CBI Recruitment /English Language

Para jumbles

Comprehension & Writing · free study notes

Para‑Jumbles – Advanced Study Notes

(English Language – Academic Level, MIT‑style)


1. What are Para‑Jumbles?

A para‑jumble is a type of sentence‑reconstruction exercise.

  • A passage of 4–6 sentences is scrambled: the order of the sentences is shuffled.
  • The solver must reorder them into a coherent narrative.
  • The sentences are single‑statement (no commas that split clauses) and usually range between 10–20 words.

Para‑jumbles appear in many standardized tests (e.g., GRE, GMAT, UPSC, Indian engineering entrance tests).


2. Theoretical Foundations

2.1 Linguistic Cohesion and Coherence

Cohesion Device Example Effect
Temporal markers (first, then, finally, afterward) First, the researchers collected data. Sequence of actions
Causal connectors (because, therefore, so) Because the sample size was small, the results were inconclusive. Cause–effect relation
Contrastive markers (however, but, although) However, the methodology was sound. Exception/contrast
Referential devices (this, it, they) It was concluded that the hypothesis was false. Pronoun referencing
Logical quantifiers (all, none, some) All participants were informed. Scope of statement

Cohesion refers to the grammatical and lexical connections that link sentences.
Coherence refers to the overall logical flow that makes the passage meaningful.

2.2 Graph‑Theoretic Model

Each sentence can be represented as a node in a directed graph.
Edges are weighted according to the cohesive compatibility between sentence pairs.

  • Edge weight = number of shared discourse markers (temporal, causal, etc.) that indicate a plausible forward direction.
  • The optimal ordering is the Hamiltonian path that maximizes the sum of edge weights.

Why this works?

  • The graph captures pairwise compatibility, allowing global optimization.
  • The Hamiltonian path problem is NP‑hard, but for 4–6 nodes exhaustive search is trivial.

2.3 Probabilistic Language Model

A n‑gram language model can be trained on a corpus of well‑structured academic abstracts.

  • Input: sentence vectorization (e.g., via TF‑IDF).
  • Output: probability ( P(\text{next sentence} \mid \text{current sentence}) ).
  • The ordering that maximizes the product of these probabilities is the most plausible.

3. Algorithmic Procedure

3.1 Pre‑processing

  1. Tokenization – split sentences into words.
  2. POS Tagging – identify verbs, nouns, conjunctions.
  3. Discourse Marker Extraction – list all temporal, causal, contrastive markers.
  4. Referential Resolution – detect pronouns and map them to antecedents (within the set).

3.2 Cohesion Scoring Function

def cohesion_score(sent_a, sent_b):
    score = 0
    # Temporal compatibility
    if any(t in sent_a for t in ['first', 'then', 'finally']) and \
       any(t in sent_b for t in ['then', 'afterward', 'consequently']):
        score += 2
    # Causal compatibility
    if 'because' in sent_a and 'therefore' in sent_b:
        score += 3
    # Contrast compatibility
    if 'however' in sent_a and 'but' in sent_b:
        score += 1
    # Referential compatibility
    if any(pron in sent_b for pron in pronouns_in(sent_a)):
        score += 1
    return score

3.3 Exhaustive Search (for (n \le 6))

import itertools

def optimal_order(sentences):
    best_order = None
    best_score = -1
    for perm in itertools.permutations(sentences):
        s = 0
        for i in range(len(perm)-1):
            s += cohesion_score(perm[i], perm[i+1])
        if s > best_score:
            best_score = s
            best_order = perm
    return best_order, best_score

3.4 Edge Cases and Handling

Edge Case Problem Remedy
Missing temporal markers Ambiguous sequence Use pronoun references and causal markers.
Multiple possible starting points Tie in scores Inspect global coherence: start with sentence that has no clear previous markers.
Circular dependencies Graph contains cycle Break cycle by selecting the edge with smallest penalty (e.g., minimal cohesion).
Pronoun ambiguity Incorrect antecedent mapping Apply coreference resolution (e.g., AllenNLP Coref) before scoring.

4. Advanced Techniques

4.1 Discourse Parsing (Rhetorical Structure Theory)

  • Build an RST tree that captures nuclear vs. satellite relations.
  • The nucleus (most central clause) should appear before satellites in the ordering.

4.2 Machine‑Learning Classifier

  • Features: POS tags, n‑gram overlap, discourse marker counts, coreference distance.
  • Labels: “appropriate next sentence” vs. “inappropriate”.
  • Train on a large set of academic paragraphs (e.g., ACL Anthology).
  • Use the classifier to score all candidate next sentences and pick the best.

5. Challenging Problem

Problem Statement
You are given the following 5 scrambled sentences from a research paper. Arrange them in the correct order.

  1. The experimental results were statistically significant, indicating a strong correlation between the variables.
  2. In addition, the control group exhibited no notable changes, thereby reinforcing the hypothesis.
  3. The researchers conducted a double‑blind trial over a period of six months.
  4. Finally, the study was subjected to peer review, which validated the methodology.
  5. The participants were randomly assigned to experimental and control groups.

Constraints

  • Use only the discourse markers and coreferences present.
  • Justify each transition with a cohesion score (see §3.2).
  • Provide the final sequence and the total cohesion score.

Solution Approach

  1. Identify markers
    • Temporal: Finally
    • Causal: thereby
    • Referential:

Your note

Page 1

Doubt hai is topic me?

Free AI teacher se poocho — 24/7, step by step.