Sound studies examines how people make, hear, record, organise and interpret sound in particular social and historical settings. Its gold standard is an explanation that connects a specific audible event to its listeners, technologies, institutions and circumstances, while showing exactly what the evidence can and cannot establish.
A recording of a railway announcement, a remembered street call or a disputed neighbourhood performance can each become a useful starting point. This guide shows how to turn those starting points into answerable questions, compare recordings responsibly, interpret voices and silences, and produce a small research project whose conclusions survive a change of context. All numerical examples and unnamed scenarios below are original teaching cases, not reports of actual communities or experiments.
This subject belongs within the Gold Standard collection and its Education parent: understanding should remain useful beyond the example in which it was learned. For practical production and listening workflows, use The Gold Standard of Sound. The emphasis here is the interpretation of sonic evidence and the histories and relationships that make sound meaningful.
Choose your sound studies route
Chapters 1–3: Frame a question and listen carefully
Chapters 4–6: Interpret histories, recordings and voices
Chapters 7–9: Study public sound, media and archives
Chapters 10–12: Compare evidence and repair an argument
Chapter index
Frame a question and listen carefully · Chapters 1–3
1. Ask what a sound does in a particular setting
2. Separate the event, the recording and the listening situation
Interpret histories, recordings and voices · Chapters 4–6
4. Reconstruct a listening history from uneven traces
5. Check playback before interpreting a recorded voice
6. Read voices, pauses and transcripts as different evidence
Study public sound, media and archives · Chapters 7–9
7. Study public sound as a question of use and authority
Compare evidence and repair an argument · Chapters 10–12
10. Compare sound environments with a fair denominator
Test transfer and continue learning · Chapters 13–14
CHAPTER 1 OF 14 · Frame a question and listen carefully
1. Ask what a sound does in a particular setting
Sound studies is an interdisciplinary field rather than a single laboratory method. Northwestern University’s introduction to the subject places it across the arts, humanities and social sciences, including histories of sound technology, mediated expression and conflicts around listening. This breadth permits several kinds of evidence, but it does not make every appealing interpretation equally persuasive.
Start with a relationship that needs explaining. A bell may signal a school transition, mark a religious occasion, announce a visitor or appear in a film soundtrack. Describing its pitch does not establish its purpose. Calling it a symbol does not establish who recognises that symbol. The research problem emerges when you identify a setting, participants, an action and a consequence.
Turn an impression into a question
“This station sounds unfriendly” is a useful personal observation. A researchable question might be: how do announcement wording, repetition, language choice and delivery affect how different passengers understand a platform change? Now there are things to examine: recordings, scripts, passenger accounts, observations of information use and the conditions under which each was collected.
An alternative question concerns authority: which voices are allowed to address everyone in the station? That requires evidence about permissions, employment, communication procedures and informal practices. The same location supports different inquiries. Choose one before gathering a large folder of recordings that cannot answer it.
Write a provisional answer in one sentence, then add what would challenge it. If your explanation is that unfamiliar wording causes confusion, evidence that regular commuters also misunderstand a clearly worded message would require another explanation. Perhaps reverberation, competing announcements or a display mismatch matters more.
The practical achievement is a bounded claim. “In these observations, passengers missed the changed platform number when two announcements overlapped” is stronger than a sweeping claim that public transport ignores its users. It states where the explanation applies and points toward a checkable mechanism. Broader conclusions can follow when broader evidence supports them.
CHAPTER 2 OF 14 · Frame a question and listen carefully
2. Separate the event, the recording and the listening situation
A sound event and a recording of that event are different objects of study. The event includes sources, movement, people, surfaces, activities and timing. The recording includes the recorder’s position, equipment, settings, selection and later processing. The listening situation adds a playback system, listening level, location, attention and purpose.
Imagine recording a conversation near a fountain. A microphone beside one speaker may capture clear speech with a distant wash of water. Another recorder beside the fountain may capture strong splashing and indistinct conversation. Both can be honest recordings. Neither establishes, by itself, what every person present heard.
Follow the chain of mediation
For each important observation, trace the path from event to claim. Where was the sound produced? Where was it captured? Which part was retained? Was it filtered or shortened? How did you listen? What language did you use to describe it? This chain helps locate the point at which an apparent fact may have been introduced by a decision.
Suppose a documentary clip sounds intimate. The intimacy might arise from a close microphone, a quiet performance, an edit that removes interruptions, headphones used by the audience, or several of these together. “The speaker was naturally intimate” jumps from an effect to a personality claim without testing the intervening choices.
Keep a short record beside the audio: file identifier, event and capture dates if known, place, recorder position, device, known processing, source of the file and unresolved uncertainties. A blank field is better than an invented answer. Label an estimated date as estimated and explain the basis.
The specialist guide to field recording extends the practical capture and custody questions. In sound studies, those details matter because they determine which interpretations remain defensible. If the original recording chain is unknown, the file can still be studied as a circulating media object; it simply cannot support every claim about the original event.
CHAPTER 3 OF 14 · Frame a question and listen carefully
3. Describe listening without making your ears universal
Listening is an activity carried out by particular people with particular purposes. A musician preparing an entrance, a commuter seeking an announcement and a caretaker checking a machine may attend to different features in the same environment. A sound that is background for one person can be information for another.
Begin with a modest descriptive vocabulary: repeated or isolated, continuous or intermittent, near or distant in this recording, speech or non-speech, identified or unidentified. Then distinguish description from evaluation. “Three sharp pulses recur after each spoken sentence” is a different claim from “the warning is irritating.” Both may matter, but the second needs a listener and a context.
Make disagreement informative
In a teaching exercise, two listeners hear a five-minute courtyard recording. One calls it peaceful; the other calls it tense. Do not average the labels into “moderately peaceful.” Ask which moments each listener attended to and what those moments suggested. One may focus on birds; the other may recognise a repeated alarm in the distance.
Replay the relevant time ranges at a comfortable level. Keep the listeners’ initial descriptions, then record any revisions separately. If someone changes their interpretation after learning the source, that change is evidence about contextual knowledge. It does not mean the first response was dishonest or the second response is the only correct one.
Avoid using listening differences as an informal diagnosis of hearing, intelligence or cultural competence. A research task can offer written descriptions, visual timelines, discussion and other participation routes without requiring everyone to perceive an identical auditory experience. Ask participants which forms of access help them contribute.
For the broader relationship among sensory experience, language and social learning, continue to Sensory Studies. Sound studies adds a sustained focus on sonic events and their mediation. Its useful question is not simply “What did people hear?” but “What were they listening for, and how do we know?”
CHAPTER 4 OF 14 · Interpret histories, recordings and voices
4. Reconstruct a listening history from uneven traces
Historical sound research often starts with incomplete evidence. A written complaint, a programme, a technical manual, a recording and a later recollection each preserve different aspects of an event. None should automatically be treated as a complete account of how the past sounded.
Consider an imagined town archive containing a letter about an evening bell, a timetable and a recording made many years later. The letter demonstrates that its author complained about the bell. The timetable may establish the intended schedule. The later recording can document a later performance. Combining them does not prove the bell sounded identical throughout the intervening period.
Build a claim with separate supports
Make three columns in your working notes: proposition, supporting trace and unresolved alternative. For “the bell marked closing time,” a contemporary timetable might be strong support. For “everyone welcomed the bell,” the complaint is counterevidence, while the timetable says little. For “the bell was audible across the whole town,” you would need more than a close recording.
Chronology matters at several levels. Distinguish when an event occurred, when someone recalled it, when the account was recorded and when the file was copied or published. A recording uploaded this year may preserve an older interview about an even older event. A website date alone cannot settle that sequence.
Read descriptive words in context. An author calling a sound “modern,” “coarse” or “respectable” provides evidence about a judgement as well as possibly about an acoustic event. Ask who wrote the description, for whom, with what expectations and with what opportunity to hear. Do not translate a historically loaded adjective into an exact measurable property.
A responsible historical account can end with an unresolved distinction. “The surviving records establish an organised bell schedule and disagreement over its use; they do not establish its precise loudness at every location.” That is a useful result. It preserves the difference between reconstructing a practice and recreating an entire sensory world from a few traces.
CHAPTER 5 OF 14 · Interpret histories, recordings and voices
5. Check playback before interpreting a recorded voice
Recorded voices invite quick judgements about pace, confidence and emotion. Before interpreting those qualities, establish whether the available file is being reproduced as intended. A speed error can change duration and pitch together in ordinary rate-change playback. Other processing can change them separately, so the mechanism must be identified rather than assumed.
The US National Archives’ recording-speed guidance treats speed as information needed for playback, recording tape speed in inches per second and disc or cylinder speed in revolutions per minute. Such metadata is part of interpretation, not merely equipment administration.
Worked case: a voice that seems unusually hurried
Assume a fictional disc was recorded at 78 revolutions per minute but transferred at 80. Assume the transfer uses ordinary speed change with no independent pitch correction, and the original passage lasted 180 seconds. The speed ratio is 80 ÷ 78, approximately 1.02564. Playback runs about 2.56% faster than intended.
The reproduced duration is 180 × 78 ÷ 80 = 175.5 seconds. A steady 440 Hz tone on the original would reproduce at 440 × 80 ÷ 78, approximately 451.28 Hz. These calculations do not tell us what the speaker felt. They show that timing and pitch evidence have been altered by a known transfer condition.
The answer is not to slow every old recording until the voice sounds plausible. That would replace an unknown with personal expectation. Investigate the carrier, documentation, transfer notes and any reliable reference signals. If uncertainty remains, describe it and avoid fine claims about vocal speed or pitch.
For a classroom comparison, retain the received file and create a clearly labelled demonstration copy with the assumed correction. Compare the interpretations, not just the sound. Which claims change? Which remain supported? A statement about the words spoken may survive, while a claim about hurried delivery may weaken. This is the central lesson: technical provenance can change the historical argument.
CHAPTER 6 OF 14 · Interpret histories, recordings and voices
6. Read voices, pauses and transcripts as different evidence
A transcript is a representation designed for a purpose. It can preserve words, identify speakers and help readers navigate an interview. It cannot automatically carry every feature of timing, overlap, emphasis or background sound. Conversely, an audio file does not automatically explain who is speaking, which words are uncertain or what happened outside the microphone’s reach.
Choose a transcription convention that fits the question. If you study how people take turns, preserve meaningful overlaps and pauses with a stated method. If you study the content of a life story, readable paragraphs may help, but record where editing removes repetitions or false starts. Do not silently turn a cleaned transcript into evidence about fluent delivery.
Worked case: what does a seven-second pause mean?
In an invented interview, a question ends at 02:14 and the response begins at 02:21. The measured gap is seven seconds. That establishes a timing observation in the available file. It does not establish embarrassment, refusal, confusion or trauma.
Possible explanations include thinking, an interruption outside the microphone’s range, a gesture, a production cut or reluctance to answer. Seek field notes, the unedited sequence and, where appropriate, the participant’s explanation. If none is available, write “a seven-second gap precedes the response” rather than assigning an emotion as fact.
The Oral History Association’s core principles describe oral history as a collaborative process and emphasise ongoing participation, consent and clarity about preservation, use and access. Apply that care to the voice itself: permission to record a conversation should not be casually treated as permission for every future presentation or transformation.
For a small research project, explain the intended audience and give participants a clear way to discuss attribution and access conditions. Keep the agreed terms with the recording. A powerful quotation is not a reason to disregard those terms. The resulting account should allow readers to distinguish what the person said, what the recording preserves and what the researcher infers.
CHAPTER 7 OF 14 · Study public sound, media and archives
7. Study public sound as a question of use and authority
Public sound connects physical space with social arrangements. An announcement, street performance, delivery vehicle and construction activity may occupy the same place while serving different purposes. A useful investigation asks who produces each sound, who depends on it, who encounters it involuntarily and who can alter the arrangement.
Begin without deciding that one group’s description represents the whole place. “Lively,” “disruptive” and “necessary” may refer to different times, positions and activities. A shop worker answering customers and a visitor enjoying a performance are not necessarily contradicting each other’s experience. They may be describing different demands within the same shared environment.
Worked case: complaints are not the entire soundscape
Suppose a fictional square receives 18 recorded complaints during a month. Twelve come from one address, four from another and two from separate addresses. There are 18 complaint events from four addresses, not evidence of 18 independent households. Nor does the absence of complaints from other addresses prove approval.
The count becomes useful when linked to time, activity, location and reporting opportunity. Were the repeated reports about one recurring event? Could people submit complaints easily? Did staff record verbal concerns? Were any complainants absent during the period being compared? These questions affect the interpretation before any policy conclusion follows.
Add observations and accounts from people with different uses of the square. A possible finding is that disputes cluster during telephone orders at two nearby shops. That suggests a more specific question about timing, orientation and communication than a general verdict that the entire square is too noisy.
Keep different kinds of evidence distinct. A measured sound level, a complaint and a description of lost concentration answer different questions. Combining them can improve an explanation, but none should be relabelled as the others. The research goal is to clarify the conflict and its conditions, including who has influence over possible changes, rather than to let the loudest argument stand in for everyone present.
CHAPTER 8 OF 14 · Study public sound, media and archives
8. Follow sound through media, work and circulation
A sound can acquire new meanings as it moves between settings. A door chime recorded for a practical demonstration might later become a podcast transition or a film cue. The waveform may be similar, yet its relationship to an audience changes. Studying that movement requires attention to selection, naming, editing, attribution and placement.
Take an invented recording of a train departing. In a transport archive it may document a vehicle and location. In a family film it may mark a farewell. In an advertisement it may suggest escape. You cannot infer the meaning of all three uses from the recording’s original circumstances alone. Equally, the later uses do not erase the need to identify the original source.
Ask who made the apparently effortless effect
Trace the work behind a finished sequence: recording, selecting takes, editing, naming files, checking permissions, mixing, describing and distributing. A polished result can conceal these contributions. Research that treats the final sound as self-producing misses both the labour and the decisions that shaped it.
For music, separate a question about sonic mediation from a question about institutions and participation. How a broadcast edit changes the apparent audience response is a sound-studies question. How venues, education and markets distribute opportunities connects directly to Sociology of Music. Many projects need both, but each strand requires its own evidence.
For reusable recordings, the guide to sound libraries helps explain why descriptive records and clear usage terms matter. A searchable label also frames interpretation. Calling a file “empty street” may hide audible work, movement or distant voices that were considered unimportant by its cataloguer.
A useful exercise is to give one short recording three different captions, then ask readers what each caption encourages them to notice. Keep the audio unchanged and disclose the exercise afterwards. The result demonstrates the influence of framing within that exercise; it does not prove that captions control every listener or that recorded evidence has no independent constraints.
CHAPTER 9 OF 14 · Study public sound, media and archives
9. Preserve uncertainty alongside the recording
An archive is more useful when future users can distinguish the received source, preservation work and edited access versions. If every file is simply named “final,” important differences become difficult to recover. A research collection needs a modest, consistent record of identity and changes before it needs an elaborate public interface.
The Canadian Heritage Information Network’s audio digitisation guide distinguishes preservation files from production versions and cautions that noise reduction can remove signal. The practical implication is to preserve an unchanged reference and document processing applied to copies. A cleaner listening experience is not automatically a more complete historical record.
Worked case: a missing background voice
Imagine an interview with a quiet second voice behind the main speaker. An edited copy removes much of the background sound to improve intelligibility. A researcher using only that copy concludes that the interview was a private exchange between two people. The conclusion exceeds the evidence because the processing may have removed signs of other participation.
Repair the record by retaining the available source, the edited version and a short change log. Identify the time range affected and describe what the processing was intended to do. If the earlier source is unavailable, state that limitation instead of reconstructing a supposedly original sound from guesses.
Treat uncertain identification similarly. “Possible bell at 03:18” is more useful than a confident label added without evidence. If a later researcher identifies the source, record the new interpretation and its basis while retaining the earlier uncertainty. The archive then shows how knowledge developed rather than pretending that every label was certain from the start.
Decide what access the collection can responsibly support. Some recordings may be available only as descriptions, restricted excerpts or supervised listening because of participant agreements or other conditions. Public availability is not the sole measure of research value. A documented collection that honours its conditions can support better inquiry than an attractive audio gallery whose provenance, permissions and edits cannot be explained.
CHAPTER 10 OF 14 · Compare evidence and repair an argument
10. Compare sound environments with a fair denominator
Comparison is powerful when it states what is being held similar and what is allowed to differ. Two recordings made at different times, positions and durations can reveal different scenes, but they do not automatically measure a change in the same environment. Begin by specifying the quantity or relationship you want to compare.
Worked case: counting announcements
A fictional researcher counts six announcements in a ten-minute station recording and eight in a twenty-minute recording. The raw count rises from six to eight. The observed rate falls from 6 ÷ 10 = 0.6 announcements per minute to 8 ÷ 20 = 0.4 per minute. Counting more events does not necessarily mean a higher event rate.
Neither calculation establishes a typical daily rate. The windows may cover different services, disruptions or passenger flows. Define an announcement consistently, record the start and end of each observation, and explain whether a repeated message counts as a new event. Sample comparable periods across more than one day before extending the conclusion.
Now consider a different question: how much of each window contains announcements? Six long announcements might occupy more time than eight short ones. Event rate and time occupied are separate measures. If two messages overlap, adding their individual durations can exceed the duration during which any announcement is audible. State whether you count source-time or elapsed occupied time.
Suppose four non-overlapping announcements each last 15 seconds within a five-minute window. Total announcement time is 60 seconds out of 300, or 20%. If two of those announcements overlap completely, time with at least one announcement becomes 45 seconds, or 15%, while the sum of the four individual durations remains 60 seconds. The definitions determine the answer.
Finally, return to the listener’s task. A lower announcement rate may still coincide with worse information access if essential messages are less intelligible or arrive too late. Numbers discipline a claim; they do not choose the right question for you. Pair the calculation with an account of what people need to do and which part of that activity the measure actually captures.
CHAPTER 11 OF 14 · Compare evidence and repair an argument
11. Design a small inquiry that another reader can follow
Choose a project narrow enough to finish with care. One public announcement system, one family recording collection or one recurring neighbourhood sound is sufficient. A useful project has a clear question, a defined body of evidence, an explanation of selection and a conclusion that answers the question at the appropriate scale.
For a first study, compare two versions of a short recorded message whose use you are authorised to examine. Ask how the editing changes the apparent relationship between speaker and audience. Keep the received files, identify their sources and describe the differences you can demonstrate. Do not attribute intentions to an editor unless you have evidence about those intentions.
Make the research process visible
Write an observation log before drafting the argument. Record file identifiers and time ranges beside each claim. Separate direct observations, participant accounts, contextual documents and your interpretation. This simple separation makes it easier to discover when an argument rests on one vivid excerpt rather than on the full material considered.
Look deliberately for a difficult example. If most passages support your interpretation of formal authority, find a passage that sounds conversational and explain its significance. It may be an exception, a different communicative task or a sign that your initial distinction was too simple. An account becomes stronger when it can accommodate variation without explaining everything away.
Plan an accessible output at the start. W3C’s transcript guidance explains how text alternatives can carry speech and relevant non-speech information and how readers should be able to find them alongside media. For your project, identify speakers, explain meaningful sounds and label uncertain words. Avoid claiming that text reproduces every quality of the listening experience.
The University of Copenhagen’s Sound Studies Lab brings together field research, critical analysis and sonic artifacts. A student can use that combination on a manageable scale: a short documented recording or description, a careful argument and a clearly labelled creative response. The creative response can explore an idea, while the argument remains accountable to the evidence.
CHAPTER 12 OF 14 · Compare evidence and repair an argument
12. Diagnose and repair weak interpretations
A persuasive sound essay can still fail if it relies on untested assumptions. Before polishing its language, inspect the link between evidence and conclusion. The most useful repair is usually more precise than “add more research.” Identify which part of the argument is weak and seek the evidence that addresses that weakness.
When a sound seems to prove an emotion
Replace the emotion label with an observable feature: a long pause, a change in delivery, an interruption or a particular phrase. Then ask what contextual evidence supports the emotional interpretation. If alternatives remain plausible, present the interpretation as one possibility. Do not make certainty a stylistic requirement.
When the archive appears silent about a group
Investigate how the collection was formed, catalogued, preserved and made searchable. A missing recording may reflect non-recording, loss, access restrictions, misleading description or a search that used the wrong terms. The absence of a search result is not sufficient evidence that people made no sound or had nothing to say. Document the search and its limits.
When a comparison changes everything at once
List the differences: site, date, duration, microphone position, processing, listener task and playback conditions. Decide which differences are central to your question and which prevent a fair comparison. Narrow the claim if you cannot reconstruct the missing conditions. A comparison of two available recordings can still be worthwhile without becoming a claim about two entire societies.
When a beautiful presentation outruns the evidence
Check whether music, montage or captions make a contested interpretation feel settled. Keep reconstructed scenes, illustrative effects and recorded source material clearly distinguishable. If the original recording does not survive, a new performance may illustrate a hypothesis, but it cannot become a recovered document simply because it sounds convincing.
The repair test is practical: could another reader locate the evidence, understand the transformation and see why you reached the conclusion? If they could reasonably reach another conclusion, explain the disagreement rather than concealing it. Sound studies benefits from interpretation that is imaginative enough to ask useful questions and disciplined enough to show where its answers stop.
CHAPTER 13 OF 14 · Test transfer and continue learning
13. Transfer the method to a changed context
Use this exercise to check whether the method travels beyond familiar examples. A community group has a recording labelled “market morning, 1986.” It wants to compare it with a new recording and publish a claim that the market has become less sociable. The old file contains many voices; the new file contains fewer. What would you need to know before supporting that claim?
Start with provenance. Is 1986 the event date or a later estimate? Was the old recording made at the busiest entrance, beside a stall or during a special occasion? Is it continuous or edited? Was the new recording made at a comparable time and position? What activities were underway? A label gives a lead for inquiry, not a complete research design.
Work through the numerical trap
Assume the old recording lasts four minutes and contains 20 identifiable conversational exchanges. The new recording lasts twelve minutes and contains 30. Under the deliberately simplified assumption that exchanges can be counted consistently, the rates are five and 2.5 exchanges per minute. The observed rate is lower in the new file, even though its raw count is higher.
Now change the information: the older file was edited from a thirty-minute source to retain busy passages. Its four-minute duration is no longer a neutral observation window. The comparison cannot estimate a historical change in conversational activity. Correct arithmetic does not rescue a denominator created through selective editing.
The defensible answer is that the two available presentations contain different patterns of audible interaction. To investigate sociability, you need an explicit definition and additional evidence: comparable observations, accounts of how people use the market, changes in opening hours or layout, and attention to interactions that are quiet or inaudible from the chosen position.
This exercise connects to the How X Works hub: a mechanism should remain understandable when the surface example changes. The transferable skill is to trace how evidence was produced, match a measure to a question and revise the conclusion when a hidden condition changes. Apply the same discipline to a classroom, a workplace recording or a family audio archive.
CHAPTER 14 OF 14 · Test transfer and continue learning
14. Continue learning: useful questions and source routes
Do I need musical training to study sound?
No. A question about announcements, recorded speech, media editing or the history of a sound-producing device can begin without musical analysis. Learn the technical or disciplinary vocabulary your question requires. Avoid treating specialised language as a substitute for describing the evidence clearly.
Is sound studies the same as acoustics?
The fields overlap, but their central questions can differ. An acoustic investigation may examine how sound propagates through a room. A sound-studies investigation may ask how people use that room, what kinds of listening are expected and how recordings represent it. A strong project can combine them while keeping physical measurement and social interpretation distinct.
Can a memory of sound be useful if no recording survives?
Yes, as a recollection with its own circumstances and limits. Ask when and how the memory was elicited, what the person associates with it and which details other evidence can support. A remembered sound can reveal meaning and experience without providing an exact acoustic reconstruction.
Does an unedited recording give an objective account?
It gives a particular record whose capture conditions still matter. Position, duration, equipment and the decision to record all shape what is available. Keeping an unchanged source helps preserve evidence, but it does not turn one microphone into every possible listening position.
How should I describe something I cannot identify?
Describe the audible features and mark the identification as unresolved. Include a time reference and any plausible alternatives without presenting them as facts. This lets another person investigate the question while keeping your account honest about its limits.
What is a realistic next step?
Choose one short recording you can appropriately use. Produce a source note, a time-based description, a paragraph interpreting one feature and a paragraph testing an alternative explanation. Add a text route for readers who cannot or prefer not to use the audio. A small, traceable study is a sound foundation for a larger project.
Read the sources for the question you are asking
For disciplinary scope, start with Northwestern’s introduction and the Copenhagen Sound Studies Lab. For playback provenance, use the National Archives recording-speed entry. For preservation and edited copies, consult the Canadian Heritage Information Network guide.
For recorded interviews, read the Oral History Association’s principles. For accessible presentation, use W3C’s transcript guidance. These sources support different parts of the method; none supplies a universal interpretation of a sound. Your task is to bring a clear question, inspect the evidence and make the reasoning available to the next reader.