For this conlang I wanted to see if I could build a language with no unbound roots; I wanted no words to ever stand on their own, but always to require affixation. In addition I wanted a special kind of affixation that wouldn’t just be attaching one morpheme to another, but would have an interesting interleaving quality.

The name of the language, Saonstat, means “the speaking tongue”. “Speak” is ostat and tongue is san. When two morphemes are combined, the final syllable of the first morpheme is combined with the first syllable of the second morpheme to create a new syllable in between them: saonstat. Another example: “speaking man” would be ostat+lenan, ostlaetnan.

Eventually I grew to like it enough that I worked out a bit of history and decided to use it for the Last Starry Night universe. But for a bit of a whimsical version of the backstory and a detailed description of the language, read on.



The single source for the language “Stlond” is a single ream of papers, tied together with twine, dating approximately to 1250 AD, written in rather vulgar Latin interspersed with stretches of 13th century French idiom. The author claims to be one Claude d’Bisque, a Frenchman of Basque ancestry, and purports to record his journey to a previously undiscovered island in the Atlantic, about two day’s sail west of France. The papers describe in tedious detail the particulars of how d’Bisque came to set out on the voyage, the Basque cod fishermen who guided him, the rather unremarkable ways of the inhabitants of the island, and the difficulties encountered on their attempt to leave. This is followed by a lengthy appendix relating details of the language of the island, which is of considerable interest, since its syntactic system is unique in the world. 

It has frequently been asserted that d’Bisque invented the language, as well as the island and the whole history related in his document; but this seems unlikely, for the following reasons. First, the linguistic appendix is written in the same handwriting and phraseology as the rest of the book, showing that d’Bisque is almost certainly the author. However, d’Bisque was no linguist, as shown by the corrupted Latin which he wrote, and the difficulty he had in describing the language Stlond. The appendix consists of long word lists, lengthy passages of transcribed speech with margins filled with attempts at translation, and some completely misguided notes on grammar. For example, d’Bisque asserts that Stlond was like Latin, in preferring to place the verb last in the sentence; but this is certainly not the case in the examples he gives. It is inconceivable that d’Bisque could have constructed this language himself from whole cloth.

In this monograph I attempt to set out an orderly description of the language. I am no comparative linguist, and cannot hope to properly place it in the taxonomy of the world’s tongues; but I hope that this description will assist some other scholar to give it the attention it deserves.

The Voyage of d’Bisque

It may interest some to hear a brief summary of d’Bisque’s voyage and description of the “terra Tlandae”, as he called the island. In the early 13th century Basque fishermen had made long journeys into the Atlantic in search of cod, highly valued for its oil and meat. They brought back many strange tales of their adventures. D’Bisque hired some of these fishermen to take him to an island they had told of; the purpose of his journey was curiosity and a “caesarum cardorum”, perhaps a reference to a broken heart. The islanders, it was told to him, were magicians and could do wonderful things. D’Bisque brought along his sons Francis and Gerard.

The voyage lasted two days, and at the end of that time they came to a single island rising from the middle of the Atlantic like the top of a mountain. On arrival at shore they were greeted by farmers, who expressed wonder at their seaworthy craft. These people, it seemed, never left their island. The Basque fishermen knew a little of the language of terra Tlandae and d’Bisque was fascinated by it; as noted above, he compiled extensive word lists and examples of sentences. The islanders could indeed do some magic, according to d’Bisque; they could call animals to them, move small objects with their minds, and see into the future. 

After spending a few days among the islanders, d’Bisque expressed an interest in climbing to the top of the mountain. The islanders warned him strongly against it, saying that it was the realm of very powerful magicians; but d’Bisque and his sons climbed to the peak anyway, believing that God would save them from heathen magic. At the top of the mountain, which took a day or so to climb, they saw no people. On their return to the village, the islanders refused to speak with them.

They then took their leave, and sailed back toward France. But almost as soon as they set out, d’Bisque’s son Francis fell ill, speaking feverishly of a great king that was calling to him in his mind, and of a lavish feast the king offered him, along with many other earthly pleasures. As the voyage went on, Francis spoke with wild eyes of birds of iron, and horses with bellies of fire, and great underground cities. At last Francis began to speak in Stlond, and babbled furiously in that language for many hours. At last he died, just as they reached France.

D’Bisque ends this rather tedious account with speculation concerning terra Tlandae – specifically, that it is all that remains of the continent of Atlantis, and that the magicians of that great land live under the ground of their submerged continent. He closes with a prayer that God punish the magicians of that land for their wickedness.

Overview of Stlond

The first and most striking thing noted by the student of Stlond is that almost all words of all grammatical categories are bound morphemes – that is, words cannot stand alone, but must be morphologically bound to other words. This rule applies to all grammatical categories except adverbs, interjections, and some elements of the determiner class. Other classes of words are somewhat like the consonantal and vowel roots of Arabic, in that they consist of sequences of consonants and vowels that must be interleaved to create a complete word. For example, the noun “dog” is the consonantal sequence taid; it cannot appear alone, but must combine with a verb such as nonan, “run”, and an aspect infix (e.g. -a-, perfective), to create tnaaiodnan, “the dog ran”. The following table elaborates on this example:

The dog rantnaaiodnan
The dog is runningtnoaiodnan
The dog will runtniaiodnan
The dog atetaaiedt
The dog slepttaaiodsliit
The man ranlenaaonan
The man atelenaaet


Phoneme inventory. Knowledge of Stlond’s phonology is necessarily limited by the orthographic representation used by d’Bisque. The language appears to have a small inventory of vowels: i, e, o, a, u. It is unknown to what extent vowel quality may vary according to environment. Sonorants m, n, r, and l may appear as vowels. Stlond does not have “ng” or “h”, or any voiced fricatives or affricates. It also has almost no coronal fricatives — only “ch”. However, it does have the voiceless velar fricative “x”, written as “hh” by d’Bisque. D’Bisque also appears to use “c” and “q” for “k”: “c” before o, a, u, and “q” before i and e. I have followed his orthography, except to use “x” for “hh”.

Clusters. Stlond admits a number of consonant clusters in both onset and coda position which English does not. In general, the language admits any onset cluster of the forms:

  • (voiceless stop)(fricative)
  • (voiced stop)(fricative)
  • (voiceless stop)(sonorant)
  • (voiced stop)(sonorant)

Codas allow the same clusters, but in reverse order. For example, the onset cluster “tl-“ is permitted, as in the noun tloc, “chowder”, and the coda cluster “-lt”, as in the verb olt, “scamper”. (This latter may be related to the noun lot, “mouse”.) Sonorants may in general not cluster, except –rn, -rm, -ln, and –lm. Other apparent clusters of sonorants should be read as vocalic sonorants. For example, nlod, “marriage”, should be read with two syllables “n-lord” in the word nlord, “of marriage”.


A note on some peculiarities of morphology are in order.

Predicatization. Nominals may be converted to predicates, either to create a new lexeme (analogous to the history of the English verb “access”), or to permit the nominal to be used attributively (see Copula Statements below). The predicatization process inserts -o- between the consonant clusters of the nominal (note that -o- is the stative aspectual marker) and prior to the first cluster. Thus the nominal lenan, “man”, may be predicatized to -olenan, meaning “to man” or perhaps “manly”, as in plo:ienan, “that boy is a man.”

Argumentization. Predicates, such as adjectives or verbs, may be converted to arguments, either to name a new object whose purpose is to perform the action, or to name the action or property itself. Argumentization reduplicates the final consonant of the predicate. Thus the verb bacar, “poke/impale”, may be argumentized to be a noun, bacar-r, “impaling”. Thus one can say bacarversos pmiotar, “spearing of fish is easy”.


In the Stlond sentence, a subject combines with a verb and an aspect infix morphologically to create a complete sentence. Modifiers to the subject generally appear to the left of the verb; modifiers to the verb appear after. Objects of the verb appear after the verb, generally marked with an adposition affix; modifiers to these objects appear after them. Word order is thus fairly strict. The above generalizations apply to simple declarative statements and wh-questions (which are in-situ). Other categories have more complex forms:

Polar questions. Here, the main verb is combined with an expletive nominal wim and moved to the beginning of the sentence. The subject of the sentence appears immediately after the verb, marked with an adposition or. For example, “Is the man eating?” would be Woemt lenaorn?

Copula Statements. In cases in which two nominals are equated, as in “Tlond is an island”, the second nominal undergoes a morphological process to change it into an adjectival element (see predicatization and argumentization below). In this case, the noun andad, “island”, is predicatized into -oandad and the sentence is Tlo:andad.