created, $=dv.current().file.ctime & modified, =this.modified
tags:languagegrammarlinguisticsaiislands
NOTE
This is a note for exploration of this concept. It is likely embarrassingly uninformed, and novice. I can only hope through my independent observation I converge on something of interest.
I think I am just trying to see what differences exist and could be explored, between language, spoken speech and writing.
I was performing ML assisted transcription using a few different models. I’ve encountered certain limitations with the models, which has made me think about how they would represent human speech.
- a model that cannot capitalize, producing just a stream of text of words detected.
- a model that cannot identify speakers (diarization)
- a model that is monolingual
Comparison of two models, one which produces results like this
Speaker 0: Okay. Thank you very much. And like everyone else, I’d like to thank the organizers for, first of all, putting together this meeting and for inviting me to participate.
What if I abandon identification of speakers? whatifijustproduceastreamofwords
My spoken language (what these transcriptions are interpreting), doesn’t include sentence start capitalization, which is also omitted in this model. (I could see all caps “HELLO” inferring speech volume.) Capitalization norms have also shifted during time, as 18th century writers would capitalize nouns. When this practice shifted the language itself did not change.
Illiterates will manage to speak a language without the use of these elements.
Most of human history existed without written language.
If the ML transcription reads: If, it takes my speech, to be, like this. If, it takes my speech to be, like this.
Written (here, type on a screen) and spoken language are not equivalent, so what is this act of transcription?
Is there any support for writing play any role in human language?
Speech is related to language, is related to writing.
Unbroken song
I remember as a kid listening to songs and wondering to myself if all of the songs that I encountered, or sought out in my life constituted a single song. In other words, saying to myself that all song experience is one song, and that gaps between music is arbitrary. In this view, listening to an entire album can be grouped into a single song (music being listened to).
It’s a bit nonsensical but it is a romantic thought, to think of this song ever sang, persisting throughout your life.
NOTE
After writing this I found this touched on in The Art of Syntax by Ellen Bryant Voigt She says “Like the musical measure, the poetic line is inherently artificial, imposed by the poet onto the language.”
Orthography
Capitalization is part of orthography, which is conventional.
Most national and international languages have an established writing system that has undergone substantial standardization, thus exhibiting less dialect variation than the spoken language
Most natural languages developed as oral languages and writing systems have usually been crafted or adapted as ways of representing the spoken language. [Orthography] often refers to any method of writing a language without judgment as to right and wrong, with a scientific understanding that orthographic standardization exists on a spectrum of strength of convention.
Orthography encompasses everything from the alphabet to spelling to capitalization to punctuation, and is certainly a part of language usage. But whether it’s an intrinsic part of the language itself is up for debate.
Do sentences exist in a language?
What if I view a language as this unbroken song above? A sentence seems to be produced through writing, and punctuation.
Sentence: A word of set of grammatically linked words expressing a complete thought.
which seems incomplete.
Also…
Utterance - a unit of speech typically with silence on the part of a speaker as boundaries.
Sentence - a unit constituting an utterance and intended to correspond to a meaningful complex structuring an idea, determined on many different levels requiring to structure one’s speech into finite units due to our physiological and cognitive limitations.
Syntactic clause/phrase or Speech clause - depending further on perspective assumed a constitutive unit of sentence.
Warlpiri Language
This language has free word order “speakers of which are reported to not even be able to repeat sentences back with the same word order.”
Totally confused about this, however - imagining it like a bag. It oddly places more emphasis on the sentence? You have a jumble of words with the bounds of comprehension being a language.
Similar to Kalkatungu , Warlpiri exhibits very free word order on the clause level , and thus a sentence consisting of subj ect obj ect-verb-instrument-time constituents could result theoretically in 120 di fferent combinations .
Avoidance register Jokes - Control Language - Islands
In Warlpiri culture, it is considered impolite or shameful for certain family relations to converse. (For example, a woman should not converse with her son-in-law.) If such conversation is necessary, speakers use a special style of the language, the avoidance register. The avoidance register has the same grammar as ordinary Warlpiri but a drastically reduced lexicon. Most content words are replaced by a generic synonym or a word unique to the avoidance register.
The configurational structure of a nonconfigurational language
There is a dissertation that undermines the idea of Walpiri as a no-configurational language. She seems to be refuting her advisor? Hale who performed the initial study.
problematic to the claims that Walpriri has a flat syntactic structure.
properties included free word order.
Warlpiri has been seen to require adding additional parameters into the typological space of human language. In this paper, I suggest that such a move is unnecessary
English Sentence Structure
>English has mostly 10 sentence patterns?
Counter-examples seem to have poetic speech, deliberate evasion. I think I might have to delve into impossible languages to express my love because these structures scare me. Sure there is variation from the root, but how many roots?
Subj+be+adverbial of time/place Adverbial is an umbrella term which covers all adverb forms, whether single forms, phrases, or clauses.
Subj+be+adjectival Adjectival is an umbrella term which covers all adjective forms, whether single words, phrases, or clauses.
Subj+be+nominal Nominal is an umbrella term which covers all words that function as a noun, whether single words, phrases, or clauses. Linking verb sentences…
Subj+linking verb+adjectival
Subj+linking verb+nominal Intransitive verb sentences…
Subj+VI VI = verb intransitive Transitive verb sentences…
Subj+VT+DO VT = verb transitive; DO = direct object
Subj+VT+IO+DO IO = indirect object
Subj+VT+DO+adjective (object complement)
Subj+VT+DO+noun (object complement)
The students are here.
She is in a bad mood.
The astronaut is an old man.
The students seem diligent.
The students became scholars.
The students rested.
That car needs new tires.
The teacher made the test easy.
They named their dog Oscar.
The judge awarded Mary the prize.
Sentence Length
The textbook Mathematical Linguistics, by András Kornai, suggests that in “journalistic prose the median sentence length is above 15 words”. The average length of a sentence generally serves as a measure of sentence difficulty or complexity. In general, as the average sentence length increases, the complexity of the sentences also increases.