Notes · Hebrew (Modern)UK · A-Levels

הבנת הנשמע: Paper 3, individual control of the recording, and a language you cannot see

This topic is Paper 3 of AQA 7672: 2 hours 30 minutes, 75 raw marks scaled to 150, 37.5 per cent of the qualification. It has three parts - listening and responding for 35 marks, a translation from English into Hebrew for 10 marks, and a 30-mark multi-skill task combining a heard source and a written source in about 200 words. The single most important structural fact about the listening section is that the recordings are studio recorded, provided to schools and colleges, and each student has individual control of the recording. That turns the section from a test of instantaneous comprehension into a test of method: you decide how many times you hear each passage and where you stop it. At least one passage involves more than one speaker. One question is a summary in Hebrew, and there the content mark is AO1 and the language mark AO3, awarded independently, exactly as in the Paper 1 reading summary. Underneath all of it sits a problem specific to this language: Hebrew spelling does not record its vowels and modern Hebrew pronunciation has merged several pairs of letters, so the relationship between the word you hear and the word you can write is looser than in most languages you might have studied.

5 sections~30 min reading time4 competenciesLevel Foundation 1 · Standard 1 · Advanced 3

T·0333 / 13

Exam profile

AO1 · Understand and respond in writing to spoken Hebrew drawn from a variety of sources - 20 per cent of the qualification, all of it collected on Paper 3AO3 · Manipulate the language accurately in written form: the listening summary carries 5 marks for quality of language, awarded independently of contentAO2 · Understand and respond in writing to written Hebrew: the multi-skill task requires a written source to be held alongside the heard oneInfer meaning from complex spoken material, including opinions, views, emotional reactions and personal experiences, and from passages with more than one speaker

Operators:identifyinfersummarisecompareevaluateconclude

basic level

In the first year the priority is exposure rather than technique: listen to Hebrew every day, ideally to material you have a transcript for, and use the transcript afterwards rather than during, so that the ear does the work first and the eye only confirms it.

higher level

In the second year turn exposure into a rehearsed three-pass routine on a controllable recording, and add the two things the grids specifically reward: a summary written to about 95 words with content and language attended to separately, and a multi-skill answer that genuinely holds two sources together rather than summarising them one after the other.

Depth

Text

Contents · 5 sections

הבנת הנשמע: Paper 3, individual control of the recording, and a language you cannot see

שליטה עצמית בהקלטה: the format, and why individual control changes everything#

FoundationAQA 7672 - 4.6 Paper 3: Listening, reading and writingAQA 7672 - 4.6.1 Listening and responding

What the ear cannot separate, and what decides it

Fig. 1Modern Israeli pronunciation has merged several pairs of letters that the script still distinguishes. The consequence for listening is that a heard word often gives you no clue how it is spelled - which matters because you must write your answers in Hebrew. Transliteration and IPA-free description only, because the figure engine has no right-to-left handling; the Hebrew letters are named in the prose.

Paper 3 carries 75 raw marks in 2 hours 30 minutes: listening and responding for 35, translation into Hebrew for 10, and the multi-skill task for 30. The listening passages come from a range of contexts and sources covering different registers and types, their content is drawn from the specification's four themes, and at least one of them involves more than one speaker. Questions target main points, gist and detail, they require inference, and they include abstract material - opinions, views, emotional reactions, personal experiences. They are set in Hebrew and answered either non-verbally or in Hebrew, and outside the summary the specification tells you to give only the information the question asks for.

Individual control of the recording is the fact to build your whole method around. You are not listening to a room-wide playback that runs once; you have the recording and you decide when to start it, when to pause it and how many times to play it. This means a passage you did not follow is not lost, and it means the useful question is no longer can I understand this at speed but how do I spend my plays. It also means the failure mode changes: candidates rarely fail this section by missing something irretrievably, and often fail it by replaying one difficult passage eight times and arriving at the last passage with no time to play it once.

The routine that uses that control well is three passes with different jobs. First pass: no writing at all, just listen for the shape - how many speakers, what the topic is, whether they agree, where the passage turns. Second pass: with the questions in front of you, stopping after each answer and writing it. Third pass: continuous, checking only the answers you were unsure of and listening for the detail you know you missed. The temptation is to start writing during the first pass, which is exactly what stops you hearing the shape, and the shape is what makes the second pass fast.

The multi-speaker passage deserves a separate preparation, because it is the one that most often goes wrong. Two speakers means two positions, and the questions will usually ask you to attribute one - who thinks what, where do they differ, what does the second speaker concede. So on the first pass of a multi-speaker passage the only note worth making is a two-column sketch with a name or a letter at the head of each column. Everything after that is attribution, and attribution is impossible to reconstruct later if you did not separate the voices on the way through.

Worked example

Planning the plays before the first passage starts

Paper 3 sets several listening passages of different lengths and one of them has two speakers. You have individual control. Build a play budget and a rule for when to break it.

Count what you have before you start anything
Look at the whole listening section first: how many passages, how many marks on each, which one is flagged as multi-speaker by having two names or two sets of questions. Divide your listening time by the marks, not by the passages. A 10-mark passage deserves twice the time of a 5-mark one however long the audio is.
Assign plays, not minutes
Three plays per passage is the default: shape, answer, check. Give the multi-speaker passage four, because the extra one buys the attribution pass. Give the shortest, most factual passage two, because the third pass on a factual passage almost never changes an answer. Write the numbers next to each passage before you press play.
Use the pause deliberately, not reflexively
On the second pass, stop after each answer rather than each sentence. Stopping every sentence destroys the discourse and makes inference questions harder, because Hebrew signals its argument with connectives at clause boundaries - אבל, לעומת זאת, מנגד, לכן - and you need to hear what follows them.
Set the break rule in advance
The rule: if a passage has taken more than its assigned plays and you have fewer than half its marks, move on and come back only if the whole section is finished. This is the same protection the reading section needs, for the same reason. Deciding it now converts a panicked judgement into an executed plan.
Reserve one play for the end
Keep enough time for one final play of the passage you were least sure about. That last play is worth more than a fourth play in the moment, because by then you know what you missed and can listen for it specifically instead of listening again in general.

Result: A written play budget beside each passage, four plays for the multi-speaker one, two for the shortest factual one, pauses at answer boundaries rather than sentence boundaries, a pre-set break rule, and one play held in reserve.

Typical mistakes

Writing during the first play, and therefore never hearing the passage as a whole.
Spending an unplanned number of replays on the hardest passage and running out of time for a later one.
Failing to separate two speakers on a multi-speaker passage and being unable to attribute a view afterwards.
Answering with everything you understood rather than with what the question asked.

Active revision

מצאו הקלטה בעברית שיש לה תמלול, והאזינו לה שלוש פעמים לפי השיטה: פעם אחת בלי לכתוב, פעם שנייה עם עצירות ותשובות, ופעם שלישית ברצף. רק אחר כך פתחו את התמלול, וסמנו בו את המקומות שבהם טעיתם. כתבו לכל טעות מה גרם לה: מילה לא מוכרת, מבנה תחבירי, או קצב דיבור

Active recall

Recall the key points — then reveal.

Sources: AQA A-level Modern Hebrew (Listening, Reading, Writing) 7672 specification (AQA)

מה שומעים ומה כותבים: from a heard word to a spelled answer#

AdvancedAQA 7672 - 3.3.1 The scripts and phoneticsAQA 7672 - 3.3.21 PhoneticalAQA 7672 - 4.6.1 Listening and responding

In this examination you hear Hebrew and answer in written Hebrew, so every listening question is also a spelling question, and modern Israeli pronunciation has removed several of the distinctions the script still makes. The pharyngeal ח and the fricative כ are both heard as the same rough sound; ט and ת are both t; ק and hard כ are both k; ס and ש in its sin reading are both s; ו and soft ב are both v; and א and ע are both silent placeholders. You cannot hear which letter a word is written with. What you can do is know the root, because the root fixes the spelling - which is why vocabulary learned as root families rather than as isolated words pays for itself twice, once in reading and once here.

The specification tolerates some of this: in comprehension questions carrying no AO3 marks, minor spelling errors which do not distort the meaning will be tolerated. That is a real safety net and you should not write less because you are unsure of a letter. But notice the limit - a spelling error that produces a different word does distort the meaning, and Hebrew's mergers are exactly the environment where that happens. Writing תיק with a tav where the word needs a tet is a minor error; writing עלה where you meant אלה changes the word. The practical rule is to write the answer rather than leave it blank, and to spend your spelling attention on the words where a merged letter would produce a real alternative.

Two structural features of spoken Hebrew regularly hide information from a listener who is thinking in English. The first is that the definite article is absorbed by the inseparable prepositions: ב plus ה becomes ba, ל plus ה becomes la, כ plus ה becomes ka. So a single heard syllable can carry both in and the, and whether a heard phrase was definite is a question you often have to answer from the sentence rather than from the sound. The second is that the present tense is a participle with no person marking, so ani kotev and ata kotev differ only in the pronoun; if the pronoun is unstressed and fast, the person can be genuinely ambiguous and must be recovered from the discourse.

Stress carries less information than in most European languages, and the specification's own 3.3.21 flags it. Standard Hebrew stresses most words on the final syllable, but ordinary speech frequently moves it earlier, particularly on names and on borrowed words, and there are minimal pairs where the stress is the only difference - the specification prints three of them in transliteration. For a listener the practical conclusion is that stress is weak evidence: use the consonant skeleton and the syntax to identify a word, and use stress only to confirm. A candidate who has learned words from a stress-marked textbook and then meets fast speech will misrecognise words they know well, and the cure is exposure to unscripted audio rather than more vocabulary.

Worked example

Recovering a heard word you cannot spell

In a listening passage about migration you hear a word that sounds like klita and you have to use it in a written answer. Work out what to write.

Isolate the consonant skeleton from what you heard
k, l, t, then the feminine ending. The k could be kuf or hard kaf. The t could be tet or tav. That gives four possible spellings before you have thought about meaning at all, and this is the ordinary situation, not an unusual one.
Use the context to fix the meaning
The passage is about people arriving in Israel and being settled. The sense is absorption or reception. Once you have the meaning, the root is available: ק-ל-ט, receive, take in - the same root as מקלט, a shelter or a receiver, and as קולט, taking in. Now the k is kuf and the t is tet, both fixed by the root and not by the sound.
Confirm the pattern before you write
The word is a verbal noun of the pattern that ends in the feminine ending, so קליטה, klita, absorption. The pattern also tells you the gender, feminine, which matters immediately because any adjective or verb you attach to it must agree. Writing קליתה, with a tav, would be a different letter but arguably a minor error; writing כליטה with a kaf would put the word in a different root family altogether and could be read as distorting the meaning.
Harvest the family while you are there
From the same root: קולט, receiving; נקלט, was absorbed; מרכז קליטה, an absorption centre, a construct chain; קולטן, a receptor. Learning the family costs almost nothing extra at this moment and means the next occurrence of any of them is recognised instantly.
Generalise the routine
Skeleton from the sound, meaning from the context, root from the meaning, spelling from the root, gender from the pattern. Five steps, all of which run in a couple of seconds once practised, and none of which requires you to have heard the letters - because you cannot hear them.

Result: קליטה, correctly spelled with kuf and tet because the root was recovered from the meaning rather than from the sound - plus four related words from the same family, obtained free.

Typical mistakes

Leaving a comprehension answer blank because of uncertainty about a letter, when minor spelling errors are tolerated there.
Learning vocabulary as isolated words rather than as root families, and therefore having no way to reconstruct a spelling from a sound.
Missing a definite article because it was absorbed into a preposition, and mistranslating a specific reference as a general one.
Relying on stress to identify a word and misrecognising familiar vocabulary in fast speech.

Active revision

האזינו לשלוש דקות של דיבור עברי לא מתוכנן, וכתבו כל מילה שלא הצלחתם לזהות בשמיעה. אחר כך מצאו את המילים בתמלול, וסדרו אותן בשלוש קבוצות: מילים שלא הכרתם כלל, מילים שהכרתם אבל לא זיהיתם בגלל הקצב, ומילים שהכרתם אבל לא ידעתם לאיית. לכל קבוצה יש תרופה אחרת

Active recall

Recall the key points — then reveal.

Sources: AQA A-level Modern Hebrew (Listening, Reading, Writing) 7672 specification (AQA)

סיכום הנשמע: the listening summary and its two grids#

AdvancedAQA 7672 - 4.6.1 Listening and respondingAQA 7672 - 4.6.2 AssessmentAQA 7672 - 4.7.3.1 Paper 3 Summary question: AO3

One question in the listening section is a summary in Hebrew of what you have understood from a passage you heard, and its mark scheme is built exactly like the Paper 1 reading summary with one substitution: the content mark is AO1 rather than AO2, because the source was spoken. Content and language are awarded independently. You must write in full sentences and use your own words as far as possible. A long summary is marked for content, or for language, only until the first natural break - usually the end of a sentence or main clause - between 90 and 100 words. Short summaries carry no automatic penalty but are self-penalising because they cannot contain all the content points.

The lifting rule applies here too, and its wording is adapted to a heard source: the AO1 mark is awarded for content points which contain the required information regardless of whether they are in your own words or partly or wholly lifted from the recording, but no AO1 mark is given where the answer includes irrelevant material or inappropriate information from the stimulus, and lifted language will not be eligible for credit when the AO3 mark is awarded. Lifting from a recording is harder than lifting from a text - you have to have written the phrase down as you heard it - which is one reason a listening summary is more often in the candidate's own words by accident than by design.

What makes this harder than the reading summary is that the source is not in front of you when you write. The consequence is that your notes are the summary's raw material, and note-taking has to be designed for it. Take content-point notes rather than transcription notes: one line per idea, in Hebrew where you can and in English where you cannot, with a mark against any point you think is central. Do not attempt to write down sentences; you will lose the next idea while writing the last one. On the third pass, fill the gaps in the notes rather than re-listening to the whole.

The AO3 grid is the same 5-mark scale as on Paper 1, and the same phrase governs it: manipulate complex language accurately where required by the task. In a listening summary the task usually requires reported speech, attribution and contrast, so the structures that earn the mark are predictable. Learn a small set and use them: לדברי, according to; לטענת, according to the claim of; המרואיין ציין ש, the interviewee noted that; לעומת זאת, by contrast; מנגד, on the other hand; שני הדוברים מסכימים ש, both speakers agree that. Six phrases, all safe, all repeatedly useful, and all of them evidence of the manipulation the grid describes.

Worked example

From two-column notes to a 95-word attributed summary

A two-speaker passage discusses whether municipal libraries should open on Shabbat. Speaker A is a councillor, speaker B a resident. Build notes and write the summary.

First pass: shape only, two columns headed
Write A and B at the top of two columns and nothing else during the first play. Listen for: how many positions, do they agree anywhere, where does the passage turn. A passage of this kind almost always has a concession in it - a point one speaker grants the other - and finding it on the first pass is worth more than any three details, because the summary's structure will hang on it.
Second pass: one line per idea, in the right column
Under A: access for people who work six days; the library is publicly funded so should serve all. Under B: staff working conditions; the character of the day in the neighbourhood; a compromise about limited hours. Note that B's third point is the concession. Write in whichever language comes fastest - notes are not marked.
Decide the structure before writing any Hebrew
Not the order in which it was said, but the order that makes an argument: A's position, A's reason, B's objection, B's concession, and a closing sentence naming what they share. That is five sentences at about nineteen words, which lands at about 95.
Write with attribution carrying the complexity
Open with לדברי חבר המועצה, according to the councillor. Turn with לעומת זאת, טוענת התושבת ש, by contrast the resident claims that. Concede with עם זאת היא מוכנה, however she is willing. Close with שני הדוברים מסכימים ש, both speakers agree that. Four attribution structures, each of them a subordinate clause, which is exactly the complex language the AO3 grid describes.
Run the checks and stop
Agreement: חבר המועצה is masculine so the verb is masculine; התושבת is feminine so טוענת and היא are feminine. Definiteness: הספרייה, המועצה, השכונה each carry the article. Et: only before definite direct objects. Count the words; if you are at 88 and mid-sentence, finish it and stop.

Result: A five-sentence summary at about 95 words carrying four different attribution structures, in which every view is assigned to the speaker who held it and the concession is placed where it does structural work rather than where it happened to be said.

Typical mistakes

Trying to transcribe the passage instead of noting content points, and losing three ideas while writing one sentence.
Writing a summary that reports the passage in the order it was spoken rather than in the order that makes an argument.
Attributing a view to the wrong speaker in a multi-speaker passage because the notes did not separate the voices.
Exceeding 100 words and losing the content points that were saved for the end.

Active revision

האזינו לראיון בן ארבע דקות עם שני דוברים, וכתבו הערות בשתי עמודות, אחת לכל דובר. אחר כך כתבו סיכום בן תשעים וחמש מילים שמייחס כל דעה לדובר שאמר אותה, ומשתמש בשלוש מילות ייחוס שונות לפחות. סמנו בסוף כמה נקודות תוכן נכנסו לסיכום

Active recall

Recall the key points — then reveal.

Sources: AQA A-level Modern Hebrew (Listening, Reading, Writing) 7672 specification (AQA)

המשימה המשולבת: holding a heard source and a written source together#

AdvancedAQA 7672 - 4.6.4 Multi-skill taskAQA 7672 - 4.7.3 Paper 3 assessment criteria

The multi-skill task is worth 30 of Paper 3's 75 marks and it is the only task in the qualification that requires you to combine two sources of different kinds. A listening passage and a written source are set on one of the themes; you listen, you read, and you then answer an unseen question in Hebrew in about 200 words. The specification names four things the response must do: analysis, evaluation, personal reaction and the drawing of a conclusion - and it adds, in its own words, that all of these require higher level skills. That sentence is a description of the mark scheme, not decoration: an answer that summarises both sources without evaluating them is doing one of the four jobs.

The commonest structural failure is a two-paragraph answer in which paragraph one reports the recording and paragraph two reports the text. Nothing in that answer holds the sources together, and the grids reward the holding. The alternative structure is thematic: identify two or three points on which the sources speak to each other - agreement, disagreement, one supplying what the other omits, one qualifying the other - and give each of those a paragraph in which both sources appear. That single change of shape does more for this answer than any amount of extra vocabulary.

Personal reaction is a specified requirement and it is not the same as opinion. What earns credit is a reaction that is argued from the two sources you have just been given, not a view you brought into the room. The safe form is a sentence that says what the two sources together lead you to think and why: לדעתי, הראיון מלמד ש, ואילו הכתבה מראה ש, ולכן נראה לי ש. That construction is doing the personal reaction and the conclusion at once, it is attributable to the sources, and it cannot be accused of ignoring them.

Two hundred words is short for four jobs, so the plan matters more than the prose. A workable shape is: one sentence naming the question and the two sources; one paragraph on the first point of contact between them; one paragraph on the second; one short paragraph carrying the personal reaction and the conclusion together. That is about four sentences of thirty-five to fifty words each. Write the plan in five lines of notes before you write any Hebrew - and note that Paper 3 gives you 2 hours 30 minutes for 75 marks, so this 30-mark task deserves close to an hour, which is far more planning time than candidates usually allow themselves.

Worked example

Planning 200 words that genuinely hold two sources together

The heard source is an interview with a school head about civics teaching. The written source is a short article reporting a survey of student attitudes. The question asks whether schools succeed in preparing young people for political life. Plan and draft.

Find the contacts, not the contents
Do not list what each source says. Ask instead: where do they meet? Contact one - the head describes what the school does, the article reports what students say they take away, and the two do not match. Contact two - both mention something outside school, the head in passing and the article at length. Two contacts, so two body paragraphs.
Decide what each contact lets you argue
Contact one supports an evaluation: an account of provision is not evidence of effect, and the article supplies the missing evidence. Contact two supports a qualification: if what shapes attitudes is largely outside the classroom, the question as posed may be aimed at the wrong institution. Notice that both of these are judgements about the sources, which is what evaluation means here.
Write the opening sentence so it does work
One sentence naming the question and both sources, with attribution built in: לפי הראיון עם מנהל בית הספר ולפי הכתבה על סקר התלמידים, אפשר לבחון את השאלה משני כיוונים שונים. That sentence has already established that two sources exist and that they will be used against each other.
Build each body paragraph around both sources
Paragraph one: המנהל מתאר, the head describes; לעומת זאת, עולה מן הסקר ש, by contrast the survey shows that; ומכאן ש, and from this it follows that. Paragraph two the same shape with different connectives. Each paragraph contains both sources and one conclusion of your own - that is the analysis and the evaluation.
Land the personal reaction and the conclusion together
לדעתי, שני המקורות יחד מלמדים ש, in my view the two sources together show that - then a claim, then a reason drawn from the sources. This is the personal reaction, argued rather than asserted, and it doubles as the conclusion. Keep it to two sentences: at 200 words you cannot afford a separate concluding paragraph, and the grid does not require one.

Result: A four-part plan of about 200 words in which each body paragraph contains both sources, evaluation is carried by two named judgements about what the sources can and cannot show, and the personal reaction and conclusion share the final two sentences.

Typical mistakes

Writing one paragraph per source and never bringing them into contact.
Summarising rather than evaluating, so the answer contains no judgement at all.
Offering a personal opinion that is unconnected to either source.
Running out of words before the conclusion, which is the sentence the grid most obviously looks for.
Under-planning a 30-mark task because it is only 200 words.

Active revision

בחרו נושא אחד מן התוכנית, מצאו עליו הקלטה קצרה וכתבה קצרה, וכתבו תשובה בת מאתיים מילים שמחזיקה את שני המקורות יחד. בנו אותה סביב שתי נקודות מפגש ולא סביב שני המקורות, וסיימו בפסקה שמביעה עמדה אישית שנשענת על שניהם. סמנו בסוף היכן בדיוק נמצאים הניתוח, ההערכה, התגובה האישית והמסקנה

Active recall

Recall the key points — then reveal.

Sources: AQA A-level Modern Hebrew (Listening, Reading, Writing) 7672 specification (AQA)

משלבים והתאוששות: registers, speakers and how to recover from losing the thread#

StandardAQA 7672 - 4.6.1 Listening and respondingAQA 7672 - 3.1 and 3.2 Themes

The specification promises spoken passages from a range of contexts and sources covering different registers and types, and the register range in Hebrew is wide. A radio news bulletin uses the passive binyanim, impersonal plurals and dense construct chains, and delivers them fast. An interview uses ordinary speech: shorter clauses, שהוא and שזה everywhere, false starts, and the future used where English would use an imperative. A studio discussion adds interruption and overlap. A recorded announcement or advertisement uses a compressed register with imperatives and slogans. Practising on one of these and meeting another in the exam is the most avoidable failure in this section.

Some features of ordinary Israeli speech are worth knowing because they change what you hear. Speakers routinely use the future form as a polite request, so תגיד לי is tell me rather than you will tell me. They front topics for emphasis, so a sentence may begin with its object. They use אז and ככה and בקיצור as discourse markers with almost no lexical content, and a listener who tries to translate them loses the thread while the sentence moves on. They contract: מה זה becomes something closer to one syllable, and אני לא becomes a run. None of this appears in a textbook dialogue and all of it appears in an authentic recording.

Losing the thread is normal and recoverable, and the recovery routine should be rehearsed rather than improvised. When you realise you have lost it, do not go back mentally to reconstruct - that guarantees you lose the next sentence too. Instead, listen forward for the next structural marker: a connective such as אבל or לכן or לעומת זאת, a proper noun, a number, or a change of speaker. Rejoin there. On a controlled recording you can then rewind and fill the gap afterwards, which is precisely what individual control is for, and which turns a catastrophic loss into a five-second replay.

Preparation should therefore be measured in the variety of what you have heard rather than the hours. A realistic weekly diet for the second year: one news bulletin, one long-form interview or podcast episode, one piece of unscripted conversation, and one piece of scripted material with a transcript. The transcript is for afterwards, never during - the ear must do the work first, or you are practising reading with sound effects. Where transcripts are unavailable, listening twice and writing a five-line summary in Hebrew after the second play is a good substitute, and it doubles as summary practice.

Worked example

The recovery routine, executed

Forty seconds into a two-speaker passage, a run of fast speech loses you completely and you no longer know who is talking. You have individual control. What exactly do you do, in order?

Do not rewind yet
The instinct is to stop and rewind immediately. Resist it for ten seconds, because the passage is about to tell you where you are: a speaker change, a connective, a proper noun or a number will arrive, and rejoining live is faster than rewinding blind. Rewinding without knowing what you are looking for costs a whole play.
Listen for the structural marker and rejoin there
Markers that carry structure: אבל and אולם, but; לעומת זאת and מנגד, by contrast; לכן and לפיכך, therefore; בנוסף, in addition; לסיכום, to sum up. Also any proper noun and any number, because those anchor the content. The moment one arrives, you know whether the argument turned and you can start following again.
Mark the gap on your notes, physically
Draw a line in your notes at the point where you lost it and write the time on the counter if there is one. You now have a precise target for the replay instead of a vague sense that something was missed, which is the difference between a five-second rewind and a whole extra play.
Finish the pass before you go back
Complete the pass to the end of the passage. The end of a discussion very often restates the positions, so a gap in the middle is frequently filled for free by the last twenty seconds. Only if it is not do you spend a rewind on it.
Replay the marked gap only
Rewind to your marked point and play only that stretch, with the specific question in mind - who said this, or what was the reason given. Listening for one thing is many times more effective than listening again in general, and it takes a fraction of the time.

Result: A five-step routine that converts losing the thread from a lost question into a targeted five-second replay: hold, rejoin at a structural marker, mark the gap, finish the pass, then replay only the gap with a specific question.

Typical mistakes

Practising only on slow, scripted teaching material and meeting unscripted speech for the first time in the exam.
Trying to translate discourse markers instead of hearing past them.
Reconstructing a lost sentence mentally while the recording continues, and losing the next two as well.
Reading a transcript while listening, which trains reading rather than listening.
Interpreting a future form as a prediction when it was a request.

Active revision

הכינו לעצמכם ארבע רשימות של מקורות שמע בעברית: חדשות, ראיונות ארוכים, שיחה לא מתוכננת, וחומר מוקלט עם תמלול. בכל שבוע האזינו לאחד מכל רשימה, וכתבו אחרי כל האזנה חמש שורות בעברית על מה שנאמר. בסוף החודש קראו את כל מה שכתבתם וראו באיזה משלב אתם עדיין הכי חלשים

Active recall

Recall the key points — then reveal.

Sources: AQA A-level Modern Hebrew (Listening, Reading, Writing) 7672 specification (AQA)

References & sources

Sources

AQA

AQA A-level Modern Hebrew (Listening, Reading, Writing) 7672 specification

הבנת הנשמע: Paper 3, individual control of the recording, and a language you cannot see

5 sections~30 min reading time4 competenciesLevel Foundation 1 · Standard 1 · Advanced 3

הבנת הנשמע: Paper 3, individual control of the recording, and a language you cannot see

Key points

Planning the plays before the first passage starts

Key points

Recovering a heard word you cannot spell

Key points

From two-column notes to a 95-word attributed summary

Key points

Planning 200 words that genuinely hold two sources together

Key points

The recovery routine, executed

Sources

הבנת הנשמע: Paper 3, individual control of the recording, and a language you cannot see

Key points

Planning the plays before the first passage starts

Key points

Recovering a heard word you cannot spell

Key points

From two-column notes to a 95-word attributed summary

Key points

Planning 200 words that genuinely hold two sources together

Key points

The recovery routine, executed

Sources