Phonology and Romanisation
Keywords: phonology, phonotactics, prosody, stress, timing, romanisation
The phonology of Classical Gluonic is often treated as an absolute known, because there are still people alive who can speak it, similar to Latin. Unlike Latin, there are actual, extant recordings of people speaking Classical Gluonic going back hundreds of years. However, the Gluonics saved only a fraction of their evidently once immense cultural heritage during the Great Departure, and in particular, a paucity of electronic records survived. In the early centuries of Gluonic society in Alt-1136, electronic records were by far the greatest bulk of all records of any kind, so the loss is devastating for any clear picture of early Gluonic society.
One disadvantage we have in confirming the original phonology of Classical Gluonic is that it has only one daughter language, Modern Gluonic, so the comparative method is of little use. Another is that the native orthography is a quasi-logographic syllabary - we know what the modern Gluonics tell us is the Classical pronunciation of each symbol, but we gain less information than if the language employed a more purely and granularly phonological writing system.
Where we do have one advantage is that the characters were type written on electronic keyboards that used special "compositional symbols" typed in sequence to identify and produce each character of the written language. In other words, they had an alphabet, that they only used to type their logographs, and for no other purpose. However, we know the "symbolic composition" of all the syllabic characters, and that gives us some additional phonological evidence about the classical language.
We certainly have a reasonably high degree of confidence in the phonology of Classical Gluonic. But there is enough uncertainty present that it has spawned a number of heterodox theories about the "real" Classical Gluonic phonology, or for a hypothetical, pre-classical phonology (for example, see Fong et al., 2001). While there are a few tantalizing clues that the conventional Gluonic view of their own classical language has some flaws, I am going to present the modern scholarly consensus, that mostly aligns with what the Gluonics themselves believe about their classical language.
Classical Gluonic conceptually had three "grades" of "strength" for obstruents except for /h/, and two grades for nasals, and rhotics. This much is non-controversial. In the table below, consonants are represented by the IPA symbol for what is thought to be the sound of their weakest grade, with a suffix of 1, 2 or 3 for increasing strength. The phoneme /r/ is different in that it is thought two have two types rather than two strengths, represented by the suffixes A and Z.
|Nasal||m1, m2||n1, n2||ŋ1, ŋ2|
|Stop||b1, b2, b3||d1, d2, d3||g1, g2, g3|
|Fricative||v1, v2, v3||z1, z2, z3||ʒ1, ʒ2, ʒ3||h|
According to what the Gluonics tell us, the threefold strength patterns of some Gluonic consonants is supposed to represent the threefold symmetry of quarks and leptons with the same charge, and the consonants with twofold strength distinctions are supposed to represent the symmetry between the series of quarks with +3/2 electric charge versus the series with -1/2 charge, the symmetry between charged and neutral leptons, and the symmetry between quarks and leptons. There are six sets to represent the six quark flavours, with symmetry between the three stops and three fricatives again evoking the twofold symmetries within the quarks and the leptons. The six remaining phonemes are supposed to represent the five gauge bosons (counting W+ and W- as two, represented by the two types of /r/), and the Higgs boson, represented ironically but appropriately by /h/.
In the Gluonic view, the notional phonemes of Classic Gluonic are not meant to represent specific entities of the Standard Model of particle physics, of which our society gained much of its knowledge from theirs, but rather to evoke its symmetries and thus tie into what they felt were its sacred properties in a more mystical and spiritual way.
Here is the phonology of Classical Gluonic in terms of how learned Gluonics tell us CG is supposed to be pronounced:
|Nasal||m, mː||n, nː||ŋ, ŋː|
|Stop||b, p, pː||d, t, tː||g, k, kː|
v, f, fː
|z, s, sː||ʒ, ʃ, ʃː||h|
This inventory is not fully accepted by Terrestrial Gluonicists, pointing out correctly that due to rules of allophony, not all of these sounds can appear in all positions, and in the stop series, aspiration appears allophonically as well. The most popular competing theory is the "ejective theory" of CG phonology, which posits that the third grade for stops was originally not a geminate consonant but an ejective consonant. While it is true that some speakers do sometimes realise the third grade as an ejective, this is not a diachronic effect, and no conclusions can be drawn. It is easier to understand CG in terms of its conventional form, and this is how modern speakers will work to render it.
For the rhotics in this scheme, /rA/ corresponds to the trill [r]. and /rZ/ corresponds to /ɹ/.
Here is the actual, phonetic consonant inventory of Classical Gluonic as it is actually spoken today when allophonies are taken into account:
|Nasal||m, mː||n, nː||ɳ||ŋ, ŋː|
|Stop||b, p, pʰ, pː||d, t, tʰ, tː||g, k, kʰ, kː|
v, f, fː
|z, s, sː||ʒ, ʃ, ʃː||ç||x||h|
Given the religious quality of the original language design, most Gluonicists believe it is doubtful that the allophonies that result in the above inventory, which ruin the symmetries of the notional inventory, were part of the language's true, original phonology; rather, it is believed that the first Gluonics probably tried to pronounce the original grades, whatever they were, exactly the same in all environments. This is part of what leads to the ejective theory, because syllable-initial geminate consonants are quite uncommon in the world's languages, yet the third grade consonants routinely appear at the beginning of syllables. Some feel that given this distribution, an original ejective realisation makes more sense.
The following basic rules apply to the realisation of CG consonants as spoken by modern Gluonic L2 speakers syllable-internally (i.e., generally not conditioned by neighbouring syllables and not affecting syllable structure and stress):
1. Third Grade Stops
Third grade stops are the strongest stops. They are realised as voiceless geminated stops, but are lenited word-initially. Let S3 represent a third grade stop consonant:
S3 > [-long][+aspirated] / #__
In other words, word-initially, the third grade is distinguished from the second grade only by aspiration, except by those speakers who sometimes represent these consonants as ejectives in their personal idiolects, in which case, the above sound change does not occur. The aspirated version of the consonant still counts as one mora in terms of syllable structure and does not impact stress.
2. Alveolar Nasal /n/
In non-careful speech, the short alveolar nasal /n/ tends to assimilate to the place of articulation of anything that follows it. It becomes /m/ before labials /v/ or /b/ (all grades), /ɲ/ before palatal /j/, and /ŋ/ before velar /k/ (but not /w/). These changes are conditioned by sounds in neighbouring syllables, as they can only occur with coda nasals, but they are listed here because they are considered to not affect syllable structure, and speakers tend to be less aware of them.
[n] > [m] / __[+labial][-nasal]
[n] > [ɲ] / __[+palatal][-nasal]
[n] > [ŋ] / __[+velar][-nasal]
A different "syllabic repair" process applies when /n/ comes before another nasal, detailed under Phonotactics, that can affect syllable structure and stress.
Note that these processes apply to short nasals only.
3. Glottal Approximant /h/
The glottal approximant [h] is in free variation with [x] and [ç] when spoken by many modern Gluonics. It is actually supposed to be [h] in all environments, but modern speakers struggle with this.
The vowels of Classical Gluonic exhibit the same symmetry of threes and twos that characterise the consonants and for the Gluonics symbolise the deeper, underlying symmetries of quantum mechanics. Gluonic vowels have a threefold symmetry between Front, Mid and Back, a twofold symmetry between High and Low in each of these positions, and a further twofold symmetry between Long and Short at each coordinate.
The following are the theoretical vowels of CG. In the theoretical representation, rather than choosing a "base IPA value," instead they are represented featurally with three symbols, F/M/B for Front, Mid and Back, H/L for High/Low and 1/2 for Short and Long. Note also that vowels come in quartets of long and short vowels of two distinct articulations that are regarded as having the same "colour" - this is noted in the column headings below and becomes important later.
|Front (Red)||Mid (Green)||Back (Blue)|
|High||FH1, FH2||BH1, BH2|
|Mid||FL1, FL2||MH1, MH2||BL1, BL2|
As with the consonants, the modern Gluonic people can tell us what they think are are the proper values of these vowels, and just as with the consonants, Terrestrial Gluonicists do not always take them at their word, however, the majority view among Terrestrial scholars does coincide with the conventional Gluonic view. This gives us the following vowel inventory:
|High||i, iː||u, uː|
|Mid||e, eː||ə, əː||o, oː|
Educated Gluonics who are speaking Classical Gluonic carefully actually pronounce the vowels in just this fashion, and will certainly tell you this is how they are to be pronounced. However, in practice, most Gluonic speakers have an enormous degree of difficulty not pronouncing CG in a way that is heavily influenced by the Modern Gluonic vowel inventory. This and other MG influences on actual pronunciation of CG can be thought of as analogous to the difference between Ecclesiastical Latin versus Classical Latin. That is a topic for another place, however.
Classical Gluonic has two diphthongs - MLFH and MLBH, realised in the common understanding as [a͡i] and [a͡u]. Diphthongs have the same length as long vowels and generally behave just like long vowels in other respects, for example, in placing stress.
Classical Gluonic is a mora timed language. It has a very regular pacing, and morae are thought to take roughly the same amount of time to pronounce (although modern experimental evidence has cast some doubt on this point of view). Modern Gluonic is not mora timed, and this is an aspect that modern speakers struggle with when trying to speak CG in a polished way.
The phonotactics of Classical Gluonics is linked to its timing. The basic syllable shape is (C)(C)V(N). A CG syllable may be light (one mora), heavy (two morae) or superheavy (three morae). Syllables with more than three morae are forbidden. The rules are:
- An onset consonant of first or second grade (other than nasals) takes up no morae.
- A third grade consonant in the onset takes up one mora.
- A second grade nasal in the onset takes one mora.
- A consonant cluster in the onset takes up one mora.
- A short vowel in the nucleus takes up one mora.
- A long vowel or a diphthong in the nucleus takes up two morae.
- A coda consonant takes up one or two morae depending on length.
Consonant clusters within a syllable can only occur in the onset. A consonant cluster takes up one mora, and therefore it must add up to a third grade consonant in weight, according to the design of CG. This fact determines the limits of the types of cluster that can be formed. In addition, CG follows the standard sonority hierarchy. The underlying sound symbolism of CG phonology also comes into play a bit.
The following rules are observed for the parts of the syllable:
- May be any vowel or diphthong.
- However, if a long vowel or diphthong, it uses up two morae and restricts the possibilities for the onset and coda, since there can only be three morae.
- Any single consonant can appear in the onset.
- The onset can also be ∅
- The maximum length of a consonant cluster is two.
- A consonant cluster in the onset must add up to three strength "units," i.e., be equivalent to a single third grade consonant.
- The first consonant must be a second grade member of one of the "threefold fermionic" phonemes - a stop or a fricative.
- The second consonant must be a member of the "gauge bosonic" phonemes (a liquid or glide), which are considered to have a grade of one, and are higher in the sonority hierarchy than fermionic phonemes.
- A further consideration is that consonant clusters must have wide "sonority spacing," so the somewhat more sonorous fricatives can only pair with the glides (/j/ and /w/), not with the liquids.
That leaves the following possible consonant clusters in syllable onsets, in conventional IPA transcription:
pl, pr, pɹ, pj, pw, tl, tr, tɹ, tj, tw, kl, kr, kɹ, kj, kw, fj, fw, sj, sw, ʃj, ʃw
The onset takes one mora and is "heavy" if it begins with any of the above clusters or a third grade consonant, otherwise it takes no morae and is "light."
The coda can be ∅, and it can be any nasal, /m/, /n/, /ŋ/, /mː/, /nː/ or /ŋː/. It takes up zero morae if ∅, one if a short nasal, and two if a long nasal.
Number of Possible Syllables
- 14 syllables that consist of just a vowel or diphthong. 6 are light and 8 are heavy.
- These can be paired with 21 light onsets to give 294 light onset open syllables.
- The 126 of these with a short vowel nucleus can pair with any nasal to give 756 light onset heavy or superheavy closed syllables. The remaining 168 with long nuclei can pair with light nasals to give 504 superheavy closed syllables.
- There are 30 strong onsets that can be combined with 6 short vowels to give 180 heavy onset heavy syllables and with 8 long vowels/diphthongs to give 240 heavy onset superheavy open syllables.
- The heavy open syllables can all be paired with a light nasal coda, to give 540 superheavy closed syllables.
This gives a total of 2528 possible syllables in Classical Gluonic, of which 132 are light, 734 are heavy and 1662 are superheavy.
Every one of these possible conventional syllables that is actually attested in the language has a separate character in the native orthography representing it. Not all are actually attested, however.
Word-Internal Limitations and Repair Strategies
In writing, there are no limits word-internally to what sounds can be adjacent to each other, as long as they are composed of valid syllables. Any syllabic symbol can be next to any other, and they are supposed to be able to be pronounced exactly as-is.
In practice, however, there are a few limits and repair strategies that the Gluonics acknowledge and use when speaking careful Classical Gluonic. This is particularly evident when two vowels are adjacent to one another. In this instance, the vowels are supposed to be separated by hiatus (even two adjacent instances of the same long vowel!) and neither blend nor insert an epenthetic consonant, even when they are of the same articulation. In practice, modern speakers cannot do this and resort to repair strategies.
- Two of the same short vowel adjacent: This results in practice in a single, merged syllable with a long vowel, if the resultant syllable would not have more than three morae. This effect can result in stress in a word shifting away from the syllable where it would normally be expected. Otherwise, an epenthetic /h/ is inserted.
- Two of the same long vowel adjacent: Separated by an epenthetic /h/
- Two vowels of any length of the same colour but different height adjacent: Also separated by an epenthetic /h/
- Two vowels of different colours adjacent: Separated by hiatus, just as in the formal understanding. Except:
- /a/ followed by /i/ (i.e., the short versions of these vowels) - This results in practice in a single, merged syllable with the diphthong /a͡i/, if the resultant syllable would not have more than three morae. This effect can result in stress in a word shifting away from the syllable where it would normally be expected. Otherwise, there is a hiatus.
- /a/ followed by /u/ (i.e., the short versions of these vowels) - same as above for /a/ and /i/, except the result is /a͡u/.
- Two short nasals adjacent: The second nasal assimilates to the place of articulation of the first and they become a long nasal, if the resulting merged sound can be attributed to one of the syllables. It will become the onset of the second syllable as long as that does not result in more than three morae in the syllable, otherwise it becomes the coda of the first syllable as long as that doesn't result in a syllable with more than three morae, otherwise the nasals remain unmerged and attributed to their separate syllables.
- A short and a long nasal adjacent, or two long nasals adjacent: Both nasals become short, and then the process in #5 above is applied. This can reduce the number of morae in either syllable and result in stress shifting.
- A long nasal followed by a third grade consonant of any kind will become short, which can reduce the number of morae in its syllable and result in stress shifting.
- A nasal coda followed by a null onset will be reanalysed as the onset of the second syllable as long as this does not result in more than three morae in the second syllable. This can change syllable weight and result in shifting stress, since the first syllable will lose weight.
- Otherwise, observe the rules of "allophony" listed under Consonants, which generally either aren't conditioned on neighbouring syllables, or else don't have the potential to impact syllable structure.
Lexical stress in Classical Gluonic is realised in higher pitch and loudness of the stressed syllable. All words have a primary stress and long words may have a secondary stress, typically seen in compounds, where the secondary stress is placed where it would have been the primary stress in the independent word.
The stressed syllable is the first syllable of the heaviest type that appears in the word. For example, if a word has one or more superheavy syllables, the first superheavy syllable will be the stressed syllable. For some speakers, if the stress falls on a short vowel, the vowel may be lengthened noticeably.
Stress is for the most part non-phonemic, completely predictable, and not represented in the orthography or (usually) the romanisation.
There are two main romanisations used by Terrestrial Gluonicists, the Bourque romanisation and the Sanderson romanisation. They were invented with different goals in mind and were originally competitive, but modern Gluonicists find both useful for different purposes, and both are in active use today.
The Bourque romanisation is designed to faithfully transliterate the orthography of Classical Gluonic, by creating a romanised transliteration of each one of the syllabic and logographic symbols of the CG language. It was very useful for discussing the CG language as it was actually written in an age before Unicode, where typesetting the well over two thousand symbols of a new and unfamiliar language was very challenging. It is less dominant than the Sanderson romanisation today, but remains very useful for discussions of the language that are focused on its literature or on its underlying ideal form.
The Bourque romanisation is a very complex topic that deserves a much more in-depth discussion in conjunction with a discussion of the native orthography. Here I will only give a very short description.
Basically, like the native orthography, Bourque is written in syllable blocks and reflects the basic pronunciation before any inherent or situational sound shifts are applied. These sound shifts are not represented in any way in writing, the reader has to be aware of them and apply them from knowledge of the rules, exactly how the native orthography is read. It uses the following phonemic symbols to build syllables:
|Theoretical||Conventional IPA||Bourque Phonemic Symbol|
The Bourque romanisation uses all capitals to create representations for Classical Gluonic syllable glyphs. Lowercase letters are reserved to create representations of the purely logographic characters which perform critical functions in CG texts but which are not written out based on the sounds of their realisation in a certain environment, and to denote aspects that would normally be literally conveyed with font colour, like grammatical colour. Where a Sanderson text would try to represent the phonology of the entire resulting words, a Bourque text would strive to be faithful to the original written representations, including non-phonetic parts.
A text is written as a string of base syllables separated by hyphens, with words separated by spaces. The Bourque romanisation also has punctuation corresponding to the CG native punctuation symbols. Here is an example of how a short phrase might look:
The reason for this stylistic choice around the capitals is to make the text resemble a CG text more, with the larger syllabic symbols and the smaller logographic function particles, in this case Green colour markers. Part of the realisation of the green marker would be to change stressed vowels in the roots to green vowels, but you can see they are represented with their "black" base realisation instead in Bourque, just as in the native system.
Bourque uses asterisks to indicate higher grade consonants in most cases, but uses apostrophes for the stop consonants, possibly because Alain Bourque was a believer in the theory that the third grade stops were originally ejectives.
The Sanderson romanisation endures as the most popular romanisation of Classical Gluonic, and the one preferred by Gluonics themselves when romanising CG. It tries to represent how CG is really spoken, including applying all the allophony and syllable boundary repair rules. John Sanderson's goal was that it be easily readable out loud. It does not try to respect or represent the syllabification represented by the native orthography. It is also intended to be convenient to write for English-speakers, avoiding diacritics, but it is utilitarian, making some choices that many people have criticised as ugly for expedience, especially with the letters x, c and q (choices influenced by Bourque). Complete unambiguity is not a goal of the system.
|Theoretical IPA||Allophonic Variations||Sanderson Phonetic Symbols|
|pː||pʰ, pː||ph, pp|
|tː||tʰ, tː||th, tt|
|kː||kʰ, kː||kh, kk|
- Allophone [ɲ] not represented.
- Allophones of /h/ not represented.
The Sanderson romanisation also doesn't represent stress, but an acute accent on the first vowel symbol of the syllable nucleus is optionally allowed to be used to indicate stress, especially to help learners. It uses the punctuation of Canadian English almost exactly, including conventions around capitalisation. The same example as given in Bourque above could be written in Sanderson as:
The actual realisation of the third grade onset and the assimilation of the 'n' in T'ÍN is shown in Sanderson, as is the colouring of the stressed consonant green, and both words have their colour suffixes written out phonetically.
Throughout this site, I will tend to use the Sanderson romanisations more than the Bourque forms, but the Bourque romanisation is indispensably useful in order to discuss the language's underlying structure and design, as well as its full, synchronic cultural context, and will also be employed.