We need to talk about vocabulary instruction

“You shall know a word by the company it keeps.” – J.R. Firth (1957)

Introduction

J.R. Firth emphasised the importance of context in understanding vocabulary & that words are best learned in meaningful contexts rather than in isolation. Words tend to appear in predictable phrases with other words, e.g. “strong coffee” rather than “powerful coffee,” & teaching students these common phrases helps them use vocabulary more naturally. Students should encounter new vocabulary in authentic texts, conversations, or real-life situations, allowing them to grasp nuances, connotations, & different meanings based on context. Additionally, repeated exposure in various contexts – through reading, listening, speaking, & writing – develops understanding & therefore retention. Some meanings are only interpretable from understanding non-compositional, fixed expressions (e.g. “kick the bucket” doesn’t mean literally kicking a bucket) & teaching vocabulary in context rather than isolated items helps to develop fluency & natural use. Ultimately, Firth’s quote suggests that vocabulary instruction should extend beyond dictionary definitions & encourage students to engage with words in context, embedded in phrases & used in a variety of discourse communities, to develop a deeper & more functional understanding of language.

The problem with typical vocabulary teaching practices

Vocabulary is important, possibly the most important learning objective in ELT. Research indicates that vocabulary size is the strongest predictor of language proficiency, communicative competence, & performance in language exams such as Cambridge, IELTS, & Trinity (Coxhead et al., 2015; Webb & Nation, 2017). Despite this, typical ELT vocabulary instruction practices, such as pre-teaching vocabulary & asking students to infer meaning from context, often fall short, failing to promote long-term retention, & fundamentally limiting the progress that students should be able to make.

Pre-taught vocabulary

Pre-teaching vocabulary (PTV) before reading or listening exercises is a prevalent practice in ELT curricula & classrooms. PTV is problematic because:

  • It takes around 2-3 minutes to present each item sufficiently well to have an effect on readers’ or listeners’ comprehension of the fore-coming reading or listening text. Therefore 6 items of vocabulary will use up 12-18 minutes of valuable lesson time.

  • It requires teachers to predict which words students are unlikely to know, a challenging task even for experienced teachers, let alone for less experienced teachers, or for students that the teacher doesn’t know well.

  • Around 70% of pre-taught vocabulary is forgotten by the end of the lesson.

While PTV may aid in immediate comprehension, it is typically ineffective for long-term vocabulary acquisition. When we consider that the main purpose of instruction & learning activities is to promote longer-term, durable retention of language, we can reasonably conclude that PTV is not an optimal technique.

Inferring meaning from context

Another popular yet ineffective strategy is inferring meaning from context. For each word that a student attempts to infer the meaning of, this typically results in one of three possible outcomes:

  • Clear & unambiguous meaning from the context & use, allowing for correct inference.

  • Ambiguous & unclear meaning, leading to misinterpretation or confusion.

  • Complex meaning due to metaphorical, poetic, or sarcastic use, thereby increasing confusion & misinterpretation.

Additionally, successful inferencing from texts requires a high level of lexical coverage – at least 98% of the words in a text must be known for reasonably accurate inference. Less than this often results in misunderstandings & frustration for students (Laufer, 1996; Webb & Nation, 2017). From personal experience in many different teaching contexts on different continents, at least 98% lexical coverage is rarely the case in real-world ELT classrooms.

A more effective method: Parallel texts

A more effective technique for vocabulary learning involves keeping words in context & providing precise meanings through parallel texts. A parallel text consists of two columns: the original target language (TL) text, e.g. in English, in the left-hand column & the student’s first language (L1) translation, e.g. in Spanish, in the right-hand column. This technique, originally developed as part of the Comparative Method in the 1920s & 1930s, explicitly contrasts the students’ first language (L1) with the TL. This comparison helps students to understand differences in vocabulary choices, linguistic structures, & expressions (Fried, 1968).

How Does it Work?

  1. Understanding first: Students read & process the original text for overall meaning through text reconstruction activities, summarisation, comprehension questions, &/or speaking activities.

  2. Identification of unknown words: Without referring to the translation, students underline words that they don’t understand &/or wish to study further.

  3. Comparison with L1: They then examine the translated version to find equivalent expressions in their L1.

  4. Critical evaluation: The students then compare & contrast the TL words & expressions with the L1, noticing or having their attention drawn towards similarities & differences between them, & the phrases & language patterns in which they’re used. They also benefit from being encouraged to critically evaluate the quality of the translations, i.e. “Is that how you’d really express that idea in your language?” Thereby further refining & developing their understanding.

  5. Vocabulary learning plan: Finally, students plan strategies, with the help of their teacher, to develop their understanding of newly learned words, ensuring retention & practical application. Additionally, principles & guidelines from the science of learning & second language acquisition research can be of great benefit to students to learn how to study new language more effectively, e.g. (Agarwal, n.d.; Henshaw & Hawkins, 2022; The Learning Scientists, n.d.).

Advantages of Parallel Texts

  • Faster & more precise learning: First & foremost, parallel texts are faster & more precise. They tend to perform better than content-oriented tasks, intended to promote incidental vocabulary learning, & explicit non-contrastive vocabulary instruction (Laufer & Girsai, 2008).

  • More efficient use of time: Unlike typical ELT methods such as PTV or guessing from context, parallel texts optimise class time, allowing more focus on practice & development (Boers, 2021; Web & Nation, 2017).

  • Clarifies ambiguities: Comparing the TL (e.g. in English) & translations in the student’s L1 highlights differences in meaning, use, & expression, e.g. if the English text uses the word “uttered” but the Spanish translation uses “dicho,” which in English would be “said” that raises questions for the students to consider, i.e. Why use “uttered” in this context rather than “said”? Thereby creating awareness of a ‘gap’ in understanding & a stimulus for further investigation & learning.

  • Enhances awareness of language structures: Students notice gaps in their knowledge of differences between their L1 & the TL, thereby deepening their understanding of how the TL expresses ideas, e.g. expressions of duration of time between English and Spanish are very different, e.g. “I’ve lived in my house since I was five years old.” vs. “Vivo en mi casa desde que tenía cinco años.” (Literally, “I live in my house from/since that I was having five years.”) Creating awareness of this ‘gap’ in their knowledge of how to express ideas in the preferred forms in the TL (English) can also stimulate further engagement with these TL forms, deeper understanding of the TL, and therefore lead to more authentic language development & production.

  • Personalised learning: Instead of relying on their teachers’ predictions, students identify vocabulary that they find challenging & can focus more time on that rather than on vocabulary that they are already confident with, ultimately leading to more efficient language development, more time spent on learning needs, and greater engagement for all students.

  • Encourages learner independence: Requiring students to create a vocabulary learning plan for their selected vocabulary items encourages more effective, systematic, and consistent study skills, essentially asking them, “How are you going to ensure that you’ll remember and be able to use this vocabulary next week, next month, and next year?” Thus helping them to plan strategies and techniques to accomplish this. This explicitly teaches strategies & techniques for vocabulary & more general language development, enabling students to apply them autonomously in future encounters.

  • Boosts motivation: More effective and optimal ELT strategies and techniques also tend to result in greater motivation. According to the ‘reward-learning’ framework of motivation creating a need for learning, i.e. a knowledge ‘gap’, provokes information seeking behaviour, then supporting information seeking with appropriate information/instruction results in successful learning experiences which, in turn, reinforces positive attitudes to learning and learner persistence, thereby creating a positive feedback loop (Murayama & von Keyserlingk, 2025). One thing that stands out in the research on motivation and learning is that prior success in learning is among the strongest predictors of motivation, i.e. successful learning motivates students to learn more (Piñeiro et al., 2019; Steinmayr et al., 2019).

What are its caveats? How can these be mitigated?

The main issue seems to be occasionally poor translations. Sometimes NLP machine translations are imprecise, e.g. giving literal, word-for-word translations, rather than their more meaningful & preferred non-compositional equivalents, e.g. translating “table tennis” into “tenis de mesa.” These are typically obvious to the learners because of the cues provided by the context & co-text & so they can challenge the translations & provide better, more preferred ones in their L1 (Spanish), e.g. “table tennis” = “ping pong.” As an additional bonus this encourages more meaningful engagement with the source texts & therefore contributes further to deeper processing & more durable vocabulary learning.

By the way, similarly to human language processing, computer natural language processing (NLP) benefits greatly from being provided with co-text. In the case of NLP machine translator models, such as Google Translate, providing plenty of co-text, ideally the whole text, results in more precise & useful translations. Understanding this can also be very useful to students who often use machine translation services in less than optimal ways.

Conclusion

Typical ELT vocabulary instruction often falls short because it isolates words from context & often gives rather generic, dictionary style definition, leading to poor understanding, & therefore poor retention & inefficient learning. A more effective approach, supported by research, involves integrating vocabulary learning into meaningful contexts. Parallel texts offer an efficient & engaging method for learning & developing vocabulary, reinforcing understanding through contextualised exposure & students’ L1 translation comparisons. By implementing this strategy, learners not only improve understanding & retention but also gain deeper insights into language structures, ultimately achieving greater proficiency & communicative competence.

References