows the epistemological theory that words are perceptual concretizations of concepts, and that concepts are based on the integration of perceptual data. It therefore recognizes entities, attributes, actions, etc. as what they are, and develops a syntactic grammar based on logical principles. All other words are concepts of method, which allow preciseness in reference.

8. Minyeva [277.8601 Hz] (5'56", 9.76 MB, mp3)

Minyeva is intended to rigorously logical,both in regularity and in expression, and also to allow complex concepts to be said in a relatively simple and unambiguous manner. The vocabulary is arbitrarily made and unrelated to natural languages. The structure is regular, with some unique syntactic features: sentence structure is a causation string with each entity as a pivot around which their actions and states are placed. The case system is a unique tripartite system with is one type of subject and two types of objects.

9. Glaugnea [416.7902 Hz] (4'47", 7.94 MB, mp3)

Glaugnea has a set of particles that reflect its 4-value logic (true; false; true and false; neither true nor false) and 3-value observationals (apparently and really; apparently but not; despite appearances). Nouns have inflectional endings to indicate three numbers: singular, dual and plural; and four cases: nominative, genitive (for alienable possession only), major object and minor object. Verbs can be inflected to indicate tense and duration, mood, negation, frequency, and emphasis. The numerals are base-13 by default, although it is also possible to use the more ordinary base-10 system. Glaugnea can be written left to right or right to left, & horizontally or vertically.

10. Roxhai [312.5926 Hz] (6'03", 9.86 MB, mp3)

Roxhai has a core vocabulary based on nouns, with verbs, adjectives and adverbs automatically derived from these. For instance: r'o means "sentiment and moral power"; ro'yo, "religious affection"; roc'o, "supernatural being or region"; roca'yo, "Christian supernatural being or region"; and rocac'o, "God". Roxhai was inspired by Solresol, Ro and Esperanto, and is still being refined.

11. Kalaba-X [468.8889 Hz] (5'31", 8.85 MB, mp3)

Kalaba-X is one of the simplest useful languages; nothing more than a formalized grammatical description [verb (modifier) object (modifier) subject (modifier)]. It is a nonce language designed for a particular training exercise: to demonstrate the trade-offs required in translating to languages with very different grammatical structures.

12. aUI [351.6667 Hz] (8'23", 12.81 MB, mp3)

aUI is based on a set of 31 proposed universal semantic primitives, each designated with a simple ideograph, from which words from any culture can be created. Mini-definitions of a word's essential meaning, analogous to chemical formulas, are composed from these elements of meaning. Overcoming the arbitrary nature of existing languages, aUI incorporates an inherently meaningful relationship between Word, Meaning, and Reality, between Sound, Symbol, and Meaning; meant to be comprehensible even by rational alien creatures.

Ben.Harper, 2009.

* According to Nicolas Slonimsky, "The American pedagogue Percy Goetschius used to play the C major scale for his students and ask them a rhetorical question. 'Who invented this scale?' and answer it himself. 'God!' Then he would play the whole-tone scale and ask again, 'Who invented this scale?' And he would announce disdainfully, 'Monsieur Debussy!'" (back)

Real Characters and False Analogues, © Ben.Harper 2004/09. Recorded in Your Dad’s Den, Clapton. A Cooky La Moo production, edition number 20.