A fair minority will maim, rape, kill, burn, torture, and destroy, just for fun. Despite modern romance novels and movies, bandits are not nice people. Those few who survive for long rapidly become tough, skillful with at least a few weapons and usually with everything short of heavy armor, although they usually stick with light armor in the interests of mobility , familiar with the area, and well-equipped with contacts to tip them off about law enforcement sweeps and rich targets — and to handle their connections with the rest of the world.
Sadly, most Bandits are fairly poverty-stricken. Mercenaries — whether companies, native auxiliaries, or bodyguards — have been around for a very long time. Just as importantly, when you come right down to it, every mercenary is an individual. Their ultimate loyalty is to themselves — and so their lives and reputations will take priority over your desires. Their skills tend to reflect that attitude, as well as the fact that they are likely to be operating unsupported, and with no one else to rely on. Like the soldier, and for much the same reasons, mercenaries tend to go for the heavy weapons and armor.
Unlike soldiers, mercenaries tend to be able to do their own planning, repair weapons and armor, act as emergency medics, and otherwise support themselves. They want to get enough money to set up in business, purchase land, or otherwise retire. That kind of job offers many perks for a clever mercenary — over and above the really big one of being able to pick and choose among possible jobs, or walk out if things get rougher then you contracted for.
Background , Character Sheets Tagged: You are commenting using your WordPress. You are commenting using your Twitter account. You are commenting using your Facebook account. Notify me of new comments via email. This site uses Akismet to reduce spam. Learn how your comment data is processed. Welcome to my roleplaying blog! The Emergence Campaign Weblog exists to easily distribute material for my players, in support of Eclipse: The Codex Persona and other Distant Horizons Games products, and to provide a home for my role-playing material - along with occasional player contributions - in general.
Queries and special requests are welcome. Here's a full post list: Create a free website or blog at WordPress. Sometimes it's hard to tell. Leave a Reply Cancel reply Enter your comment here Fill in your details below or click an icon to log in: Email required Address never made public. Eclipse d20 — Binding Mysterious Spirits V. Eclipse d20 — The Founders. The Immortal Rants of Sean K.
Categories
Epic Spell Conversions, Part V. This ensures that there can be no triple consonant clusters. There is a lot of work that I have yet to do with the phonotactics. There are some pretty significant restrictions on onset consonant clusters, but I haven't fleshed out all of the details on that yet. So far I've only been able to list permissible word-initial clusters. They are for the time being:. These clusters appear in ancient forms of Salhari words and certain loanwords, but otherwise cannot occur at a syllable onset. I chose these sounds primarily for aesthetic purposes, but also for ease of pronunciation.
I don't see the point in creating a language that I can't even pronounce. Some may criticize this phoneme inventory as too IE or not "exotic" enough; and if that's your opinion, you're welcome to it. But my purpose in choosing these sounds was not to make use of weird phonemes, it was to create something that is phonoaesthetically pleasing to me. In terms of sociolinguistics, I do have some notes on things like dialectical variation, but very little of it will make sense until I tell you a bit more about the language.
Please excuse my formatting. No matter what I do, reddit keeps fucking it up to where what I typed in the comment field does not look at all like what comes out when I submit it. I've tried reformatting it several times, but to no avail. It's pissing me off too much, so here you go. Words are kept sounding similar to the languages they are pulled from, when they are changed it is to shorten them or to make them more pleasing to the ear of their creator.
Laminan has a two hiated? This can present a problem when a verb or word stem ends in 2 separate vowel sounds and must have an ending beginning with a vowel attached. Because the language is learned by mere presence in Lamina, there are no distinct dialects of the language or any major differences throughout the islands or regions. I picked these sounds, because I wanted to give my conlang a very formal, British feel, while still making it completely unique.
There some variations to how you would speak depending on social class. While coming up with the language's vowel system, I started with the basic English five syllable system, a, e, i, o, u, and inserted those letters into words. I spoke in Dzarian enough that I noticed natural changes in the sound of the language, and adopted those changes as the normal sounds; this made the language sound more natural.
Also, Dzarian is notorious for having anomalous sounds and letter orders which can only appear in a couple of words in the language, these are often not mentioned by me when documenting the phonology. Notice, in this dialect, how vowels at the beginning and middle of the word tend to centralize, whereas sounds at the end tend to do the opposite. This is entirely dialectal, in Central, West, and Southern Dzarian, the opposite usually happens. Just kinda what I thought fit the script I had made for it. Plosives are fun, liquids help when forming consonant clusters, and the nasal because there was still a character left.
Liquids don't start or end words and the nasal may never be attached to plosives. Plosives are never connected The nasal really is any nasal sound. Liquids tend to be held out longer. Because your character is just labeled "nasal" and those are all nasals. Or is a speaker reading the written form just supposed to know which particular nasal comes next? They are completely free in variation.
In my vocab section I alternated a little to try and show this, but I'm new at this and wasn't really sure how to explain it. Mostly it was laziness. I figured if I leave it open now I can play with it more later. It may end up as being different nasal phonemes that share the same written character but pronounced differently in different words, or being different nasals for different intent, or something else.
But I didn't feel like working it out, so right now it is just free variation. Perhaps there can be different allophones depending upon the environment. Yeah, it will most likely become something along those lines. Thanks for using all of the proper vocabulary so I can know what to look up later.
Length is contrastive for fricatives. There are no diphthongs. Thus, all possible consonant clusters are as follows: Various rules regarding allophones:. One common approximation of a pronoun in Classical Timavikan is to conjugate the word for "one" into the appropriate word class. No, I don't have numbers figured out yet, so I can't give an example. In modern dialects of Timavikan, pronouns have developed. I purposely put a lot of fricatives in the phonology to make that one of its defining traits. I wanted to make a relatively simple phonology while still alluding to the complicated consonant clusters of many Caucasian languages.
The Dominion War
So I didn't go for anything too outlandish. Everything else is as it is in IPA. If I had paid a little more attention in Phonetics and Phonology I'd be able to describe the actual rules, but I'm blanking so I'll have to describe them. Pronouns I briefly toyed with having grammatical cases, but I ultimately decided against it. However, as with English, pronouns still retain their declensions.
I designed the phonology principally with that in mind. Pretty much just basic Romance sounds, trying to stick to things similar to languages like French, Spanish, and Italian.
- The Legacy of Shandar - Andrea Cernuschi - Google Книги;
- QUEST FOR THE HYDRA (WIZARD OF EARTHS END Book 2).
- The Real World User Guide for ebikes - ebike book 2010.
- Achieving Quality Results in Large Budget Short Term Developmental Projects!
- If God is Good, Why is there so much Suffering and Pain?.
- The Sixth Sense: Accelerating Organizational Learning with Scenarios.
- Net Profit: How to Succeed in Digital Business.
There's not much that's weird about Uesian phonetics, except that diphthongs are exceedingly common. The first 20 above words, spoken aloud. So is this supposed to be a Romance language itself? Have you come up with a method to regularly derive your vocabulary from Latin via sound change? I've thought about making my own plausible Romlang via that method - the results would be very satisfying - but it seems like it would be a lot of work.
Neisii words are constructed in a similar method to Japanese words, using CV syllables. As a result, consonants need to be easily distinguishable from one another, which is why there are so few fricatives. Dental, alveolar, postalveolar and retroflex fricatives all sound similar, as do palatal, velar, uvular and pharyngeal fricatives. I chose to use [f][v][s][z] because they are the most contrasting fricatives and most suitable to the rhythm of Neisii. The wide range of vowel combinations are a necessity in order to provide a large enough number of CV combinations to prevent homophones.
Because Neisii includes and depends upon such a wide variety of vowels and combinations thereof to differentiate meaning, sometimes making it very difficult for non-native speakers to distinguish different but similar sounding words. Here is my post on Session 3. There are also some dialects much fewer, though that retain another fricative: I'm debating whether or not to include pitch accent in the language.
I really want to, but having that on top of triconsonantal roots makes it seem like it's kind of overkill. I have it currently, but if I decide to get rid of it, it will definitely have stress in its place. There's also a ton of allophony, too much to get into in a brief post such as this, though I thought that it would make the language sound really unique, I guess. I really love retroflexes, though, which is why I have a whole series of them. Oh, here's where I was supposed to put all the stuff I mentioned earlier.
One of the things I'm trying out is having completely different phonotactics at the phonemic and phonetic levels, but I'm not sure it's going to work out because with a triconsonantal language, the words are getting pretty unwieldy. Other allophony mostly involves consonants at the ends of words being reduced the aspirated stops being reduced to voiceless fricatives, the ejectives being reduced to voiced stops, stuff like that. One other thing that I seem to always do in conlangs also shows up here: There are certainly dialectal variations briefly touched upon in the phoneme inventory section , but I haven't figured them all out yet.
I know there are going to be three groups of dialects which I've creatively named "Northern", "Southern," and "Eastern". In one of the groups, geminate consonants don't exist and final short unstressed vowels are systematically elided. So as you can see, this language is heavily root-based in deriving new words yay triconsonantal root language , which is why I have a bunch of random rarer words in my dictionary right now, and no words for like colors and food and family members and stuff.
As such, you're kind of getting a glimpse of some bits of morphology, that will be covered in later sessions I'm sure. The extent of my pronoun system is possessive suffixes that are attached to nouns. They divide for 2 numbers and 3 persons, like the standard European language. They are, in order of 1st sing - 3rd sing, 1st pl - 3rd pl: I don't have anything actually written down, but I know exactly how the script is going to work.
It's going to be a combination orthography, similar to Japanese, in that any word that is composed of a root and a "harmony" what they call the "root fillers" in the language is represented by a single character, and any other information is represented by a syllabary. The characters aren't really going to be truly logographic or anything like that, except for the base forms. The derived forms will be the character of the base form in various shapes or with certain meaning radicals. The syllabary is pretty self explanatory, I think. I couldn't pronounce this language to save my life can't manage ejectives that well , so I'll spare you all recordings.
The problem with this is obviously compounds, for which I have the following solutions:.
These are supplemented with occasional lone vowels or concluding ''n''s or ''s''s. Phonically, both voiced and unvoiced consonants are allowed in i, however, the phoneme inventory is voicing-invariant, meaning 'pa' and 'ba' are the same phoneme. This also applies to voicing onset time VOT , and aspiration differences. Along the same lines, rhotics and laterals are classified as the same phoneme. By allowing a wide range of phones to correspond to the same phoneme, any machine learning algorithm will be more capable of effectively and rapidly classifying incoming audio streams.
Presumably i will evolve over time, changing as it is used, and as classification algorithms begin to develop categories for different phonemes in connection with their users. I have been fighting with the formatting of the tables, so some of them might be a bit off. Let me know if you see something wrong. If grouped with another consonant, they will match the voicing of that constant; otherwise, they tend to be unvoiced.
So, I think it probably has a slightly boring phonetic inventory. But, I think it will be good enough for some fun. At the moment, I plan for just using the latin alphabet to transcribe the language. Maybe I'll get frisky and create one later. Length is contrastive in vowels and consonants, considering having vowel harmony as well but I will wait to see how that might interact with the noun class system to decide for sure. Trying to keep it fairly simple because it's supposed to be spoken by very diverse group around the world with many native languages, but the people that learn it take it pretty seriously and most develop a very good accent.
Lots of dialectical variation though amongst those who don't. I'll come back to add my 20 words, pronouns and greetings when I've got time, but finals are looming at the moment. This tendency appears to colour some parts of their vocabulary. It is suspected that there is another layer of contextual sounds that fall outside regular phonemic description, but these appear to be very difficult to create without the correct physiology. These are comprised of a number of short grunts, exhalation, and inhalation. This basically makes it nearly impossible for a non-Traaz to speak with native fluency, although native speakers will be able to understand what is being conveyed, the subtleties will be lost.
It's safe to assume that any ' that follows a vowel is a glottal stop. The only exception to this rule is OO', due to how it can be followed by further vowels. However, all vowels are ended pretty short; ' merely ends them quicker. An attempt at decreasing the number of consonants from 20 to Vowels however, have almost doubled in amount due to the glottal stop. This is for Kaujasas. The phoneme inventory is based off of Finnish constraints, so very few things will be found here that won't be found in Finnish.
The consonant inventory is sparse, and voiced plosives are almost completely absent. This impacts how loan words are imported. Consonant clusters are heavily restricted: They are sorted by unrounded and rounded as well. Unlike in Finnish, there is no real vowel harmony; diphthongs however are impacted by vowel harmony.
The inventory is straight forward. The only series that has secondary articulation are the central plosives, which include: Other than that no other secondary articulation is phonemic. Although, vowels are lengthened when they are under primary stress. So from the get-go I wanted ejectives in my consonants because i pulled a lot of inspiration from Navajo. I ended up spamming my inventory with a ton of ejectives and after researching Cree, I decided to pull back on the ejectivity and go for a more minimal inventory.
You might notice that Ru doesn't have 1 complete series and that was intentional because I wanted Ru to seem alien because it is missing certain consonants rather than containing several "anal-attentive" "hard-to-prounounce" sounds. Other Remarks Ru has a vast system of morphophonology. Most consonants are not allowed to come into contact with each other. The only exceptions are the combinations: In most cases when two consonants come into contact two things may occur. The frequency of these changes is not regular however the changes themselves are quite regular.
However, there are exception. Ru vowels also have a tendency to form glides or coalesce when they come into contact with another. A small detail but this allophone plays a big role in Ru's phonetic history. Orthography The orthography for Ru is straight forward as well. All letters correspond to their IPA equivalents except for: A unique aspect of Ru is that allophones are accounted for in the orthography. Small Lexicon Ru is a verb heavy languages with most nouns being derived from verbs or verbs themselves. Either way here is a list of the first 20 words I developed for Ru:.
Consonants in parenthesis are allophones to other consonants and the ones marked with asterisks are marginal phonems. I chose the vowels i'm used to as a Spanish speakers as it is my first conlang. For the consonants I did quite a mixture, but basically I picked the ones that sounded the best to me. Even though dipthongs aren't rare, long vowels are, and most of them come from an ancient glotal approximant h lost within time. The list of allophones if mostly nasals, and there explanation is the following:.
Pronouns are declined as nouns except for the first person , i'll leave hear only the singular nominative case as we'll go through declention later. Okay — I am finally ready to post my phonology. Been working on it for a few weeks now. I wanted Aranuen to sound like the sounds of nature.
It emerged gradually from the sounds of wind and leaves and water. So I tried to select sounds that could best emulate these sounds to me. The consonants are simple — basic plosives with no too outlandish sounding phonemes. Fricatives are also mostly straightforward, though augmented slightly with the inclusion of bilabial fricatives. The movement of air is a very basic part of the sound of nature, and so Aranuen has more sounds to represent this movement than some other languages.
There are few approximants, and fewer nasals. The vowels do not show a lot of logic in their selection, other than being some of the easier to pronounce options. Most of the vowels I chose are sounds that you learn to produce very early in life. They can work as standard consonants, or be sustained to provide a different sound.
This only applies to the aforementioned phonemes — fricatives cannot be used this way. For now, I just want to catch up on the remaining challenges, and then I can always come back. Now if someone could help me with the IPA. This is a bit unclear but I'd appreciate if someone could find some IPA symbols for me. The letter combination is how the sound will be written, but isn't supposed to describe the sound exactly.
Here is the phoneme inventory for Reshuv. There are no dental sounds though there is a column. The phoneme inventory of Reshuv is intended to be very similar to that of Hebrew. So, I took the consonantal inventory of Hebrew and adopted it for this language. I am chiefly aiming for a natural sounding language. I haven't started on vocabulary yet. So, I will edit this post later, once I've made a vocabulary. Reshuv uses the Hebrew alphabet. This assignment of letters and sounds will be the standard writing system of Reshuv. Vowel marks can be used to indicate vowel sounds, but these are rare.
As you may have noticed, there are a lot of palatal pairings in this phonology. Palatalization does not spread through the cluster. As I understand Turkish uses the circumflex sometimes to force a palatalization of the preceding consonant; I had put a circumflex in the language name and wanted to justify it, and that worked out nicely. Only the two nasals for ease of writing i. Soramentish phonology is mostly inspired by European languages with a few palatals thrown in. The vowel system in particular I pretty much completely ripped off from German a few days ago, but I'm planning on altering it significantly over time.
Honestly I mostly just picked this phonology because I like the way it sounds, and as it's my first conlang I was trying not to do anything too outlandish. Pronouns are distinguished for three persons and three numbers. First-person dual and plural pronouns are distinguished by clusivity- you and I; this person and I but not you; you and I and these people; these people and I but not you. Third-person pronouns are distinguished by animacy and have both proximate and obviative forms- the proximate form is assigned first, and when another referent of the same number and animacy is added to the discussion, it is assigned the obviative form.
This helps cut down on "he said to him and he had seen him"-type obscurity. Finally, I've been attempting to come up with a script since even before I started work on the language, but it always comes out either hideous or far, far too Tengwar-y for my liking, so I end up scrapping it and starting over. Oh, it's just one of those curved lines you can put over diphthongs and affricates and what have you.
Words consist of any number of equal syllables.
- Community Guidelines.
- WindSwept Narrows: #24 Spring.
- ?
- tandjfoods.com: Sitemap!
- Ach so fromm from the opera Martha - Score!
Syllables have a strict SVC structure with the letters meaning this:. Most consonants are produced according to IPA. The fact that all syllables have to end in a consonant with the next syllable starting in one too, gives the language a rather hard sound. This opens the possibility for a unambigious dialect which produces them unvoiced. Also [r] has no phoneme on its own. Other dialects might place it differently. For the syllables symbols are composed of parts similar to hangul.
These are then layed out top-down left-to-right or alernatively left-to-right top-down. The words are seperated with lines but there are no other punctation marks. This image shows how the symbols are composed. Inspiration I wanted to create something that resembled European languages.
So no tone, clicks, ingressives, ejectives, voiceless nasals, unfortunately. I am trying to do something with these, but so far just as allophones in certain dialects. Allophones In some reduplication morphemes, it is required to change the vowel. There are several of these vowel changes: Alphabet I might upload the alphabet, but I don't have a fancy phone with camera. There is both a tonal and length distinction, leaving 4 varieties of each vowel: Long high, long low, short high, short low. There are no diphthongs aside from when semivowels act as the consonant.
The glottal plosive is not marked since it can only occur word initially and is often fortified later. There is no gemination of consonants. This is how the lateral fricative ended up in there and why there is a high central vowel. This is also why I made it tonal. It wouldn't be fun without these wrenches because then I would have nothing to tool with haha. Short vowels are subject to a shift in certain environments.
In these environments, the vowels shift in this way:. For short high vowels, this occurs when it is directly preceded by a long high syllable. This shift does not occur across word boundaries. It can be most clearly observed when this situation occurs at the end of a word. For short low vowels, which are generally the weakest vowels in the sonority hierarchy, this reduction occurs whenever it follows a high vowel.
The Dominion War | Memory Beta, non-canon Star Trek Wiki | FANDOM powered by Wikia
In careful speech, it will not happen word initially, but in faster speech it is common to hear this lenition even word initial for, especially very common use words like pronouns. Tone sandhi is a process in quick speech to make long vowels more distinct and to facilitate fast, clear speech. If a long vowel is followed by a syllable with a different tonic height, the long vowel will dip or rise slightly in tone towards the end of its timing in anticipation of the tonic step.
Short vowels do not do this, thus illustrating the sharp contrast in the two levels possible. Due to this, it is possible to have complete sentences that have a soft almost musical gliding up and down. Low vowels that do not have an upswing as described above also have a secondary effect. They tend to lower the overall tone of the utterance slightly. After a long utterance with enough low vowels of this nature, the actual frequency of high tones might end up being lower than what the low tones were at the beginning of the utterance.
This is possible because the two tonal levels are realized in contrast to each other. However, if a similar thing were to happen with high vowels, the utterance would not raise in pitch, but stay roughly the same. To account for this, some words with high vowels will cause a drastic rise in pitch to occur so that the pitch of the utterance might remain pronounceable.
This is most likely to occur between sentences, or at the end of various clauses, though it does not always occur. If a noun clause is close enough to the end of the utterance, the sharp upswing will probably be ignored. Long vowels are more likely to cause this to occur, and almost always the upswing will occur across a word boundary.
For it to happen midword would indicate that not only should the word be split into two, but that the before and after are potentially in different elements in the sentence. This upswing can also be induced by a mid sentence pause if it should be needed. Though not mandatory, due to effects of tonal drift, certain statements will follow slightly different tonal patterns than others even though function words may otherwise denote the type of comment. For this we must understand the tonal levels. There are 9 tonal levels, named , where 9 is the high frequency end, and 1 is the lowest.