Applied linguistics investigates practical problems involving language and communication, then uses evidence to improve what people can understand and do. Its gold standard is a defensible connection between a real problem, an explanation, a suitable response and an observable outcome. A successful project does more than apply a linguistic label: it shows why a change helps particular people in a particular situation, and where that conclusion stops.
This guide follows original fictional cases involving a community workshop, language learning, assessment and translation. All numerical examples are constructed for teaching. Start with the task you need to improve, then follow the relevant chapter route. The Gold Standard master connects this subject to the wider collection, and the Education parent explains the goal of learning that transfers into unfamiliar situations.
Choose your applied linguistics route
Chapters 1–3: Define the problem and evidence
Chapters 4–6: Design learning and assessment
Chapters 7–9: Improve communication in context
Chapters 10–12: Evaluate, repair and transfer
Chapter index
Define the problem and evidence · Chapters 1–3
1. Begin with a communication problem, not a favourite solution
Design learning and assessment · Chapters 4–6
4. Design language learning around an actual capability
5. Use vocabulary measures without mistaking them for comprehension
Improve communication in context · Chapters 7–9
7. Improve multilingual communication without flattening people’s repertoires
Evaluate, repair and transfer · Chapters 10–12
10. Evaluate change with a transparent worked comparison
Answer questions and investigate · Chapters 13–14
CHAPTER 1 OF 14 · Define the problem and evidence
1. Begin with a communication problem, not a favourite solution
A community workshop receives many incomplete bookings. Staff conclude that applicants need better English. They propose a vocabulary lesson. Before accepting that explanation, an applied linguist would examine what applicants are being asked to do, what the booking form says, which fields are missed and whether the process works for people who already understand the words.
Perhaps the form asks for an “attendance designation”, although staff only need to know whether someone is attending as a participant or a helper. Perhaps the two options appear below a button that looks like the end of the form. Perhaps applicants cannot decide because the workshop itself has not explained the roles. These are different problems. A vocabulary lesson would address only part of the first, and might not be the best response even there.
The International Association of Applied Linguistics describes an interdisciplinary field concerned with practical language and communication problems. Its scope includes learning, literacy, multilingualism and language policy, among other areas. The field is broader than language teaching, and its practical orientation can also generate new questions for linguistic theory.
State the problem at an observable level
Replace “Applicants have poor English” with “Applicants frequently leave the role field blank.” The second statement can be checked. It does not yet explain the pattern, and that is an advantage. It leaves room for evidence about wording, layout, background knowledge, technology and the organisation’s own instructions.
Write the intended outcome just as carefully: “Applicants can select the role that matches what they intend to do.” A higher completion rate is helpful only if the choices are accurate. Forcing everyone to click a default answer could improve the database while making the booking information worse. The practical goal determines which measurements matter and which apparent improvements are misleading.
CHAPTER 2 OF 14 · Define the problem and evidence
2. Conduct a needs analysis with the people doing the task
A needs analysis asks what people must accomplish, what they can currently accomplish, what conditions shape their performance and which gaps matter most. It should include the people affected by the eventual intervention. Staff see incomplete records; applicants experience the wording and uncertainty that precede them. Neither perspective alone supplies the whole explanation.
For our workshop, collect the current form, the invitation, the confirmation message and the instructions volunteers receive. Walk through the entire sequence as a new applicant. Ask a small, varied group to explain what they think each step requires. Observe where they hesitate without immediately supplying the answer. Record the difference between a person’s interpretation and the organisation’s intended meaning.
Include constraints. Some applicants use a phone, some share a device, some read the invitation in another language and some rely on a friend to explain it. A solution requiring a long video with sound may exclude people who could easily read a short message. A translation can help with language access while leaving an ambiguous process unchanged.
Separate needs, preferences and assumptions
An applicant may prefer a paper form, need an accessible interface and be assumed by staff to need simpler vocabulary. These are three different statements. Ask what evidence supports each. A preference deserves attention, but it does not automatically establish the cause of an error. An assumption should be tested before it becomes a programme of lessons.
End this stage with a short task specification: who needs to do what, using which materials, under which conditions, and with what acceptable result. For example, a first-time applicant should be able to choose a workshop role on a phone, explain the choice and identify the next step. That specification is much more useful than a general ambition to “improve communication”. It also exposes disagreements between stakeholders before money and effort are committed to a solution.
CHAPTER 3 OF 14 · Define the problem and evidence
3. Build an explanation from several kinds of evidence
Applied work often combines close analysis with observation, interviews, counts and practical testing. Each method answers a different question. Text analysis can identify an ambiguous phrase. An interview can reveal how one person understands it. A collection of completed forms can show how often an error appears. A comparison of alternative forms can help test whether a revision changes performance.
Suppose ten applicants interpret “participant” as “anyone who attends”, while the organiser uses it to exclude helpers. That is evidence of a mismatch in category use. It does not show that the applicants lack general vocabulary. In ordinary language, their interpretation may be entirely reasonable. The organisation has introduced a local distinction without explaining it.
Keep an evidence notebook with three separate entries: observation, interpretation and proposed action. “Seven applicants left the field blank” is an observation. “They were uncertain about the categories” is an interpretation. “Add short descriptions beside the options” is a proposed action. Joining these entries too quickly makes an attractive solution look inevitable.
Look for the case that does not fit
Find applicants who completed the field correctly and ask what helped them. Perhaps they had attended before, heard a verbal explanation or guessed successfully. Correct performance does not always indicate clear design. Conversely, someone may understand the categories but leave the field blank because neither option fits their circumstances.
The Corpus Linguistics root explains how to investigate patterns across systematically assembled language material. Use that route when the question concerns recurring wording or genre patterns. The Cognitive Linguistics root helps distinguish conceptual categories and alternative presentations of a situation. Neither replaces direct evidence about whether this booking process works for its users.
CHAPTER 4 OF 14 · Design learning and assessment
4. Design language learning around an actual capability
Sometimes the evidence does identify a learning need. Workshop volunteers may understand written requests but struggle to clarify a spoken request politely when the room is noisy. Teaching twenty definitions will not directly practise that capability. A better task gives learners a reason to listen, identify uncertainty, ask a focused question and confirm the resulting understanding.
Create an original role-play. One learner needs equipment for a demonstration; another controls a fictional equipment list. The request contains one missing detail, such as quantity or collection time. The listener must obtain that detail before completing a record. The outcome is a correct record and an appropriate clarification, not merely a fluent conversation.
Begin with support: a model, a short list of clarification expressions and a visible record template. Later remove some support, vary the equipment and introduce an unfamiliar speaker. Record which changes cause difficulty. This sequence helps distinguish memorising a dialogue from using a communicative strategy. It is an instructional design example, not a claim that one sequence is universally optimal.
Make the feedback point to the next action
“Be clearer” gives the learner little to work with. “You confirmed the item but not the quantity; ask how many before completing the record” identifies the missing move. Follow the feedback with another attempt containing a different quantity. If the learner simply repeats the corrected dialogue, you have not yet checked transfer.
The existing guide to collocations in listening introduces why recognising familiar word combinations can matter. Use such specialist explanations to generate a diagnostic question: does the learner recognise the expression in print but miss it in speech? Do not assume that every listening difficulty has the same cause. Noise, unfamiliar pronunciation, attention, vocabulary and task complexity can interact, and the lesson should address the bottleneck the evidence actually reveals.
CHAPTER 5 OF 14 · Design learning and assessment
5. Use vocabulary measures without mistaking them for comprehension
Vocabulary is important because words carry information, but a single vocabulary score cannot describe every reading problem. A reader may know the ordinary meaning of a word while missing its technical sense. Another may identify every word but lose track of the relationship between a condition and an exception. Applied analysis asks which knowledge is required for this text and this reading purpose.
Consider a fictional 500-token workshop guide. If a reader does not know 25 of those running word tokens, estimated coverage is 475 divided by 500, or 95 percent. If ten tokens are unknown, coverage is 490 divided by 500, or 98 percent. These calculations describe the proportion of tokens counted as known under the chosen measurement procedure. They are not direct percentages of meaning understood.
The lexical coverage specialist develops this distinction. A primary research route is Schmitt, Jiang and Grabe’s 2011 study, which examined word knowledge and comprehension using two texts and participants from several countries. Such studies investigate a relationship; they do not turn one coverage value into a guarantee for every reader and task.
Inspect where the difficulty occurs
Suppose the unknown word is repeated in a background anecdote. Its effect may differ from an unknown word that names the required action. Similarly, misunderstanding “unless” can reverse a condition even when nearly all other words are familiar. Ask the reader to explain what they would do, then identify the part of the text supporting that answer.
A useful repair might combine a glossary, a clearer sentence and an example showing the exception. Evaluate each against the original communication goal. Replacing all specialist terms can make a document less useful if readers need those terms on labels or equipment. Explain necessary terminology at the point of use, and test whether readers can connect the explanation to the real object or action.
CHAPTER 6 OF 14 · Design learning and assessment
6. Make assessment evidence fit the decision
An assessment samples performance and supports an interpretation. Its quality depends partly on what someone intends to do with the result. A short classroom exercise can identify a useful next lesson without supporting a broad claim about someone’s overall language proficiency. The same score can be informative for one purpose and unsuitable for another.
For volunteer training, suppose the target is accurately clarifying an equipment request. A multiple-choice vocabulary quiz samples recognition of selected meanings. It does not directly show whether the volunteer can ask a follow-up question in conversation. A role-play may be closer to the target, but it introduces other influences: partner behaviour, familiarity with the scenario and how the assessor interprets the performance.
The International Language Testing Association’s Guidelines for Practice emphasise matters including validity, consistency, fairness, accessibility and appropriate communication of results. These considerations apply to the use made of an assessment, not merely to whether its questions appear professional. Follow relevant institutional requirements when an assessment has consequential uses.
Build a small transparent rubric
For our fictional training task, use three separate checks: the required detail was identified, the clarification question made the uncertainty clear, and the final record matched the confirmed information. Keep pronunciation or stylistic preferences out of the score unless they materially affect the defined target. Explain the criteria before practice so learners understand what successful performance means.
Ask two assessors to score the same sample independently, then inspect disagreements. Perhaps one accepts an implied confirmation while another requires an explicit one. Clarify the criterion and reconsider affected ratings. Agreement is useful, but two assessors can consistently apply a bad criterion. You still need to ask whether the assessment represents the capability and whether the resulting decision is justified. Reporting separate dimensions often provides more useful feedback than hiding everything inside one total.
CHAPTER 7 OF 14 · Improve communication in context
7. Improve multilingual communication without flattening people’s repertoires
People may use different languages or varieties for different activities, audiences and relationships. A person who prefers one language for reading instructions may use another confidently in conversation. Treating them as simply fluent or not fluent can conceal the resources available to them and the precise support they need.
The Council of Europe’s account of plurilingual and pluricultural competence treats a person’s repertoire as interconnected and evolving. This is a useful perspective for asking how people combine resources. It does not remove the need to specify which language is required for a particular task or to evaluate whether everyone can participate.
In our workshop, two volunteers might discuss a difficult instruction in a shared language before producing an English equipment record for the team. The relevant questions are whether the discussion helps them understand, whether the final record is accurate and whether other team members can use it. Banning the discussion could remove useful support; leaving the final record inaccessible to colleagues would create a different problem.
Establish the common information everyone needs
Identify the information that must be shared across the group: item names, quantities, collection times and changes. Decide how those details will be recorded and confirmed. Allow flexible routes to understanding while maintaining a dependable shared record. The goal is participation and accurate coordination, not performance of one preferred identity.
Do not assume a translation is enough simply because a language label matches someone’s background. Literacy, variety, terminology, script and familiarity with the activity matter. Ask users to demonstrate the task and explain difficult points. Nor should one bilingual participant automatically become the unpaid interpreter for everyone else. Planning language access includes considering who is doing the additional work and whether they have agreed to it.
CHAPTER 8 OF 14 · Improve communication in context
8. Translate the intended task, preserving its conditions
A workshop instruction says, “Return unused materials to the desk after your session.” A translator needs to preserve the action, the object, the destination and the timing. A fluent version that says “Return all materials before your session” would fail despite sounding natural. The difference concerns the operation people will carry out, not a minor stylistic preference.
Before translation, identify unresolved information in the source. Does “unused” include opened packets? Which desk is meant? Is the instruction for participants, helpers or both? A translator cannot reliably repair a policy the organisation has not decided. Keep a list of questions and obtain an authorised answer rather than quietly choosing whichever interpretation reads best.
The existing source-text reading guide provides a detailed route into this preparation. For the workshop, create a brief containing audience, purpose, terminology, essential conditions and the form in which the translation will appear. A caption, a spoken announcement and a printed checklist may require different arrangements while preserving the same operational content.
Review the meaning and test the use
A bilingual review can compare the source and translation for omissions, additions and altered relationships. A user test asks whether intended readers can carry out the task. Both are useful. A translation may preserve a confusing source accurately, while a reader-friendly rewrite may accidentally remove an exception. The checks need to address both fidelity and usability.
If machine translation assists the work, preserve the source version, identify who reviewed the output and examine numbers, negation, names and conditional relationships particularly carefully. Do not upload private material to an unapproved service. For a low-stakes classroom exercise, invented instructions provide enough complexity to practise these checks without using real personal information. The broader lesson is that speed of production and reliability of communication are separate outcomes.
CHAPTER 9 OF 14 · Improve communication in context
9. Design instructions that readers can find and act on
An instruction can be linguistically accurate and still fail because it is difficult to locate. If a return deadline is buried in a long introductory paragraph, readers may miss it while scanning for what to bring. Applied work therefore considers document structure and interface behaviour alongside vocabulary and grammar.
The W3C’s writing guidance for web accessibility recommends meaningful headings and link text, clear instructions and concise content. These are useful starting principles, not proof that any particular rewrite is accessible to everyone. Accessibility also depends on implementation and the ways people interact with the material.
For our fictional booking form, replace an unexplained field label with “How will you take part?” Then describe the options: “Participant: join the activities” and “Helper: support the organisers.” This revision still needs review. Perhaps helpers also join activities, so the categories overlap. If so, allow the relevant combination or ask a more specific question. Clear wording cannot make a false distinction true.
Test the error path as carefully as the happy path
Suppose someone submits the form without selecting a role. “Invalid input” identifies no repair. “Choose whether you will join as a participant, help the organisers, or do both” is more actionable if those are the actual options. The interface must also make the relevant field reachable and retain the other information already entered.
Run the task with representative users and appropriate accessibility checks. Ask them to find the role description, make a choice and recover from a missing answer. Do not count a task as independently completed if a facilitator repeatedly points to the right place. Record assistance separately. A form that works only when a helpful researcher stands beside the user has not yet demonstrated that it works in ordinary use. This is why careful testing observes the process, not just the final submitted record.
CHAPTER 10 OF 14 · Evaluate, repair and transfer
10. Evaluate change with a transparent worked comparison
Here is a constructed example. Before a form revision, 48 of 80 applicants select a role that matches their stated intention. Afterwards, 68 of another 80 do so. Assume the same definition of accuracy and complete follow-up for both groups. The observed rates are 48 divided by 80, or 60 percent, and 68 divided by 80, or 85 percent. The difference is 25 percentage points.
That is a useful descriptive result. It does not yet prove that the wording caused the improvement. The later applicants might have received a verbal explanation, have more previous experience or be applying for a different kind of workshop. A before-and-after comparison confounds the revision with whatever else changed over time.
Add a comparison without overstating it
Suppose a second, comparable workshop keeps its form unchanged. Its accuracy rises from 40 of 80, or 50 percent, to 48 of 80, or 60 percent. Its improvement is ten percentage points. Subtracting that from the revised workshop’s 25-point improvement gives a difference in changes of 15 percentage points. The calculation is 25 minus 10, not 85 minus 60.
This comparison can help investigate shared changes, but the 15-point figure is not automatically a causal effect. An interpretation of that kind would require assumptions about how the groups would otherwise have changed, alongside evidence about comparability and other events. The second workshop might have a different audience or receive a different publicity campaign. Show the four counts so readers can inspect the calculation instead of presenting one impressive number.
Count the costs and unintended effects
Imagine the revised form improves accuracy but takes much longer to complete. Or fewer people complete it at all. Report those outcomes too. Define the denominator carefully: accuracy among submitted forms differs from successful accurate submissions among everyone who starts. Excluding people who abandon the process may make the result look better while concealing a new barrier.
A stronger next evaluation could allocate users to alternative versions under an appropriate design, keep relevant conditions stable and decide the outcomes in advance. It could also collect explanations from users to understand remaining errors. Numerical comparison estimates a pattern; qualitative evidence can help explain how that pattern arose. Neither should be used as a decorative addition to a conclusion already chosen.
CHAPTER 11 OF 14 · Evaluate, repair and transfer
11. Diagnose six common failures in applied work
The first failure is defining the person as the problem. “These users cannot read properly” closes inquiry too early. Return to the actual task and inspect the materials, assumptions and conditions. A user may be interpreting a label reasonably while the organisation uses it in an undocumented special sense.
The second failure is choosing the intervention before examining the evidence. A team may prefer an app, workshop or translation because it is familiar or visible. Ask which observed difficulty the proposal addresses and what mechanism would connect the change to the desired outcome. If no clear connection can be stated, further diagnosis is needed.
The third failure is measuring what is easy instead of what matters. Attendance at a lesson does not establish learning. Satisfaction with a form does not establish accurate completion. A polished translation does not establish preserved meaning. Keep process measures, experience measures and outcome measures separate so that each contributes what it can actually show.
Repair the explanation, not just the presentation
The fourth failure is treating one successful demonstration as evidence of broad transfer. A learner may recognise a rehearsed request and fail when the quantity, speaker or medium changes. Introduce a relevant new condition and identify which part of the skill survives. Then adjust teaching at the point where performance breaks down.
The fifth failure is hiding variation inside an average. A revision might help experienced readers while creating difficulties for first-time users. Inspect patterns that matter for the task, while avoiding unreliable claims from tiny subgroups. A small sample can reveal a problem worth investigating without measuring its prevalence accurately.
The sixth failure is failing to maintain the intervention. A glossary becomes misleading when the organisation changes its role names. A translation becomes stale when the source changes. Assign responsibility for keeping related versions aligned and define what triggers review. Applied work continues after the attractive report is finished. A result is dependable when people can still use it correctly after ordinary changes in the environment, not only on the day it was launched.
CHAPTER 12 OF 14 · Evaluate, repair and transfer
12. Transfer the approach from a form to a spoken briefing
Now change the situation. Volunteers receive a spoken briefing before a workshop. Several collect equipment at the wrong time. The organiser wants to reuse the successful form revision and distribute a vocabulary list. Your task is to decide what should transfer and what needs a new investigation.
The transferable method is to define the action, inspect the communication, observe performance and test an appropriate repair. The specific explanation may not transfer. A spoken message unfolds over time. Listeners may miss an early condition, hear competing noise or have no opportunity to ask a question. A wording change that helped readers scan a form may not address these constraints.
Start by reconstructing the message accurately. Suppose it says, “After the first activity, unless your team is assigned to setup, collect the materials from the side desk.” Ask volunteers to explain when each team should collect materials. If people disagree, inspect the condition, the exception and the earlier explanation of team roles. Do not infer a vocabulary deficit solely from the wrong collection time.
A worked repair and its limits
One candidate revision is to separate the groups: “Setup team: collect your materials now. All other teams: collect materials after the first activity. Use the side desk.” This is appropriate only if it preserves the actual arrangement. The original sentence did not itself specify what the setup team should do, so the organiser must confirm that added instruction before using the rewrite.
Then ask representatives from each group to state their next step in their own words. A correct chorus of “yes” is a weaker check than an explanation of the action. If the room is noisy, a visible written summary may complement the briefing. Evaluate whether people can find and use it when needed rather than assuming that adding another channel always helps.
The answer to the transfer task is therefore not “use simpler language everywhere”. It is “carry over the diagnostic method, re-establish the task conditions and test a repair that fits the new medium”. This distinction is central to applied linguistics. The field’s usefulness comes from disciplined adaptation, not from a universal list of communication tricks.
CHAPTER 13 OF 14 · Answer questions and investigate
13. Answer practical questions about the field
Is applied linguistics just language teaching?
No. Language teaching is a major area, but the field also examines practical problems in assessment, translation, multilingual communication, literacy, policy and other settings. The common thread is a language-related problem investigated with appropriate theory and evidence. A practical project can involve several disciplines and several kinds of expertise.
Must a project begin with a linguistic theory?
It needs a clear problem and a defensible way to investigate it. Theory can help identify mechanisms, make distinctions and generate predictions. Starting with a favourite theory and forcing every observation into it is less useful. Let the evidence constrain the explanation, and state when more than one account remains plausible.
Is the simplest wording always best?
No. Removing necessary distinctions can make a message shorter and less accurate. Readers may need specialist terms to match a document to equipment or a later procedure. Explain essential terms and preserve conditions. Judge the wording by whether intended users can understand and act appropriately, not by sentence length alone.
Does a language level describe everything someone can do?
No single label captures every task, medium, topic and support condition. A person can read familiar material well and struggle with spontaneous conversation in an unfamiliar setting. The CEFR Companion volume overview includes descriptors across multiple communicative activities, including mediation and online interaction. Any level claim still needs suitable evidence and a clear account of its use.
What is mediation in language use?
The Council of Europe’s mediation account concerns enabling communication and helping construct or convey meaning, within or across languages. In our workshop, explaining a complicated instruction to a colleague can involve mediation. Successful explanation preserves relevant meaning and helps the recipient participate; it is not merely repeating the original more loudly.
Can a small project produce useful evidence?
Yes, if the claim matches the evidence. A few observed users can reveal an ambiguous label or a broken step. They usually cannot establish how common the problem is across a large population. Describe the sample, retain the examples and use the findings to improve the next investigation. Small honest evidence is more useful than a large conclusion built on unstated assumptions.
CHAPTER 14 OF 14 · Answer questions and investigate
14. Complete a project that another person can inspect
Choose a modest task: a club invitation, a classroom instruction, a public information sheet or a fictional translation exercise. Define success before redesigning the material. Gather the current version and enough context to understand its purpose. Identify who should be consulted, what information is necessary and how their contributions will be handled.
The British Association for Applied Linguistics’ good-practice guidance provides points for reflection on research practice and responsibilities. It is intended to be used alongside relevant institutional and local requirements. For an informal learning exercise, invented examples often allow substantial practice without collecting identifiable personal material or placing participants under pressure.
Create a short report with the original problem, materials, observations, competing explanations, proposed change and evaluation. Include one example where the change did not work. Explain what you would revise next. Preserve a distinction between the evidence collected and the conclusions you hope a later study will establish. Another reader should be able to follow the reasoning without needing access to your private assumptions.
Build a practical learning route
For language learning, begin with a capability and design practice that progressively changes the conditions. For assessment, define the interpretation and decision before choosing tasks. For translation, establish the source meaning and review both fidelity and use. For public communication, observe how people locate, interpret and act on information. These routes share a method while requiring different subject knowledge.
Use the primary and official sources linked in the relevant chapters for their specific purposes: AILA for field scope, ILTA for assessment practice, the Council of Europe for communicative descriptors and mediation, W3C for accessible writing, and the cited vocabulary study for one empirical investigation. No single source certifies every intervention, and none of the fictional examples here should be reported as published research findings.
Continue through the How X Works hub or the verified specialist guides linked throughout this article. The final check is practical: can you identify what went wrong, explain why your proposed response fits, show what changed and recognise when a different situation requires a different solution? That is the capability a sound applied linguistics project should develop.