The corrected figure: monitoring self-repair in HL listening
Twenty-two students sit a mock Paper 2 listening. Track 2 is a three-minute radio interview with a marine geologist from Galicia about beach erosion. At 1:12 the interviewer says «un 40 % de las playas»; at 1:40 the geologist corrects herself: «perdón, hablé de un 40, pero la cifra actualizada es el 45 %». She names two causes, one of which she later calls «secundaria». Students hear the track once, as in the real exam. Question 6 asks for the percentage of beaches affected according to the scientist (1 mark); question 7 asks candidates to state the main cause and explain why the second cause is mentioned (3 marks). Marco writes 40 % and, for question 7, lists both causes as equally important. The transcript shows the speaker also uses «o sea» and «es decir» before restating, and her /s/ is frequently aspirated. Marco scores 9/20 overall, with most losses in the second half of the paper.
1. What distinguishes Marco's 40 % from guesswork, and what does the pattern of his errors reveal about how he listens?
Read the worked answer
He heard the number. 40 % is a correct transcription of what was said at 1:12; this is not a decoding failure. The error is strategic: he treated a first figure as final and stopped monitoring. The speaker flags the repair twice - perdón and la cifra actualizada - and both o sea and es decir introduce restatements later in the track.
What it reveals. Marco writes content words the moment he hears them, which consumes working memory and leaves nothing for monitoring whether the message is later revised. HL listening texts are selected precisely because information is hedged, revised or contradicted, so a candidate who cannot hear repair loses the denser second half of the paper - which matches his 9/20 profile exactly. The misreading of question 7 has the same root: secundaria is an evaluation by the speaker, not a label to be copied onto a list, and he flattened it.
The accent question. Aspirated /s/ is a real decoding challenge (loh datoh for los datos), but it cannot explain this error, because he got the number right. Before spending lesson time on phonology, check whether he catches repairs in a different, clearer accent. If he does, the problem is strategic and phonological work would be wasted. If he does not, he needs bottom-up drilling of repair phrases in fast speech.
2. How would you use this track in class to build repair-monitoring, without turning listening into transcription practice?
Read the worked answer
Stage 1 - orientation before playing (5 min). Show only the question stems and the topic. Students predict what kinds of numbers and evaluations a scientist might revise. This creates slots to listen into, so they are not writing blindly.
Stage 2 - first listen for stance, not data (5 min). Who is more cautious, interviewer or scientist? Where does the scientist disagree with the premise? No numbers written. This trains the habit of hearing the interview's shape before hunting for facts.
Stage 3 - focused listen on the repair window (10 min). Play only 1:05–1:50 three times. Students note the number, then note the change. Establish a note convention that preserves uncertainty: figures written with «?», a struck-through number, a two-column «dicho / corregido» layout. Explicitly list repair markers: perdón, mejor dicho, más bien, la cifra actualizada, o sea, es decir, lo que quiero decir.
Stage 4 - transfer (15 min). A different track in a different accent in which a date, a place and a cause are each revised once. Success criterion is not «all answers right» but «every revision marked in the notes».
Why this order matters: isolating the repair window teaches the marker; the transfer track tests whether the marker is now heard in flow, which is the only evidence that matters.