Synthesis: Writing From Multiple Sources
C1 · Writing Lab · about 45 minutes
1 · Learn
Synthesis is not summary in a trench coat. The test is simple: if your paragraphs could be rearranged by source without loss, you have summarized; if every paragraph is organized by an IDEA and the sources enter as witnesses who agree, qualify, or collide, you have synthesized. Matrix first: rows = sources, columns = the questions that matter; the empty cells and the contradictions are where your essay lives.
Attribution is the ethics of the craft. Every borrowed claim carries its owner (“For Chen, the driver is cost; Alvarez dissents…”); paraphrase must change structure, not just synonyms; and your own voice — the synthesis proper — should be identifiable as the one arranging the furniture, not hiding under it.
2 · See an Example
Synthesis: Why Do Startups Cluster? (excerpt)
Ask why startups cluster in a few expensive cities when the technology they sell promises placelessness, and the literature offers three answers that sound like rivals and behave like layers. The oldest is economic: Saxenian's comparison of Silicon Valley and Boston's Route 128 (1994) locates the advantage in networks — engineers who change jobs without changing neighborhoods carry knowledge with them, so the region learns faster than any firm. A second tradition, labor economics, points to thickness: in a deep market, a failed founder is re-employed by Friday, which quietly lowers the personal price of risk (Glaeser, 2011). The third, most recent account stresses signaling: an address in a known cluster certifies a young company to investors who cannot yet judge its product (Nanda & Rhodes-Kropf, 2013).
Read together, the three accounts describe one mechanism at three altitudes. Networks move knowledge, thickness insures failure, and signals convert both into capital — and the cluster is simply where all three compounds interest fastest. The dissent in the literature is real but narrower than it looks: remote-work studies (e.g., Bloom et al., 2015) show individual productivity surviving distance, yet none measures the regional learning that Saxenian described — the variable on which the whole cluster effect is argued to rest. The open question is thus not whether work can leave the cluster, but whether learning can.
3 · Analyze the Example
- Question first, sources as witnesses. The paragraph opens with the QUESTION (why cluster?) — each source enters to answer it, in one clause each, with dates. No source gets its own paragraph.
- Paraphrase with structure changed. Saxenian's 300-page comparison becomes “engineers who change jobs without changing neighborhoods carry knowledge with them” — the idea digested, not decorated.
- Synthesis as altitude. “Three altitudes of one mechanism” — the writer's own contribution is the RELATION between sources, something no source says alone.
- Dissent weighed by its variable. The remote-work studies are not dismissed; they are located — they measure productivity, not regional learning. Friction resolved by precision, the synthesis way.
Useful Vocabulary
Synthesize the three accounts.
Alvarez dissents on this point.
Tech firms cluster in a few cities.
One mechanism at three altitudes.
Paraphrase the claim, then cite it.
Useful Expressions
- Ask why …, and the literature offers …
- Read together, the accounts describe …
- … is real but narrower than it looks.
- The open question is thus not … but …
- For X, the driver is …; Y dissents.
Common Mistakes
✗ Patchwriting: taking a source sentence and swapping synonyms (“the region learns faster” → “the area gains knowledge quicker”) while keeping its skeleton.
✓ Close the source. Write the idea from your matrix notes. Then open the source and check you kept the meaning — and lost the skeleton.
Synonym-swapping is plagiarism with a thesaurus. Structural paraphrase — different sentence, same claim — is the actual skill being examined.
✗ Quote stuffing: six quotations in a page, connected by “Another scholar says…”
✓ Quote only what loses power in paraphrase (a definition, a verdict, a phrase under analysis). Everything else: paraphrase, attribute, move on.
A synthesis should be 90% your sentences carrying their claims. Quotations are exhibits, not the argument.
✗ False balance: “Scholars disagree about whether smoking is harmful, with one 1970s industry paper dissenting.”
✓ Weight the witnesses: “The consensus is X; a minority tradition, largely industry-funded, maintains Y — a distribution the reader deserves to know.”
Synthesis reports the SHAPE of agreement, including its lopsidedness. Equal time for unequal evidence is its own distortion.
4 · Guided Practice
Build the matrix row: Source = a mayor's speech claiming bike lanes “gave the city back to its people”. Columns: Claim / Evidence offered / What it assumes.
See a model answer
Claim: bike lanes return urban space to residents. Evidence offered: photographs, two shopkeepers' quotes. Assumes: that street space 'belongs' to residents rather than through-traffic — a premise a synthesis would name.
Paraphrase structurally: “Knowledge is carried between firms by engineers who change employers without relocating.”
See a model answer
Example: “When engineers switch jobs locally, what they know travels with them — so the region, not just the new employer, gets smarter.” (Different skeleton, same claim.)
Stage the friction in one sentence: A finds remote work raises productivity; B finds it lowers innovation.
See a model answer
Example: “Where A measures output per worker rising at a distance, B measures the collisions that produce new products falling — a disagreement that dissolves once 'work' is split into execution and invention.”
5 · Write — Your Task
✍️ Choose a question with at least three findable positions (a debate in your field, a local controversy, a question from your studies) and write a synthesis (300–380 words): a matrix in your notes; paragraphs organized by ideas; each claim attributed; one passage staging a real friction between sources and resolving or precisely locating it; a closing sentence that is yours alone.
Goal: 3–4 paragraphs + matrix notes kept · Length: 300–380 words
- If a paragraph can be labeled with one source's name, rewrite it until it can't.
- Let verbs carry attribution (argues, documents, dissents) — they are also your evaluation.
- Your own voice appears most clearly in the relations you draw between sources.
6 · Check — Self-Check
Work through your checklist honestly, then run the Self-Check. It counts words and sentences and spots mechanical points — it does not grade your English. You are the editor.
Read a model answer for the task
Why do some cities' bike lanes fill while others stand empty? The planning literature answers at street level: lanes succeed where they form a NETWORK — a lane to somewhere (Pucher & Buehler, 2008). Transport economists answer at wallet level: cycling rises where driving is priced honestly, through parking and congestion charges (Shoup, 2005). A third, sociological account answers at mirror level: people cycle where cycling looks normal, which is to say where people like them are seen cycling (Aldred, 2013). Staged together, the three are a sequence, not a quarrel: pricing makes the first riders rational, networks make them safe, and visibility makes them unremarkable — after which the lanes fill 'by themselves', and the city congratulates its culture for what its prices began.
7 · Improve & Rewrite
Weak version — rewrite it in your notebook first, then compare:
According to Smith (2018), sleep is important for memory. According to Lopez (2019), students who sleep more get better grades. According to Kim (2021), phones at night are bad for sleep. In conclusion, sleep is very important for students and they should sleep more and use phones less.
See an improved version
Three findings, often cited separately, compose a single argument when read together. The mechanism is laboratory-solid: memory consolidates during sleep, with the deepest gains in the hours students most often surrender (Walker, 2017). The behavioral evidence agrees in direction — later school starts, which simply return those hours, raise both sleep and grades (Wheaton et al., 2016). The third finding explains why information alone changes nothing: the surrendered hours are not lost to laziness but bought, nightly, by devices engineered against sleep (Twenge, 2017). The synthesis writes the policy by itself: you cannot educate a student out of a market. Either the hours are protected structurally — start times, platform design — or the consolidation literature remains a description of what other people's children get.
Quick Quiz
Keep going
📘 Grammar deep-dive: Reported Speech Advanced: Reporting Verbs🌱 Vocabulary Builder: the B1 Learning Goals set🎓 TOEFL Writing: the exam version of this skill← Previous lesson · C1 — all lessons · Next: Formal Reports at C1: Analysis, Caveats, and Decisions →
Ad space