Home

Pronunciation

• Basics
• Consonants
• Vowels
• Orthography in Foreign Languages
• Phonetic Structure of Words
• Timing
  • Pitch, Stress and Comma
• Phonetic Rules & Pronunciation in Dialects
• Generating Stems
• Foreign Words
  • Typical Adjustments for Foreign Sounds
  • Semantics of Sounds
  • Short Words

Pronunciation

This  section describes how words are pronounced in Tyl-Sjok.

Briefly:

Vowels (all unrounded):
      Romanized         X-Sampa
     FRN CTR BCK      FRN CTR BCK
HGH   i   y   u        i   1   M
LMD   e   w   o        E   3   V
LOW       a                a

Palatized Vowels:
      je, ja, jo, ju, jy, jw

Consonants:
   Romanized/X-Sampa
     ALV  VEL  GLT      Remarks
STP   t    k    ?       usually voiceless, always unaspirated
FRC   s    x    h       usually voiceless
NAS   n   ng/N
LAT   l

Clusters:
      ts   kx
      nt   ngk

Phonotactics:
     C(j)V(C+(V(C)))

Accidentally, the vowels look very similar to those of Rumanian. The differences are: a) the low-mid vowels are mid and b) the back vowels are rounded in Rumanian. The Rumanian system thus is:

      Romanized         X-Sampa
     FRN CTR BCK      FRN CTR BCK
HGH   i  î/â  u        i   1   u
MID   e   ã   o        e   @   o
LOW       a                a

Basics

The goal of the design of Tyl-Sjok phonemes is to have a language that does not use the lips to form sounds. Therefore, no labial or labiodental consonants exist and all vowel are unrounded.

A word in Tyl-Sjok always starts with a consonant. The syllabic part of a syllable is always a vowel group, either monophthong or diphthong, thus no syllabic consonants exist (e.g. in contrast to colloquial German: `einen guten Lappen' [?aIn= "gu:t_nn= "lap_nm=]).

Note: Pronunciation is given in X-Sampa notation.

Consonants

Tyl-Sjok uses the following consonants. The consonants should be quite easy to pronounce for most speakers as they are common in many languages. German has most of them, most dialects of English have all but the velar fricative.

The sounds where chosen in order to be easy to distinguish even when mumbled and spoken without a lot of effort. Tyl-Sjok speakers are lazy. So uvular sounds or only used when they are closed to back vowels (not in all dialects), because they are too much effort otherwise.

Accordingly, palatal sounds are left out, because they are usually accompanied by lifting the jaw a bit in order to help the tongue reach the uppermost parts of the mouth.

The same holds for retroflex sounds which where left out, because it is exhausting to flip the tongue back (most speakers of Mandarin are also too lazy to pronounce any of the retroflex fricatives and affricates).

Tyl-Sjok Roman IPA Description

t t [t] alveolar voiceless plosive

s s [s] alveolar voiceless fricative

n n [n] alveolar voiced nasal

l l [l] alveolar voiced lateral approximant

k k [k], [q] velar (or uvular) voiceless plosive

x x [x], [X] velar (or uvular) voiceless fricative

N N, ng [N], [N\] velar (or uvular) voiced nasal

? ? [?] glottal plosive (glottal stop)

h h [h] glottal voiceless fricative

Note: Pronunciation is given in X-Sampa notation.
To clarify the pronunciation, it should be said that you should be lazy when pronouncing sounds of Tyl-Sjok. The [s] is pronounced like in some dialects of Dutch or of Spanish, so it is less tense than in German. All others are pronounced like in French (i.e., not like in German, where voiceless plosives are sometimes postaspirated in German, but are not in French and Tyl-Sjok).

For the usage of uvular sounds see Section 2.7.

The following affricates are pronouncible in Tyl-Sjok (only between vowel groups):

Tyl-Sjok Roman IPA Description

ts ts [ts] alveolar voiceless affricate

kx kx [kx] velar voiceless affricate

The following nasal-plosives are pronouncible in Tyl-Sjok (only between vowel groups):

Tyl-Sjok Roman IPA Description

nt nt [nt] alveolar nasal-plosive

Nk Nk, ngk [Nk] velar nasal-plosive

Vowels

The following vowels exist in Tyl-Sjok. They were chosen to get a set of easy to pronounce vowels that are distinguishable easily (they are separated far enough from each other), to keep the lips out of duty and to keep movement of the jaw as small as possible. It should be possible to generate the vowels with the use of the tongue alone without moving the lips or the jaw.

Tyl-Sjok Roman IPA Kirshenbaum X-Sampa Description

i i [i] i i unrounded close front vowel

y y [1] i" 1 unrounded close centre vowel

u u [M] u- M unrounded close back vowel

e e [E] E E unrounded open-mid front vowel

w w [3] V" 3 unrounded open-mid centre vowel

o o [V] V V unrounded open-mid back vowel

a a [a] a a unrounded open centre vowel

[i] and [E] occur in German: `bieten' ["bi:t_nn=] and `bäten' ["bE:t_nn=].

[M] sounds like `u' in German `Ruf' [Ru:f], but with unrounded lips.

[V] occurs in English `but' [bVt].

[3] sounds like `e' in German `Blume' ["blu:m@], but is a bit more open.

[1] sounds like `i' in German `blind' [blInt] but father back. In Russian this sound exists (transcribed as `y'). It is usually diphthonguised in Russian but is a clear monophthong in Tyl-Sjok.

Note that the vowels are pronounced only by movement of the tongue. The ideal position is having your mouth almost closed, with the teeth at a distance of 1 to 5 millimetres. The sounds then may shift away a bit from the descriptions given above.

All vowels are short (but lengthened a bit by a stress).

The following palatised vowels are pronouncible in Tyl-Sjok:

Tyl-Sjok Roman IPA

je je [jE]

ja ja [ja]

jo jo [jV]

ju ju [jM]

jy jy [j1]

jw jw [j3]

Orthography in Foreign Languages

When using Tyl-Sjok romanisation in other languages, it is allowed to change the Romanisation a bit to match the rules of those languages. In particular, the following changes make life easier:

• ' (glottal stop) may be written as `q'
• `N' (velar nasal) may be written as `ng'

These changes are explicitly allowed, because capitalisation and punctuation should not be messed up.

Other changes should not be performed to prevent an explosion of possible orthographies. The following typical changes some languages would apply are explicitly forbidden, except for explanations of pronunciation.

• `x' (velar fricative) may not be written as `ch' or `kh'.
• `w' (open-mid centre vowel) may not be written as `eu'.
• `t' (voiceless alveolar stop) may not be written as `d'.
• `k' (voiceless alveolar stop) may not be written as `g'.
• `e' (open-mid front vowel) may not be written as `ä', `é', `è' or `ê'.
• `u' (close back vowel) may not be written as `ou'.
• `y' (close centre vowel) may not be written as `ı'.

Phonetic Structure of Words

Because Tyl-Sjok is an isolating language, its words are pure stems. We now describe the structure of such a stem.

A stem consists of an arbitrary count of syllables following the pattern CV or CVC, where C is a single consonant and V is a monophthong or a diphthong. The following constraints reduce the number of possible syllables.

• Glottal consonants ([h] or [?]) do not occur at ends of syllables.
• Velar nasals ([N]) do not occur at the beginning of syllables.
• Diphthongs only occur in the first syllable of a stem.
• Stems always end in a consonant.

The last rule is not true for particle stems which may end in a vowel if that particle always precedes another word.

When several syllables are combined to a stem by concatenating them, the following constraints are placed on two adjacent consonants:

• Two adjacent consonants inside a stem must either form an affricate, or
• be a nasal followed by a plosive at the same place of articulation (i.e., [nt] and [Nk] are pronouncible).

Tyl-Sjok stems are mono- or bisyllabic. Longer stems only occur with foreign words. The syllables of such words are written in one character each.

Timing

 All mora should have about the same length. A mora is assigned by consonants: the patterns CV, CjV, C are one mora each: palatization does not add a mora.

Examples: Hwk tje xok u set are 8 morae: hw, k, tje, xo, k, u, se, t. Also note that single plosive mora tend to be devoiced and sometimes aspirated, depending on dialect. Dialects that like to aspirate also tend to devoice isolated `l' mora.

The timing rule applies to all stems, mono- and polysyllabic.

The sequences `t+s' and `k+x' will unite into an affricate, and therefore eliminate the single consonant mora of the plosive. This happens inside words and across word and phrase boundaries, too. (This applies to `t+sj', too).

FIXME: This rule may be subject to change. I'm still experimenting..

Pitch, Stress and Comma

   Each syllable is assigned a pitch. There are two pitch levels, low and mid. These are vaguely equivalent to the IPA pitches 1 and 3. Southern dialects have also a high pitch (IPA 5). Syntax defines which pitch is to be used by defining which syllables are stressed.

Words that are the last child of a node in the syntax tree are accented. Non-accented words get mid pitch.

In southern dialects, the last word of an accented sequence is assigned low pitch, all others high pitch.

For northern, standard dialects, all accented syllables are assigned low pitch.

In polymoraic stems, all mora are pronounced equally accented/pitched.

Pauses may be inserted after the last low pitched stem of a low-pitch sequence. Commas must be placed after all accented words in the romanisation.

Figure 2.1 depicts the pitch assignment for southern dialects.

  
Figure 2.1: Yellow syllables get low pitch, green ones high pitch, all others mid pitch.

Usually, northern pronunciation will be depicted.

Phonetic Rules & Pronunciation in Dialects

 Tyl-Sjok is designed to have a standard pronunciation and several dialects. Some of the following phonological rules are used by all dialects, some rules only by a few.

None of the following phonetic rules is expressed in the orthography of Tyl-Sjok. This section might also me names `From Phonemes to Phones'.

• A mandatory rule for all dialects is that nasals at word boundaries adjust their position to that of a following plosive.

  [n], [N] + [t] [nt]

  [n], [N] + [k] [Nk]

  [n], [N] + [q] [N\q]

• In all dialects, `hj' is pronounced [C].
• In most dialects, the sequence `sj' is pronounced [S], others keep [sj].
• In some dialects `hi' is pronounced [Ci] and `si' is pronounced [Si]. This is not considered the standard pronunciation.
• In most northern dialects, duplicate consonants at word boundaries are spoken as long consonants (geminates).
• In most colloquial dialects, the glottal stop is dropped almost completely. All kinds of morae collapse when a syllable starts with a vowel or with `j':
  • C# + V -> CV (e.g. ni.n.a.t ni.na.t)
  • Ci# + V -> CjV (e.g. ni.e.xu.l nje.xu.l)
  • C# + jV -> CjV (e.g. a.t.ja a.tja)

  Only between other vowels, the glottal stop is sometimes pronounced, especially if the colliding vowels are identical. Additionally:

  • y + i -> y + ji (e.g. y.i y.ji)

  C + C C:

  In contrast to that, most other dialects simply eliminate one of the consonants.

  C + C C

  (Note that this eliminates morae, which is important to know for poems. Most poems are written for the standard dialect.)

• In southern dialects, voice assimilation applies. Voice is non-phonemic, so it is fully comprehensible to always keep t, s, k, x, sj and hj devoiced. However, in a voiced context, these consonants tend to be voiced, too. E.g. `king ka' is usually pronounced [kiNga] and `kin sa' becomes [kinza]. This voicing also applies for fricatives: `kitsa' = [kidza].

  In northern dialects, voice assimilation only applies to plosives and affricates: t, ts, tsj, k, kx. So `kin sa' is pronounced [kinsa].

  Please note that there is a strict difference between voiced `hj' (voiced palatal fricative) and `j' (voiced palatal approximant).

  (If the dialect likes to devoice single consonant morae, this voicing does not occur after such a mora, of course. So the pronunciation of `kik ta' would be [kikta]).

  (This voicing rule is similar to that of Korean.)

• In a few northern dialects that are not exposed to much communication with other dialects, velar consonants (and even consonant clusters) move backwards to become uvular if they follow a back vowel. This uvularisation moves left-to-right.

  V[+back] + C[+velar] V C[+uvular]

  Ca[+uvular] + Cb[+velar] Ca Cb[+uvular]

  Note that this rule also applies across word boundaries. So kul ?il xuk kyng ?wng xuk (the hexadecimal number 0x510) is pronounced [kMl ?il xMq:1N ?3N\ XMq]

  Most speakers from other dialects fail to imitate this.

• Some southern dialects do not distinguish [3] and [1]. Both sounds collapse to a central mid vowel, a schwa: [@].
• Some eastern dialects do not distinguish [1] and [i] but pronounce both as [i]. The same rule applies to [3] and [E] which both become [E]. The rule is:

  V[+central] V[+front]

The normal northern dialect (long consonants, but no uvularisation) is considered the standard way of speaking. The standard, northern pronunciation is usually clearer and more elegant wrt. phonetic constraints that most other dialects, which are less distinctive.

Please note that there are quite a lot of articulation points of fricatives (s, sj, hj, x, h), but no phonemically different fricatives at the same point of articulation.

Generating Stems

There is a computer program that generates stems for Tyl-Sjok. It comes with the Tyl-Sjok distribution and is called s2makestem.cc; a Makefile is provided for compilation.

The programs first command line parameter is the number of syllables (defaults to randomly generate one to three syllable words), its second parameter is the number of words to generate (defaults to 100).

If the option -t is added, the output becomes IPA compatible LaTeX code.

If the option -p is added, a simplified particle stem is generated.

Foreign Words

Sounds from foreign language, like names, are incorporated into Tyl-Sjok by either shifting the sounds to the closest pronouncible pattern, by occasionally dropping sounds, by separating vowel groups with glottal stops and seldom by introducing new syllables duplicating a vowel in order to be able to articulate consonants clusters. Dropping consonants is more frequent, though. Rules for monosyllabic stems may be used to judge what is pronouncible. Each syllable is written in one character. The first one contains the semantic information, the others are pure pronunciation syllables.

E.g., the German name `Matthias' becomes `na-ti-?as'. `Henrik' becomes `hen-lik'.

Typical Adjustments for Foreign Sounds

bilabial sound

alveolar sound,

rounded vowels

unrounded vowels,

vowel clusters

either the most prominent vowel or several syllables, using glottal stops for separation. Sometimes, diphthongs are rendered as V1 + h + V2: e.g. `ai' is rendered as `ahi',

consonant clusters

either some consonants are dropped or several syllables are used, using the close center vowel or some vowel from the neighbourhood for syllable completion,

palatal sound

alveolar or velar sound (depending on position) e.g. Icelandic `banki' would be become tahuN-ki , but Mandarin `xin' would become sin,

uvular, pharyngeal, laryngal sound

velar sound.

[j]

sometimes `i', sometimes left out,

[w]

sometimes `n', sometimes `u', sometimes left out,

Other features are simply left out (tones, vowel length, stress, aspiration of plosive, etc). Long consonants are usually also not taken into Tyl-Sjok names.

Usually, foreign words are split into several monosyllabic stems instead of using bisyllabic ones. This happens, because stress on parts of foreign words sounds strange to Tyl-Sjok speakers. Collapsing into one word sometimes happens when a word is not felt to be foreign anymore.

Examples of Adopted Names

• Linus Torvalds li.FN nus<> tol<> sal<> tys<>
• Björk tw.FN lyk<>
• Reykjavík lei.FN kiha<> sik<>
• Beijing teit.FN siN<>
• Taiwan tahi.FN nan<>
• Francois Mitterand sa.FN lan<> suha<> ni<> tw<> loN<>
• Heidemarie Wiecorek-Zeul hahi.FN tw<> na<> li<> sit<> so<> lek<> sohil
• Saddam Hussein sa.FN tan<> hu<> sein<>

Semantics of Sounds

 In many words in Tyl-Sjok, the sounds are related to the meaning. Here is a vague, incomplete list. Note that this are preferences only.

h, ?

indicate closeness to agent or speaker

alveolar consonants and front vowels

indicate `after', `leaving'
`large', `high', `strong', `good', `open'

velar consonants and back vowels

indicate `before', `coming' (the air in the mouth comes from the back)
`small', `low', `weak', `bad', `close'

central vowels (y, w, a)

indicate average

palatized vowels

indicate change

plosives

relate to location

fricatives

relate to time

high vowels

relate to location

mid-low vowels

relate to time

Several regular changes occured when ancient Tyl-Sjok underwent sound changes.

diphthong iu

became i or ju

diphthong oe

became o or jo

diphthong ie

became i or je

diphthong uo

became u or jo

initial N

became n or k

Short Words

The following lists the CV (one consonant, one vowel) words:

i e a y w u o

l EMPH turn, become to last do

t NEG become not to last not do not

s REF so, correct n

k together DET

x SKIP START

? NULL be-at and

h YN happen-when

The following lists the CjV (one consonant plus palatized vowel) words:

je ja jy jw ju jo

t go-from

s start-at n l

k open go-to close

x last-until

you he/she/it I h

The following lists ?VC (glottal stop, vowel, consonant) words:

t s n l k x ng

i say, talk power of

e near

a WHICH water

y jacket w

u harass

o END h

Home