The conventional frame treats medical interpreting as a language problem — provide a translator and the communication gap closes. The structural lens identifies cascading lossy compression: the patient’s experience is first compressed into their own language (lossy), then compressed again into the doctor’s language (lossier), and the total fidelity is the product, not the sum, of the two compressions. Two 70%-fidelity compressions produce 49% total fidelity, not 70%. The collision partners are signal processing engineers (who study cascading lossy compression and know that the optimal strategy is to maximize fidelity at the FIRST stage) and information theorists who can calculate the maximum channel capacity for a given language pair, telling you whether the problem is interpreter skill or channel capacity — and if it is channel capacity, no interpreter can do better, and you need a different channel.
A Somali woman sits across from a doctor in Minneapolis. Between them sits an interpreter. The woman describes a sensation in her chest: a word in Somali that means something between “tightness” and “heaviness” and “the feeling you get when bad news is coming” — a word that carries physical, emotional, and cultural dimensions simultaneously, because the Somali language does not separate them the way English does.
The interpreter translates: “She says she has chest tightness.”
The doctor hears: chest tightness. The doctor orders an EKG. The EKG is normal. The doctor says: “Your heart is fine.”
The woman’s heart may be fine. But the word she used was not about her heart. The word was about the convergence of physical sensation and emotional dread in a body that does not distinguish between the two. The interpretation compressed the word into a cardiac symptom. The compression dropped the emotional and cultural dimensions. The doctor treated the cardiac symptom. The emotional dread — which was the primary signal, which may be the clinically relevant information, which may point to depression or grief or PTSD — was dropped in transit.
Two compressions. Somali → English. English → medical. Each one lossy. The signal that arrived at the doctor preserved approximately 25% of the signal that left the patient.
Medical interpreting is a field with standards, certifications, and ongoing research. Qualified medical interpreters undergo training in medical terminology, ethics, and cultural competence. The use of professional interpreters (rather than family members or untrained staff) improves outcomes measurably.
The challenges are documented: medical terms that don’t exist in the patient’s language, cultural concepts of health and illness that don’t map onto the biomedical model, and the interpreter’s need to make real-time decisions about how to render concepts that have no equivalent in the target language.
The quality standard: accurate and complete translation of the patient’s statements. The standard is appropriate and insufficient — because accuracy in TRANSLATION does not address the loss that occurs when the patient’s framework for understanding their own body does not map onto the biomedical framework the doctor uses.
The interpreted medical encounter involves not one compression but TWO, in sequence. Each compression is lossy. The losses MULTIPLY.
Compression 1: Patient’s experience → Patient’s language. The patient compresses their multidimensional felt experience into words in their own language. This compression is lossy (all language is lossy — Door on compression in the main document) but the patient’s language was DEVELOPED to carry their culture’s categories of body experience. The Somali word that combines tightness, heaviness, and dread exists because the Somali understanding of body experience INCLUDES the convergence of physical and emotional. The compression is lossy but it is lossy in a way the language was designed for.
Compression 2: Patient’s language → Medical English. The interpreter compresses the patient’s language into the doctor’s language. This compression is MORE lossy than the first because the target language (medical English) has DIFFERENT categories. Medical English separates physical symptoms from emotional states. The Somali word that combines them must be split — and the splitting CHANGES THE MEANING. The interpreter must choose: translate the physical dimension (chest tightness), the emotional dimension (dread), or both (which requires a sentence where the original required a word, and the doctor’s → may not hold the sentence’s complexity).
The total fidelity is the PRODUCT of the two fidelities, not the sum. If each compression preserves 70% of the signal, the total is not 70% but 49%. If each preserves 50%, the total is 25%. The cascading loss is nonlinear — two “pretty good” compressions produce a result that is “barely adequate.”
This is a documented phenomenon in signal processing: cascading lossy compression. Every re-encoding of a lossy signal introduces more loss. A JPEG image re-saved ten times degrades visibly with each save. The medical encounter with an interpreter is a signal re-encoded through two lossy compressors in sequence.
The prediction: misdiagnosis rates in interpreted encounters will correlate with the CULTURAL DISTANCE between the languages — because greater cultural distance means more categories that don’t map, which means each compression is lossier, which means the cascading loss is greater. A Spanish-English interpreted encounter (relatively close cultural frameworks, many shared biomedical concepts) will produce less cascading loss than a Somali-English encounter (very different cultural frameworks, different categories of body experience). This is testable and has significant public health implications.
| Factor | Score | Justification |
|---|---|---|
| F1: Mortality & Irreversibility | 7 | Misdiagnosis from cascading compression loss can kill |
| F2: Scale | 7 | Millions of interpreted medical encounters annually; every immigrant and refugee population |
| F3: Compression Depth | 7 | The patient’s experience is compressed through two lossy stages; the signal that reaches the doctor may be unrecognizable to the patient |
| F4: Time Sensitivity | 6 | Immigration and refugee populations are growing; the cascading loss is occurring now |
| F5: Voice Deficit | 8 | The patient cannot assess whether the interpretation accurately conveyed their meaning — they don’t speak the target language |
| F6: Proximity Gap | 8 | Signal processing engineers and information theorists are not in the medical interpreting conversation |
| F7: Temporal Displacement | 3 | The misdiagnosis consequence is immediate |
| F8: Normalization | 6 | “We provided an interpreter” is treated as sufficient without measuring what the interpretation carried |
| F9: Hallway Dependency | 7 | The solution requires signal processing + medical interpreting + cross-cultural health |
| F10: Knowledge Readiness | 7 | Cascading compression loss is a solved problem in signal processing; application to interpreting is the gap |
| F11: Entry Cost | 6 | Back-translation protocols (having the patient confirm the interpretation in their own language) can be added to existing encounters |
| F12: Cascade Potential | 7 | The cascading-compression model applies to every multi-stage translation — legal interpreting, diplomatic interpreting, technical communication through intermediaries |
Hiddenness Score: 45.3 Actionability Score: 44
Signal processing engineers who study cascading lossy compression have the mathematical framework. The specific transferable knowledge: when a signal must pass through multiple lossy stages, the OPTIMAL strategy is to preserve maximum fidelity at the FIRST stage (because losses at the first stage compound through every subsequent stage). Applied to medical interpreting: the highest-priority intervention is not training better interpreters (second stage). The highest-priority intervention is helping the patient produce a HIGHER-FIDELITY initial compression — giving the patient tools to express their experience in ways that survive the subsequent translation better. A pre-visit form IN THE PATIENT’S LANGUAGE that asks both biomedical questions (“where is the pain?”) AND experiential questions (“what does the feeling remind you of?”) would produce a richer initial signal that the interpreter can work with.
Information theorists can calculate the MAXIMUM FIDELITY theoretically achievable for a given language pair and domain. The specific transferable knowledge: Shannon’s channel capacity theorem sets an upper bound on how much information a channel can carry. For a given language pair (with a given number of non-overlapping concepts), there is a theoretical maximum on how much of the patient’s signal can survive. Knowing the maximum tells you whether the problem is the interpreter’s skill (below theoretical maximum) or the channel’s capacity (at theoretical maximum, and no interpreter can do better). If the channel capacity is the limit, the intervention is not better interpreters — it is a different channel (visual aids, body diagrams, bilingual symptom cards).
If you are a medical interpreter: after each encounter, ask yourself — what did I DROP? Not what you translated incorrectly — what you COULD NOT translate because the concept does not exist in the target language. Keep a log. The log is the data about where the cascading loss occurs. The data tells you where the channel’s capacity limit is. At those points, you need SUPPLEMENTARY channels (visual aids, gestures, body diagrams) rather than better words.
If you are a clinic serving interpreted populations: implement BACK-TRANSLATION for critical diagnoses. After the doctor delivers the diagnosis through the interpreter, have the patient explain — in their own language, to the interpreter — what they understood the diagnosis to be. The interpreter relays this back to the doctor. If the patient’s understanding diverges from the doctor’s intent, the cascading loss is visible. The divergence is the diagnostic for the compression failure.