Language contact is the interaction of speakers and their linguistic resources, and the changes or patterns of use that can follow. Its gold standard is an explanation that connects an observable feature to a plausible pathway of interaction, checks competing explanations and states what the evidence cannot establish. A borrowed word, a bilingual conversation and a community changing its main language are related phenomena, but they require different observations.
This guide helps readers recognise those differences, investigate them and avoid treating multilingual speech as defective. The dialogues, invented language forms and numerical datasets below are teaching constructions, not records of real communities. The Gold Standard master provides the wider collection; the Education parent explains why understanding should survive a change of context.
Choose your language contact route
Chapters 1–3: Identify contact and borrowing
Chapters 4–6: Understand translation and interaction
Chapters 7–9: Investigate structure and history
Chapters 10–12: Measure evidence and test transfer
Chapter index
Identify contact and borrowing · Chapters 1–3
1. Begin with people interacting, then define the phenomenon
2. Map the pathway through which a feature could travel
3. Understand borrowing without imagining a word must be returned
Understand translation and interaction · Chapters 4–6
4. Recognise calques and transferred meanings
5. Analyse code-switching before assigning a motive
6. Separate conversational accommodation from lasting change
Investigate structure and history · Chapters 7–9
7. Look for structural influence beyond shared words
8. Distinguish contact, inheritance and coincidence
9. Treat contact languages as structured languages with histories
Measure evidence and test transfer · Chapters 10–12
10. Calculate rates that answer the question you actually asked
11. Design a small investigation that someone else can follow
Answer questions and continue · Chapters 13–14
CHAPTER 1 OF 14 · Identify contact and borrowing
1. Begin with people interacting, then define the phenomenon
Two languages printed on the same map do not tell us who speaks to whom. Start with an interaction: a customer negotiating a purchase, friends exchanging messages, colleagues discussing a machine, or a child listening to relatives. Identify the participants, their relevant language experience, the activity and the expression being investigated. That gives the inquiry a scale small enough to examine.
Contact linguistics connects several fields. The 2023 Language Science Press volume on language contact places conversational interaction alongside language shift, contact languages and geographical patterns. These are different timescales and levels of analysis. An adjustment during one conversation is not automatically a lasting change in a language.
Give the question a unit and a timescale
Consider an invented observation: a workshop uses a name for a tool that resembles a word used by visiting technicians. At least four questions are possible. Where did the name originate? Who currently uses it? How is it pronounced or combined with other words? Does its use extend beyond this workshop? Each question needs its own evidence.
A recording of today’s meeting can establish an occurrence. Several speakers using the term independently might establish wider distribution in the sample. Older manuals can help investigate chronology. None of those pieces alone establishes the entire history. Write the question before gathering a large collection of examples, otherwise the most memorable example can quietly become the conclusion.
This subject also has boundaries. Historical Linguistics examines change and relationships over time. Sociolinguistics examines language in social life. The particular task here is to explain what interaction between linguistic systems contributes, while retaining those neighbouring methods.
For a first exercise, write three sentences about the workshop: one observation, one proposed explanation and one missing check. A good answer is: “The term occurs in this recording. Contact with technicians may explain its introduction. We need earlier records and evidence of alternative sources.” The distinction prevents a plausible story from becoming a fact merely through confident wording.
CHAPTER 2 OF 14 · Identify contact and borrowing
2. Map the pathway through which a feature could travel
A contact explanation needs more than a statement that two groups lived nearby. Ask how an expression could have reached particular speakers. Possible pathways include repeated conversation, schooling, employment, migration, translated material and media. These are leads to investigate, not a universal ranking of causes. The pathway must fit the dates and people in the claim.
Imagine a fictional harbour with three groups: boat crews, warehouse staff and inland shopkeepers. The crews and warehouse staff meet daily, while shopkeepers mainly read shipping notices. A new expression appears first in notices and then in shops. Direct imitation of crew speech is possible, but printed communication is another route. A diagram with everyone connected to everyone would conceal that difference.
Separate exposure, adoption and spread
Exposure means an opportunity to encounter a form. Adoption means using it in the relevant sense. Spread concerns its appearance beyond the initial users. In the harbour example, a notice may reach a thousand readers while only a few use its terminology. A large audience is not a count of adopters.
Suppose the fictional evidence consists of eight notices and twelve interviews. Record the publication dates, intended readership and distribution of the notices. For interviews, distinguish participants recalling a term from participants using it spontaneously. Prompting someone with a word and receiving recognition is a different observation from hearing it in unprompted talk.
Social categories should not replace this evidence. A person’s birthplace, surname or appearance does not establish their language repertoire, conversation partners or source of a particular expression. Ask relevant questions voluntarily and explain why they matter. A participant can know a borrowed expression without knowing the language from which it historically came.
The useful output is a modest pathway statement: “The notices provide a documented route by which these participants could have encountered the expression before the interviews.” This is stronger than a vague claim about cultural influence, and narrower than a claim that the notices definitely caused adoption. If the interviews predate the notices, the proposed sequence needs repair. Contact research improves when chronology can actually contradict the favoured explanation.
CHAPTER 3 OF 14 · Identify contact and borrowing
3. Understand borrowing without imagining a word must be returned
Lexical borrowing concerns words entering one language’s vocabulary from another. “Borrowing” is a technical metaphor: the source language does not lose the word, and the recipient language need not return it. Yaron Matras’s discussion of linguistic matter explains why incorporation can become established even when speakers cease to recognise the source.
Keep historical origin separate from present familiarity. A word can feel entirely ordinary to its users while having a contact history. Conversely, an unfamiliar-looking spelling is not sufficient proof of borrowing. A learner needs evidence about the form, meaning, chronology and route, rather than an intuition about whether it looks foreign.
Work a fictional chain of transmission
Suppose Language A has the invented word “naku” for a particular container. A dated merchant’s notebook records Language B speakers using “nako” for that container. Later B-language instructions use “nako” with a regular B plural ending. These stipulated observations make adoption from A into B plausible, but they still leave questions. Was the notebook accurate? Could both communities have encountered the term through a third language? Does the earlier A record predate B usage or merely survive better?
Now suppose Language C later acquires the word through B-speaking traders. For the proposed C borrowing, B is the immediate donor; A belongs to an earlier part of the chain. Those labels answer different questions. The World Loanword Database glossary explicitly distinguishes the immediate source from earlier transmission and records degrees of evidence for borrowing.
The answer to “Where did the C word come from?” therefore depends on the intended depth. A short dictionary explanation might name the immediate route and then the earlier origin. A classroom answer should not collapse the two into a direct A-to-C encounter that the records do not support.
Continue with the site’s Loanwords in English guide for vocabulary applications. When checking an actual English word, use a reputable historical dictionary and its dated evidence. The invented “naku” example teaches the structure of the reasoning; it supplies no etymology for any real language.
CHAPTER 4 OF 14 · Understand translation and interaction
4. Recognise calques and transferred meanings
Contact can influence an expression without bringing across its original sound sequence. A calque, or loan translation, renders an expression’s meaningful parts through the receiving language’s resources. This differs from adopting the source form itself. However, the existence of a literal translation does not show that the translated expression became conventional.
The site’s Calques and Loan Translations guide develops that distinction. For analysis, keep three possibilities available: an established expression with a supported contact history, a one-off translation, and an independently created expression that happens to resemble another language’s construction.
Compare three deliberately invented outcomes
Assume a fictional source language calls a portable lamp “night-seed”. A translator has three options for a fictional receiving language. They might adopt a version of the source sounds; translate the two components into the receiving language’s own words; or describe the object as “a lamp carried in the hand”. These choices transfer different aspects of the source expression.
If the component-by-component version appears once in a translated leaflet, we can describe that translation decision. We cannot yet declare that the receiving language has acquired a new conventional term. If independent writers later use it without explanation, that provides a different kind of evidence. If an older receiving-language text already contains the same expression, the proposed direction becomes doubtful.
Meaning can also change while a word’s form stays familiar. In a second invented case, a word previously used for a physical doorway starts appearing in texts about entering a computer system. Contact with another language’s terminology is one hypothesis; a locally developed metaphor is another. To distinguish them, investigate translation dates, users’ working practices and earlier meanings. Do not settle the issue by noticing that both languages use a similar metaphor today.
For a translation task, the practical check is current intelligibility and appropriateness for the reader. A historical borrowing story cannot rescue wording that the intended audience misunderstands. For a historical inquiry, the practical check is evidence of transmission. These two jobs overlap, but successful communication today and a proven origin are not the same result.
CHAPTER 5 OF 14 · Understand translation and interaction
5. Analyse code-switching before assigning a motive
Code-switching broadly concerns using more than one language or variety within an interaction. Researchers differ over the boundaries between switching, borrowing and other forms of mixing. Alexiadou and Lohndal’s research on language mixing illustrates the debate, including mixing within words. Pronunciation alone is not a dependable universal test for assigning every instance to one category.
Begin with what participants actually do. Preserve the order of turns, the words around the switch and the task. A transcript that removes the surrounding explanation can make a quotation, a technical label and an instruction look like the same event. Where the language boundary is uncertain, mark that uncertainty instead of forcing a label.
An invented repair sequence
Consider a fictional bilingual exchange, represented in English with language labels rather than fabricated translations. Person A asks in Language A which valve to close. Person B gives a component label in Language B, points to a diagram and then continues in A. Person A repeats the label and closes the indicated valve in the imagined training exercise.
The visible sequence suggests that the label and diagram helped identify a referent. It does not establish that B lacked the word in A. The label might match the equipment manual, recall previous training or be the term both people normally use. To investigate, compare similar exchanges, inspect the manual and ask participants about their practice without treating retrospective answers as perfect recordings of thought.
An appropriate analysis might say: “The switch occurs at the component label during a referential clarification.” That is a specific claim tied to the sequence. “The speaker switched because Language A is inadequate” is a much larger claim unsupported by the example.
For a classroom discussion, ask what additional evidence would distinguish three explanations: a momentary retrieval difficulty, an established shared term, or quotation of the manual. A useful answer proposes observations that could differ across the explanations. If a student merely repeats “people switch for identity”, prompt them to show where that interpretation becomes visible in this exchange. Social meaning is something to examine, not an automatic caption applied to every multilingual sentence.
CHAPTER 6 OF 14 · Understand translation and interaction
6. Separate conversational accommodation from lasting change
Accommodation involves adjustments in how people communicate with one another. A speaker may move toward an interlocutor’s pronunciation or wording in a particular interaction. The step from such short-term behaviour to community-wide change needs separate evidence. Ruch and de Benito Moreno’s review identifies this connection as an important research question while emphasising methodological variation and remaining gaps.
Imagine a fictional study of speaking rate. Before a paired task, one participant produces 120 syllables per minute and the other 180. During the task their rates are 140 and 160. The difference has narrowed from 60 to 20 syllables per minute, a reduction of 40. Relative to the initial gap, that is 40 divided by 60, or about 66.7%.
Interpret the calculation at the right scale
Those arithmetic results describe the stipulated measurements. They do not show why the rates changed. The paired task may require shorter answers, participants may become more familiar with the material, or measurement windows may differ. A strong study would examine comparable speech and define how pauses, hesitations and syllables were counted.
Now add a delayed recording in the original task. The rates return to 122 and 178. The new gap is 56, close to the initial 60. This fictional result would weaken a claim that the interaction produced a lasting shared speaking rate. It does not erase the observed convergence during the paired task. Both results can be true at their respective timescales.
The repair is to make the conclusion match the design: “In these measurements, the participants’ rates became closer during the paired task; the later recording did not show a comparable persistent convergence.” Avoid calling either speaker’s initial rate defective or assuming that moving closer is always socially desirable.
For learners, this example demonstrates why a before-and-during comparison is insufficient for a claim about lasting change. Add a later observation, compare the same kinds of material and consider alternative causes. These habits transfer to accent studies, vocabulary uptake and classroom practice, where an immediate imitation may be visible long before independent use is established.
CHAPTER 7 OF 14 · Investigate structure and history
7. Look for structural influence beyond shared words
A contact investigation can concern grammar and patterns of meaning as well as vocabulary. Matras and Sakel’s study of pattern replication distinguishes copying linguistic material from reorganising a language’s own resources along an external model. This distinction helps explain why a sentence can contain familiar local words yet still raise a contact question.
Use an invented construction to see the issue. Suppose Language A habitually marks a reported event with an expression meaning “they say”. Language B already has words meaning “people” and “say”, but older B texts rarely use them as a compact report marker. Recent bilingual conversations show that compact use more often. The proposed influence concerns distribution and function, not necessarily a newly borrowed word.
Make the alternative explanation concrete
Before claiming replication, ask whether older B evidence is genuinely comparable. Perhaps the old collection contains legal contracts while the new one contains gossip and personal narratives. Report markers would have different opportunities to appear. A difference in genre could produce the apparent historical trend without any new grammatical convention.
Repair the comparison by looking for earlier and later narratives with similar reporting tasks. Define what counts as the compact marker, distinguish literal statements about people speaking, and inspect examples in context. If comparable earlier evidence is unavailable, say that the historical claim remains provisional. Absence from a small surviving collection is weaker than absence from all use.
Next examine the contact pathway. Were the relevant speakers bilingual? Did they encounter the proposed model before the change? Does the distribution fit that route better than an independent development? Several clues can support an explanation together even when no single clue is decisive. Conversely, one attractive parallel can remain insufficient despite a long discussion.
The useful learning outcome is the ability to separate form, function and frequency. A word may be old while its grammatical role changes; a construction may be old while its frequency rises; a newly recorded form may have existed previously in speech. Each possibility suggests a different follow-up. This is why contact analysis benefits from close grammatical description and historical comparison rather than a collection of look-alike sentences.
CHAPTER 8 OF 14 · Investigate structure and history
8. Distinguish contact, inheritance and coincidence
Similarity is a starting observation, not a completed history. Two forms may resemble each other because they descend from a common earlier form, because one was adopted through contact, because both reflect another source, or because the resemblance is accidental. The Historical Linguistics root develops recurring sound correspondences and the comparative method, which help prevent isolated resemblances from carrying too much weight.
Here is an original evidence puzzle. In two invented languages, five pairs of everyday words show a recurring initial sound relationship: Language A has “p” where Language B has “f”. A sixth pair, naming a recently introduced tool, begins with “p” in both. Assume that the dates and meanings of the five everyday pairs support an inherited relationship. The tool pair is therefore worth investigating separately, but its exceptional sound is not by itself a proof of borrowing.
Build a conclusion that survives a new document
Additional fictional evidence says the tool was introduced through A-speaking traders, the A term is recorded earlier, and B’s first records use the same form for the same object. Together these facts strengthen the borrowing hypothesis. Now imagine discovering an earlier B notebook containing the term. The direct route may need revision, depending on the notebook’s date and reliability. A good explanation makes room for that update.
Do not equate first surviving record with first spoken use. A word written in one language in 1850 and another in 1880 need not have originated in the first language in 1850. The dates establish known attestations. The inference about earlier speech depends on the quality and coverage of the historical record.
The same caution applies geographically. A cluster of similar features across neighbouring languages invites an areal explanation, but shared ancestry and independent developments must also be considered. A modern border does not reconstruct past interaction. Historical movement, multilingual practices and the distribution of several features are more useful than a coloured map alone.
The exercise’s answer is therefore conditional: borrowing is supported by the combined fictional sound, meaning, chronology and interaction evidence; no single part establishes it alone. Learning to combine evidence without pretending it is complete is a central competence of the subject.
CHAPTER 9 OF 14 · Investigate structure and history
9. Treat contact languages as structured languages with histories
Terms such as pidgin, creole and mixed language name research categories with different histories and contested boundaries. They are not grades of linguistic worth. The Atlas of Pidgin and Creole Language Structures compares structural features across contact languages and stresses comparison both among them and with languages involved in their histories. A label does not replace a description of grammar, use and development.
In broad introductory usage, a pidgin is associated with communication across groups without a shared language and initially limited functions; a creole is used as a community language and may be acquired from childhood. The research review Pidgins and Creoles: Debates and Issues discusses different definitions and the importance of social functions. Actual histories are varied. Do not assume that every creole followed an identical, neatly documented sequence or that mixing words in a conversation creates a new mixed language.
Read a real case with dates and scope intact
Alan N. Baxter’s APiCS survey of Papiá Kristang describes a largely Portuguese-derived lexicon with Malay and other contributions, alongside a grammatical history involving several influences. That is already enough to challenge a simplistic picture in which a language is just one source vocabulary with a few additions. Its words, constructions and community history require related but distinct descriptions.
The survey also includes speaker estimates tied to older observations. Those estimates should not be silently presented as today’s population. When using any historical survey, record the observation date separately from the date on which you accessed the page. A webpage still being available does not make every number on it current.
For local reading, the eduKate Baba Malay and Peranakan living heritage article offers a route into language, translation and performance. Read it as a specific community context, not a licence to infer an individual’s language abilities from their cultural affiliation.
An effective museum or classroom activity compares two descriptions: what a source says about a language’s historical development, and what a present participant chooses to say about their own repertoire. Keep both attributed. Neither must be squeezed into the other. This preserves the difference between a research account of a language and the identity or experience of a particular person.
CHAPTER 10 OF 14 · Measure evidence and test transfer
10. Calculate rates that answer the question you actually asked
Counts become useful when the denominator is clear. Suppose an invented set of conversations contains 60 occurrences of a contact-related expression in 20,000 running words. Another contains 90 in 60,000 words. Using the same counting rule, the first rate is 60 ÷ 20,000 × 10,000 = 30 occurrences per 10,000 words. The second is 90 ÷ 60,000 × 10,000 = 15. The larger raw count belongs to the collection with the lower normalised rate.
The conclusion is limited to these invented collections. It does not show which community borrows more, which language is changing faster, or whether the difference is statistically reliable. Genre, topic and speaker selection may explain the contrast. The site’s Corpus Comparison guide provides a fuller route through those design choices.
A change of denominator changes the meaning
Imagine that all 60 occurrences in the first collection come from two people discussing one machine. The second collection’s 90 occurrences are spread across fifteen people and several topics. Frequency per running word and spread across speakers now answer different questions. Report both if both matter. Repeating a term thirty times does not turn one speaker into thirty independent adopters.
For a grammatical choice, eligible contexts may be a better denominator than all words. Suppose a report marker appears in 18 of 30 passages where the task requires reporting someone else’s statement. Its rate is 60% of those eligible passages. Dividing 18 by every word in the recording would not describe the same choice. Define eligibility before judging the forms, or the denominator may be selected to favour the result.
Keep uncertain cases visible
In a fictional sample of 50 eligible passages, 20 clearly meet the coding rule, 10 are ambiguous and 20 clearly do not. The definite-positive proportion is 20 ÷ 50 = 40%. If all ambiguous cases qualified, the proportion would be 30 ÷ 50 = 60%. Thus the classification uncertainty spans 40–60% under this simple sensitivity calculation. This is not a statistical confidence interval and does not measure sampling uncertainty.
The honest answer retains the uncertainty category and explains how it might be resolved. A single percentage obtained by quietly discarding inconvenient examples would make the arithmetic look cleaner while making the evidence less useful.
CHAPTER 11 OF 14 · Measure evidence and test transfer
11. Design a small investigation that someone else can follow
A manageable study begins with one feature and a justified sample. For example: “How do consenting members of this fictional hobby group use one technical label in explanations to experienced and new participants?” That question is more workable than “How has bilingualism changed society?” It specifies a contrast and leaves room for a negative result.
Prepare a short research note before collecting material. State the question, inclusion rule, task, language-identification convention and unit of counting. Decide how quotations, proper names, repetitions and unclear audio will be treated. Include an uncertainty label. For a school activity, invented materials can teach these decisions without recording private conversations.
Check the annotation process
Suppose two coders independently classify 40 invented examples and agree on 34. Their simple agreement is 34 ÷ 40 = 85%. Inspect the six disagreements. If all concern quotations from a manual, the coding instructions may need a more precise quotation rule. Revising the rule and recoding the entire set is preferable to changing only the inconvenient cases.
Agreement is not the same as correctness. Two coders can share the same mistaken assumption, and a dataset dominated by one category can produce high agreement easily. The purpose of this elementary calculation is to locate decisions needing attention. A formal study may require additional reliability measures and a design suited to its question.
Consent and access are part of the study design. Explain the intended use, who can access recordings and whether examples will be published. Removing a name may leave a recognisable voice or story. Do not make a private family chat a public dataset simply because one participant is willing to share it. Agree the appropriate permissions before collecting or distributing material.
Finally, preserve the relation between a reported count and its examples. A reader should be able to understand which decisions produced the total, even when privacy prevents release of the original recording. Provide an anonymised coding description, carefully chosen permitted examples and a clear account of exclusions. If the evidence is too restricted for a strong claim, narrow the claim. The aim is a reproducible reasoning process, not a dramatic conclusion.
CHAPTER 12 OF 14 · Measure evidence and test transfer
12. Repair weak explanations and test a changed context
Three common failures have different repairs. A weak etymology treats resemblance as origin; repair it with chronology and comparison. A weak conversation analysis assigns a motive from one switch; repair it with surrounding turns and plausible alternatives. A weak trend claim compares unequal collections; repair it with comparable sampling and an appropriate denominator. Naming the failure precisely helps choose the next observation.
Language shift also needs its own evidence. It concerns changes in the distribution and transmission of language use, not merely the presence of borrowed words. Andreia Caroline Karnopp’s review of language shift emphasises that conditions do not produce identical outcomes in every community. A contact history does not by itself predict a community’s future.
Transfer the method from conversation to a digital form
Return to the fictional workshop term from earlier chapters. Now the workshop introduces an online booking form. The form uses the source-language equipment label, while staff usually use an adapted local form. Several users choose the wrong item. Does this demonstrate that borrowing has failed? No: the immediate practical problem is whether this interface enables users to identify the intended equipment.
Construct two clearly labelled prototype descriptions using the relevant forms and a plain explanation. Ask consenting testers to locate the item in a controlled task. Observe selections and requests for clarification. Keep the task, item set and instructions comparable. Do not equate a preference for one label with a complete account of a user’s linguistic competence.
If one wording yields fewer misidentifications, that supports a design decision for the tested task. It does not prove the origin of the term or justify removing other community forms from dictionaries. The historical question and the usability question require different outcomes, even though they concern the same word.
This is the transfer check: identify what stays constant and what changes. The expression and its possible contact pathway remain relevant. The audience, medium, task and success criterion change. A learner who can name those changes has moved beyond memorising definitions.
Where a real community wants to strengthen opportunities to learn and use a language, follow the Language Revitalisation root. That is an action programme directed by people concerned, rather than a prediction made from counting loanwords.
CHAPTER 13 OF 14 · Answer questions and continue
13. Answer common questions with useful limits
Does using a borrowed word mean someone is bilingual?
No. A person can acquire an established word as part of the language they normally use without knowing its earlier source. To describe bilingual experience, investigate the person’s repertoire and relevant tasks. Word history and an individual’s language knowledge are different subjects of evidence.
Does code-switching show that someone cannot speak properly?
A switch alone cannot establish that conclusion. Ask what happened in the interaction and which task is being evaluated. If a formal assignment requires a particular register, teach and assess that register explicitly. Do not turn one requirement into a judgement about every setting in which a speaker communicates.
Can languages influence each other without exchanging many words?
An investigation may concern patterns of construction, meaning or use rather than obvious imported forms. But this is not permission to declare every shared pattern a contact effect. The grammatical description, historical comparison and interaction pathway still have to support the explanation.
Is a calque the same as a bad literal translation?
No. A translated construction can become established, while another literal translation remains awkward or misleading for its intended readers. Check current convention for the communication task and historical evidence for an origin claim. One judgement cannot substitute for the other.
Is the language with more loanwords less authentic?
A borrowing count does not measure cultural authenticity or speakers’ legitimacy. It depends on what was counted, how origins were established and which domains were sampled. Communities can debate preferences for particular words, but the analysis should not present a percentage as a scientific ranking of people’s belonging.
Can an AI system discover a word’s origin?
It can suggest possibilities to investigate, but a fluent explanation is not a historical source. Check the cited dictionary, record or study directly. Generated examples may help teach a distinction if labelled; they must not be added to a corpus as independent observations of community speech.
What if the evidence never settles the question?
Report the strongest supported conclusion and the remaining alternatives. “Consistent with borrowing through B, but earlier evidence is missing” is informative. The aim is not to eliminate uncertainty by choosing a confident label. It is to make the uncertainty specific enough that a future discovery could improve the account.
CHAPTER 14 OF 14 · Answer questions and continue
14. Build a learning route from observation to explanation
Begin with a word or expression you can investigate responsibly. Write its current meaning and context before guessing its origin. Use the loanword and calque guides to identify possible mechanisms. Then consult a historical source and record what it actually supports. A satisfying story without a source remains a question, however often it appears online.
For a conversation project, choose a short permitted or invented exchange. Preserve the sequence and distinguish a description of the switch from an interpretation of its function. Compare a second situation with a changed audience or task. State which interpretation becomes stronger and which remains uncertain. Avoid creating a public collection of other people’s private speech merely to make the exercise feel authentic.
For a quantitative project, reproduce the calculations in Chapter 10, then alter one assumption. Double a corpus size without adding occurrences; the normalised rate halves. Redistribute the same occurrences across more speakers; the token rate stays constant while dispersion changes. These manipulations show which quantity each measure describes and prevent a familiar formula from becoming an automatic answer.
Produce a one-page evidence account
The final product should contain a clear question, the observed material, its source and date, the proposed contact pathway, a competing explanation and a justified conclusion. Add a short sentence explaining what further observation would change your mind. This makes the work inspectable and gives the next learner a concrete starting point.
Read Sociolinguistics when the main question concerns audience, register and social interpretation. Read Historical Linguistics when the main question concerns change and relationships over time. Use How X Works to explore mechanisms in neighbouring subjects and return to the Gold Standard master for the broader collection.
The research links throughout this guide lead to the relevant primary publications, author descriptions and research databases. They support the distinctions being taught; the fictional exercises are original demonstrations rather than reported findings. When pursuing a real language case, continue into its language-specific scholarship and credit the researchers and community contributors whose evidence makes the account possible.
The final test is practical: can you explain what the evidence shows without turning an occurrence into a convention, a resemblance into an origin, or a conversational adjustment into permanent language change? If you can also name the next useful check, you have a method that can travel beyond the first example.