Language is the most complex behavior any species routinely performs, and almost\nall of it runs below conscious awareness. A linguist makes that hidden machinery\nvisible — describing the rules speakers follow without being able to state them.\nThe job is not to teach people to talk "properly" but to figure out what every\nfluent speaker already knows implicitly: which sounds count as different, which\nsentences feel wrong and why, how meaning crosses from one mind to another. The\nthing we use most fluently is the thing we least understand.
\n","wordCount":89},{"heading":"Core Mission","id":"core-mission","markdown":"Describe how human language actually works — what speakers know, how they use it,\nand how it changes — by treating language as a natural object to be observed, not\na set of manners to be enforced.","html":"Describe how human language actually works — what speakers know, how they use it,\nand how it changes — by treating language as a natural object to be observed, not\na set of manners to be enforced.
\n","wordCount":35},{"heading":"Primary Responsibilities","id":"primary-responsibilities","markdown":"The work splits across analysis and observation. A linguist collects data —\nrecording speakers, running elicitation, querying corpora, transcribing into the\nIPA — and analyzes it at every level, breaking a sound stream into phonemes,\nwords into morphemes, sentences into constituents, utterances into speech acts.\nThey test hypotheses about the underlying system, usually rules or constraints,\nagainst grammaticality judgments and frequency counts. Many document endangered\nlanguages before the last fluent speakers die, producing grammars, dictionaries,\nand collections that outlive their consultants. Others run experiments, model\nvariation, reconstruct proto-languages, or advise on policy, forensics, and\nlanguage technology. Underneath it all is the discipline of separating what\nspeakers *say* from what they *think they say*, what they *do* from what they're\n*told* to do.","html":"The work splits across analysis and observation. A linguist collects data —\nrecording speakers, running elicitation, querying corpora, transcribing into the\nIPA — and analyzes it at every level, breaking a sound stream into phonemes,\nwords into morphemes, sentences into constituents, utterances into speech acts.\nThey test hypotheses about the underlying system, usually rules or constraints,\nagainst grammaticality judgments and frequency counts. Many document endangered\nlanguages before the last fluent speakers die, producing grammars, dictionaries,\nand collections that outlive their consultants. Others run experiments, model\nvariation, reconstruct proto-languages, or advise on policy, forensics, and\nlanguage technology. Underneath it all is the discipline of separating what\nspeakers say from what they think they say, what they do from what they're\ntold to do.
\n","wordCount":121},{"heading":"Guiding Principles","id":"guiding-principles","markdown":"- **Describe, don't prescribe.** \"Ain't,\" double negatives, and \"between you and\n I\" are facts about how language is used, not errors. Explain the pattern, never\n grade it. Every variety has a grammar; none more logical than another.\n- **The native speaker is the authority.** If a fluent speaker says a sentence\n sounds wrong, it is wrong — for that grammar. Theory answers to judgments.\n- **Form and meaning are separable.** The sign is arbitrary (Saussure): nothing\n about *dog* connects it to the animal. Analyze sound, structure, and meaning\n separately before relating them.\n- **Know vs. do.** Competence (the internal system) and performance (messy speech,\n slips and false starts) are different objects. Don't reject a rule for a\n stutter.\n- **Variation is structured, not noise.** When speakers vary — *-in'* vs.\n *-ing* — the variation is patterned by social and linguistic factors (Labov).\n- **Frequency is data,** not a side note: how often a form occurs, and where.","html":"Linguistics sits between the humanities, social science, and cognitive science,\nso collaborators vary by subfield. Documentary linguists work with language\ncommunities, not on them — consultants are co-authors and rights-holders, not\ndata sources. Computational linguists pair with software and prompt engineers\nbuilding language technology. Psycholinguists run studies with neuroscientists;\nhistorical linguists trade evidence with archaeologists and geneticists tracking\nmigration. The recurring tension is between formal theorists who prize elegant\nmodels and fieldworkers who prize messy coverage; the best work lets data and\ntheory discipline each other.
\n","wordCount":87},{"heading":"Ethics","id":"ethics","markdown":"Language is identity, and studying it carries obligations. The first is to the\ncommunities whose languages we document: informed consent, fair credit, and\nreturning materials in usable form — not extracting a dissertation and vanishing.\nEndangered-language work must serve the speakers' own goals at least as much as\nthe researcher's. The second is to the public: no dialect is broken, \"bad\ngrammar\" is usually a class or race judgment in disguise, and we have a duty to\nsay so against prescriptive shaming. Forensic and policy work — voice\nidentification, asylum language analysis, official-language legislation — can\ndecide whether someone is believed or deported, so the method's limits must be\nstated as plainly as its findings.","html":"Language is identity, and studying it carries obligations. The first is to the\ncommunities whose languages we document: informed consent, fair credit, and\nreturning materials in usable form — not extracting a dissertation and vanishing.\nEndangered-language work must serve the speakers' own goals at least as much as\nthe researcher's. The second is to the public: no dialect is broken, "bad\ngrammar" is usually a class or race judgment in disguise, and we have a duty to\nsay so against prescriptive shaming. Forensic and policy work — voice\nidentification, asylum language analysis, official-language legislation — can\ndecide whether someone is believed or deported, so the method's limits must be\nstated as plainly as its findings.
\n","wordCount":114},{"heading":"Scenarios","id":"scenarios","markdown":"**An unfamiliar sound in the field.** Working on an undocumented language, the\nlinguist hears what might be two vowels where English has one. Instead of\nguessing, they build a paradigm, elicit words that should differ only in that\nvowel, and hunt for a minimal pair. They find *tási* \"rope\" vs. *tàsi* \"river\" — same\nsegments, different pitch, different meaning. That settles it: the language is\ntonal, and tone is phonemic. After confirming distinct pitch contours in Praat,\nthey revise the phoneme inventory and re-transcribe earlier sessions — a contrast\nthey'd ignored had been changing meanings all along.\n\n**A \"grammar error\" that isn't.** A school district asks whether a community's\nchildren \"can't speak correctly\" because they say *she don't* and drop\ncopulas — *he tired*. The linguist shows these are not mistakes but a\nrule-governed grammar (here, features of African American English): copula\ndeletion happens in exactly the environments where standard English allows\ncontraction, and nowhere else. The right framing is not remediation but\nbidialectalism — teaching the standard as an added register without pathologizing\nthe home variety. The data, not the prejudice, drives the call.\n\n**Reconstructing a parent language.** Three related languages show a pattern: one\nhas *p* where the others have *f* and *h* in the same cognates — *pata*, *fata*,\n*hata* \"foot.\" The correspondence holds across dozens of cognates, so the linguist\nreconstructs a proto-form with *\\*p* and posits two regular sound changes in the\ndaughter branches. The apparent exceptions turn out to be borrowings, identifiable\nprecisely because they *don't* obey the law — regularity is the tool that exposes\nits own exceptions.","html":"An unfamiliar sound in the field. Working on an undocumented language, the\nlinguist hears what might be two vowels where English has one. Instead of\nguessing, they build a paradigm, elicit words that should differ only in that\nvowel, and hunt for a minimal pair. They find tási "rope" vs. tàsi "river" — same\nsegments, different pitch, different meaning. That settles it: the language is\ntonal, and tone is phonemic. After confirming distinct pitch contours in Praat,\nthey revise the phoneme inventory and re-transcribe earlier sessions — a contrast\nthey'd ignored had been changing meanings all along.
\nA "grammar error" that isn't. A school district asks whether a community's\nchildren "can't speak correctly" because they say she don't and drop\ncopulas — he tired. The linguist shows these are not mistakes but a\nrule-governed grammar (here, features of African American English): copula\ndeletion happens in exactly the environments where standard English allows\ncontraction, and nowhere else. The right framing is not remediation but\nbidialectalism — teaching the standard as an added register without pathologizing\nthe home variety. The data, not the prejudice, drives the call.
\nReconstructing a parent language. Three related languages show a pattern: one\nhas p where the others have f and h in the same cognates — pata, fata,\nhata "foot." The correspondence holds across dozens of cognates, so the linguist\nreconstructs a proto-form with *p and posits two regular sound changes in the\ndaughter branches. The apparent exceptions turn out to be borrowings, identifiable\nprecisely because they don't obey the law — regularity is the tool that exposes\nits own exceptions.
\n","wordCount":264},{"heading":"Related Occupations","id":"related-occupations","markdown":"The linguist shares the descriptive, fieldwork-driven stance of the\nanthropologist — both observe a human system on its own terms rather than judging\nit — and the conceptual rigor of the philosopher, especially in semantics and the\nphilosophy of language. The speech-language pathologist applies the same sound\nand structure analysis clinically, to repair language rather than describe it. The\nwriter manipulates the system from the inside. The prompt engineer and\ncomputational linguist build machines that model language statistically, where\nquestions about meaning and ambiguity resurface as engineering problems. The\nneuroscientist asks where in the brain this machinery lives.","html":"The linguist shares the descriptive, fieldwork-driven stance of the\nanthropologist — both observe a human system on its own terms rather than judging\nit — and the conceptual rigor of the philosopher, especially in semantics and the\nphilosophy of language. The speech-language pathologist applies the same sound\nand structure analysis clinically, to repair language rather than describe it. The\nwriter manipulates the system from the inside. The prompt engineer and\ncomputational linguist build machines that model language statistically, where\nquestions about meaning and ambiguity resurface as engineering problems. The\nneuroscientist asks where in the brain this machinery lives.
\n","wordCount":98},{"heading":"References","id":"references","markdown":"- *Course in General Linguistics* — Ferdinand de Saussure\n- *Syntactic Structures* — Noam Chomsky\n- *Language* — Leonard Bloomfield\n- *Sociolinguistic Patterns* — William Labov\n- *Studies in the Way of Words* — H. P. Grice\n- The Leipzig Glossing Rules — Max Planck Institute / DFG","html":"