How I Improved my English Speaking in 3 Weeks

Improving English speaking in three weeks comes down to a simple pattern: message widely across countries, then turn those conversations into structured practice with pre-planned topics, targeted vocabulary capture, and visible progress tracking. Over a 28-day stretch, the learner reports adding 193 words—up from 143 words across a 365-day period—crediting the jump to speaking practice that creates “positive competition” with conversation partners. The motivation is practical: if a partner uses a word the learner doesn’t know, the learner is pushed to look it up and keep pace, spending more time speaking and learning rather than drifting into passive study.

The process starts with who to talk to. Practicing with compatriots—people who share the learner’s native language—tends to steer conversations toward the easiest option: switching to the native tongue when difficulties arise. Instead, the learner recommends pairing with people who don’t share the native language, so both sides must communicate in English. Language exchange can help, but it has drawbacks: it splits time between English and the native language, which the learner says makes practice less efficient. It can also be harder to get reliable answers from non-English native partners when questions require knowledge beyond a shared vocabulary subset.

To make conversations productive, messages sent to partners should be treated as learning material. When the learner encounters unfamiliar words or phrases in incoming messages, those items get checked in a dictionary and added to a flashcard system. That turns everyday chat into a steady stream of vocabulary targets.

Conversation sessions themselves should be pre-planned. Early sessions focused on food from each country, but later sessions required a system to avoid running out of topics. Without preparation, the learner says conversations often collapse into basic wording—such as searching for an ingredient name in Persian and showing a picture—rather than learning the English equivalent. Pre-planning is where tools come in: Google Gemini, Bing, and ChatGPT can generate translations and speaking prompts, but the learner emphasizes double-checking outputs for naturalness using dictionaries or additional searches.

Finally, progress needs to be tracked to sustain motivation. The learner uses Obsidian with a spaced repetition plugin to manage flashcards and monitor word growth over time, using the visible trend as proof that the method works.

For apps, two platforms get compared. High Local offers instant access to themed “tables” for listening and speaking without scheduling, but the learner criticizes frequent mismatches between advertised level/topic and what appears in practice, plus unpredictable discussions (including politics) and a lack of predefined structure. Speaky, by contrast, lacks tables and instead relies on filtering people by English level and other criteria, then messaging them to schedule conversations via Google Meet or Skype. The tradeoff is that many users appear to be there for dating or casual chatting rather than speaking practice, leading to low response rates. Still, when a real speaking partner is found, scheduled sessions deliver more speaking time and better learning than dropping into a table and waiting for others to talk.