ISSN 2953-6367

http://revistainvestigo.com

Octubre 2026

Vol. 7 No. 21, PP. 50-63

https://doi.org/10.56519/4c0d0b95

APRENDIZAJE MÓVIL DE VOCABULARIO EN INGLÉS CON

LINGODEER EN UNA ESCUELA RURAL DEL VALLE DEL CHOTA,

ECUADOR: ESTUDIO PRIMARIO PRETEST-POSTEST

MOBILE VOCABULARY LEARNING WITH LINGODEER IN A

RURAL SCHOOL OF THE CHOTA VALLEY, ECUADOR: A

PRIMARY PRETEST-POSTTEST

Carlos Leomar De La Cruz Espinoza¹, Martha Viveros Almeida²

{carlos.delacruz@upec.edu.ec¹, martha.viveros@upec.edu.ec²}

Fecha de recepción: 29/07/2026

/ Fecha de aceptación: 21/08/2026

/ Fecha de publicación: 14/10/2026

RESUMEN: En las escuelas rurales, las limitadas oportunidades de exposición al inglés y

de práctica autónoma del vocabulario justifican la exploración de herramientas móviles

integradas a la enseñanza. En este contexto, este estudio cuantitativo primario examinó el

cambio en el logro de vocabulario en inglés durante un periodo de enseñanza apoyada por

LingoDeer en una escuela pública rural del Valle del Chota, Ecuador. Este estudio

cuantitativo primario examinó el cambio en el logro de vocabulario en inglés durante un

periodo de enseñanza apoyada por LingoDeer en una escuela pública rural del Valle del

Chota, Ecuador. Se analizaron las respuestas a 30 reactivos y los puntajes emparejados de

70 estudiantes, así como un cuestionario diagnóstico de 71 respondientes. Los totales se

verificaron reactivo a reactivo. La media aumentó de 12,26 (DE = 5,12) a 25,80 (DE = 1,42)

puntos sobre 30; la ganancia media fue 13,54 (DE = 5,22), t(69) = 21,69, p < 0,001, IC 95 %

[12,30; 14,79], r pre-post = 0,067 y dz = 2,59. Los 70 estudiantes mostraron diferencias

positivas (rango 2-23; prueba exacta de signos, p < 0,001), con una proporción de 7,1 % en

AAR o DAR, que se incrementó a 100,0 %. En el cuestionario, las respuestas «Nunca»

oscilaron entre 45,1 % y 67,6 % para mediación digital y entre 56,3 % y 63,4 % para

desempeño léxico autopercibido. La consistencia interna fue alta para mediación digital

(alfa = 0,966) y desempeño léxico (alfa = 0,968), mientras que el postest de logro produjo un

coeficiente negativo (alfa = -0,395). La misma prueba se administró en ambas ocasiones, los

puntajes finales estuvieron próximos al techo y el archivo analítico no conservó los

identificadores de condición ni medidas verificadas de exposición. Por ello, el artículo

informa un cambio primario pretest-postest, pero no atribuye causalmente ese cambio a

LingoDeer.

Palabras clave: aprendizaje móvil de idiomas; LingoDeer; vocabulario en inglés; TPACK;

educación rural; Ecuador

¹Unidad Educativa Guardiana De Los Saberes San Gabriel de Piquiucho: Docente de Inglés, https://orcid.org/0009-0001-1171-

8451; +593981172433.

²Universidad Politécnica Estatal del Carchi: Profesor Ocasional, https://orcid.org/0000-0002-2441-6954, +593979289520.

50

Revista Científica Multidisciplinaria InvestiGo

Riobamba – Ecuador

Cel: +593 97 911 9620

revisinvestigo@gmail.com

APRENDIZAJE MÓVIL DE VOCABULARIO EN INGLÉS CON LINGODEER EN UNA ESCUELA RURAL DEL VALLE

DEL CHOTA, ECUADOR: ESTUDIO PRIMARIO PRETEST-POSTEST

ABSTRACT: In rural schools, limited opportunities for exposure to English and

independent vocabulary practice justify exploring mobile tools integrated into teaching. In

this context, this primary quantitative study examined the change in English vocabulary

achievement during a period of instruction supported by LingoDeer in a rural public school

in the Chota Valley, Ecuador. This primary quantitative study examined change in English-

vocabulary achievement during a period of LingoDeer supported instruction in a rural public

school in Ecuador's Chota Valley. Thirty item responses and matched scores from 70

students were analysed, together with a diagnostic questionnaire completed by 71

respondents. Item totals were verified item by item. Mean achievement increased from

12.26 (SD = 5.12) to 25.80 (SD = 1.42) out of 30; mean change was 13.54 (SD = 5.22), t(69) =

21.69, p < .001, 95% CI [12.30, 14.79], pre-post r = .067, and dz = 2.59. All 70 students had

positive differences (range 2-23; exact sign test, p < .001), and the proportion in AAR or DAR

increased from 7.1% to 100.0%. For the questionnaire, 'Never' responses ranged from 45.1%

to 67.6% for digital mediation and from 56.3% to 63.4% for self-perceived lexical

performance. Internal consistency was high for digital mediation (alpha = .966) and lexical

performance (alpha = .968), whereas the posttest achievement items produced a negative

coefficient (alpha = -.395). The same test was administered at both occasions, final scores

were near the ceiling, and the analytic file did not retain condition identifiers or verified

exposure measures. The article therefore reports primary pretest-posttest change without

attributing that change causally to LingoDeer.

Keywords: mobile-assisted language learning; LingoDeer; English vocabulary; TPACK; rural

education; Ecuador

INTRODUCCIÓN

Vocabulary knowledge is the foundation for comprehension and production in listening,

speaking, reading and writing. Lexical knowledge is not available for communication until

learners have repeated experiences with the form, meaning, pronunciation, and contextual

use (1-4). In contemporary research, word knowledge is considered as a multidimensional

phenomenon and is differentiated from the productive mastery needed for communication

(1,3). The repeated presentation of word lists with spaced intervals of time and attention and

retrieval is necessary for acquisition; the presentation of word lists alone is not sufficient for

the acquisition of memory (2,4). In a low-proficiency English as a foreign language (EFL)

classroom, the limited amount of time for class instruction and the few opportunities for

individualized practice can result in a classroom that produces receptive fragments that are

not easily transferred to the classroom's oral and written environment.

The present study is based on two theoretical strands. Well-sequenced applications

operationalize this principle, which is known as the input-oriented approach, by systematically

sequencing and recycling input that is just a little above the learner's level. The sociocultural

view of learning sees the zone of proximal development as the space in which learning occurs

as a result of mediation (6), which can be provided by software, but requires teacher

orchestration to transform individual practice into shared communicative competence.

51

APRENDIZAJE MÓVIL DE VOCABULARIO EN INGLÉS CON LINGODEER EN UNA ESCUELA RURAL DEL VALLE

DEL CHOTA, ECUADOR: ESTUDIO PRIMARIO PRETEST-POSTEST

Mobile-assisted language learning (MALL) can provide portable, repeatable, multimedia

practice in the classroom. The recent reviews and meta-analyses, in general, suggest positive

results for second-language learning and, in particular, for vocabulary (7-13), although the

results are also characterized by significant differences in the effects across learners' ages,

intervention duration, learning design, and assessment. Mobile practice as a pedagogical tool

is not a treatment itself, but rather a pedagogical tool, and these syntheses do not exclude the

possibility of publication bias, weak comparison conditions or incomplete reporting, which can

inflate effect estimates. Convenience and gamified feedback are added to the commercial

applications, but the mere presence of an application does not guarantee learning. Some of

the concerns that persist are small samples, brief exposure, attrition, novelty effects, and

weak comparisons (14,15). Incomplete reporting of statistical assumptions, effect sizes,

measurement reliability and validity also are common issues identified in reviews of

quantitative MALL research (16,17).

The field has evolved from feasibility reports to controlled evaluations. An early synthesis

revealed mostly positive results across implementations (18), and more recent meta-analyses

indicate moderate to large average effects across implementations, with significant

heterogeneity, but not effects similar to the corrected within-person dz reported here (9,10).

The proposed mechanisms are: more time on task, immediate corrective feedback,

multimedia coding, and spaced review (19). The evidence at the platform level is still

inconsistent and is contingent on persistence, dosage, learning design and comparison quality

(15). Therefore, a large pooled pre-post standardized change should not be compared directly

or interpreted as application efficacy.

LingoDeer organizes each language course into a set of language units that are fixed in terms

of difficulty level and content from beginner to intermediate. Each unit includes explicit

grammar notes, presentation of words in context, native-speaker recordings, immediate item-

level feedback and progress analytics, as well as a dedicated spaced-repetition review module

(21). This is a structured, explanation-rich profile and is not as much an implicit, streak-driven

platform as it is a platform that might be articulated in the classroom, as unit topics can be

mapped onto national curriculum blocks. However, empirical assessment is still limited. A

recent qualitative case study reported that Indonesian junior secondary learners showed

greater interest, involvement and reported word recall, while the difficulties of pronunciation

and cost of premium version were reported (22). The present study fills in this gap in

quantitative achievement evidence and in evidence from Latin America.

These concerns are more significant in rural public-school environments where access to

devices, connectivity, and technical support can be inconsistent. The Ecuadorian EFL

curriculum is structured around five strands of learning (communication and cultural

awareness, oral communication, reading, writing, and language through the arts) and has

progressive expectations across General Basic Education (23) that align with the Common

European Framework of Reference for Languages (24). The most defensible use of digital

practice is when it is used to fulfill these curricular goals and is mediated in a purposeful way

by the teacher.

The technological pedagogical content knowledge (TPACK) framework supplies that rationale.

TPACK treats effective technology integration as the situated interaction of content,

52

APRENDIZAJE MÓVIL DE VOCABULARIO EN INGLÉS CON LINGODEER EN UNA ESCUELA RURAL DEL VALLE

DEL CHOTA, ECUADOR: ESTUDIO PRIMARIO PRETEST-POSTEST

pedagogical, and technological knowledge rather than as the isolated use of a digital tool

(25,26). Applied to MALL, TPACK implies that an application's affordances produce learning

only when the teacher aligns them with content standards and appropriate pedagogy. In the

present context, LingoDeer supplied structured multimedia exercises and immediate

feedback, while the teacher connected those activities to curricular vocabulary, the ERCA cycle

(experience, reflection, conceptualization, and application), and communicative classroom

tasks.

This adaptation was especially important in the predominantly Afro-Ecuadorian rural

community of the Chota Valley where opportunities to use English outside of school may be

limited, and devices, connectivity, and technical support may be uneven. As a result,

LingoDeer was designed to be a teacher-led, curriculum-based tool, not a standalone

application that requires constant internet connectivity.

The research team conducted the study in a public educational unit in Piquiucho, a

predominantly Afro-Ecuadorian community in the Chota Valley of Carchi Province. Classroom

control and intervention assignments were planned, but the final participant-level analytic file

retained grade only (8, 9, or 10), not class section or condition. The article thus reports a

primary pooled pre test-post test study as opposed to a controlled treatment-effect estimate.

Three research questions were addressed: (a) What were baseline patterns of teacher digital

mediation and self-reported lexical performance? (b) What were pooled and grade-specific

vocabulary-score changes from pretest to posttest? and (c) What were the measurement and

design constraints for the interpretation of those changes?

MATERIALES Y MÉTODOS

Design, setting, and data collection. The present article presents the main quantitative one-

group pretest-posttest analysis carried out at the Guardiana de los Saberes San Gabriel de

Piquiucho Educational Unit. This study used a baseline contextual questionnaire and a 30-item

vocabulary assessment, which were given before and after a period of instruction supported

by LingoDeer. The research team documented the responses for each item on the instrument

at the participant level and matched the achievement records with coded identifiers. These

basic study records were used to calculate all numerical results presented in this article. No

qualitative data were gathered or analysed.

Participants and analytic samples. There were 70 students who completed the matched 30-

item pretest and posttest responses for the achievement analysis, and 71 respondents who

completed the contextual questionnaire. The one-participant difference was due to the

absence of a student who completed the contextual questionnaire on the day of the

achievement posttest, so that student could not be included in the matched pretest-posttest

analysis.

The 70 complete achievement pairs were included, including four students who were

identified as having special educational needs (SEN) in the enrolment summary. They were

assessed as part of the school's regular inclusive classroom practices and the assessment

content and scoring criteria were not changed so that the assessment was comparable. There

53

APRENDIZAJE MÓVIL DE VOCABULARIO EN INGLÉS CON LINGODEER EN UNA ESCUELA RURAL DEL VALLE

DEL CHOTA, ECUADOR: ESTUDIO PRIMARIO PRETEST-POSTEST

was no SEN based exclusion or score adjustment. Unfortunately, individual SEN status and

specific support measures were not stored in the participant-level analytic file, meaning that

SEN-specific sub-group analyses were not possible; this is acknowledged.

Table 1. Participant distribution, curricular target, and planned classroom assignment.

Grade / class

group

Boys

5

Girls

5

Total

10

SEN

2

Planned classroom role / analytic status

Planned control; section/condition not

retained in analytic file

Eighth A

Planned intervention; section/condition

not retained in analytic file

Eighth B

Ninth A

Ninth B

5

10

2

Planned control; section/condition not

retained in analytic file

4

6

10

0

Planned intervention; section/condition

not retained in analytic file

5

10

0

Tenth (single

group)

LingoDeer-supported instruction; no

comparator

12

31

18

39

30

0

70 complete pairs included; no

participant-level condition or SEN field

Total

70

4

Note. Planned classroom roles, class sections and SEN are recorded as part of the enrollment records. The

participant-level analytic file includes coded ID, grade, item responses, and total score but lacks section, condition,

sex, and SEN fields. All 70 complete pairs were included, so there is no possibility of separating the planned control

and intervention classrooms for a between groups analysis that is reproducible.

Instruments. A locally developed objective test with 30 items and a scoring of 1 point for each

correct answer was used as the achievement instrument (0-30). The same form was

administered at pretest and posttest; no parallel form or delayed retention test was used.

Expert content review and alignment with CEFR A1-A2 and the Ecuadorian curriculum were

reported, but independent criterion validation was not conducted. Analysis of the primary

binary item responses gave alpha/KR-20 = .814 at pretest and alpha = -.395 at posttest. The

negative posttest coefficient is a negative average inter-item covariance and is therefore not

a valid estimate of internal consistency for this administration. Unstable inter-item

covariances may have been due to the ceiling effect (M = 25.80/30, SD = 1.42) and the lack of

variance in scores. The total scores for the post test were therefore considered exploratory.

The diagnostic questionnaire comprised 14 items: 1-8 were related to teacher digital

mediation, 9-13 were related to self-perceived lexical performance and Item 14 was related

to resource preference. Alpha values were .966 for Items 1-8 and .968 for Items 9-13, with a

13-item alpha of 0.983. These coefficients describe internal consistency only and do not

establish unidimensionality or construct validity.

54

APRENDIZAJE MÓVIL DE VOCABULARIO EN INGLÉS CON LINGODEER EN UNA ESCUELA RURAL DEL VALLE

DEL CHOTA, ECUADOR: ESTUDIO PRIMARIO PRETEST-POSTEST

Table 2. Measurement framework and data-collection instruments.

Quality evidence

Instrument

Construct and content

Scale

reported

Primary item-level

responses; pretest

alpha/KR-20 = .814;

posttest alpha = -.395

(not acceptable)

Meaning recognition; contextual

vocabulary; simple-sentence

production; auditory discrimination

Achievement pre-

/post-test

30 items; 0-30

Items 1-8 alpha = .966;

Items 9-13 alpha =

.968; all 13 items

alpha = .983; factor

structure untested

Teacher digital mediation (Items 1-

8); self-perceived lexical

Diagnostic

questionnaire

4-point frequency

scale (13 items)

performance (Items 9-13)

Audio was the modal

response: 35/71

(49.3%)

Resource-preference

item

Preferred resource for vocabulary

development (Item 14)

Single categorical

choice

Note. Resource-preference results were calculated from the 71 primary questionnaire responses: audio 35

(49.3%), reading 19 (26.8%), dialogues 13 (18.3%), and dictation 4 (5.6%).

Figure 1. Study procedure and analytical scope.

Note. The diagram presents the primary-data study sequence and the pooled analytic scope. A condition key and

documented exposure are required before a control-versus-intervention effect can be estimated.

Table 3. Curriculum-to-activity alignment in the instructional guide.

Curriculum thread

Grade 8 (A1.1)

Grade 9 (A1.2)

Grade 10 (A2.1)

Communication and

cultural awareness

Greetings and

introductions

Food origins and healthy

choices

Months, festivals, and

birthdays

Animals and sound-based

practice

Buying clothes: buyer-

seller exchanges

Oral communication

Reading

Family dialogues

Schedules, time, and

agendas

Short descriptions of pets

Menus and food labels

Classroom objects and

demonstratives

Routines and short

messages

Writing

Personal diary entries

Language through the

arts

Colours and visual

labelling

Seasons and comparative

descriptions

Weather vocabulary

55

APRENDIZAJE MÓVIL DE VOCABULARIO EN INGLÉS CON LINGODEER EN UNA ESCUELA RURAL DEL VALLE

DEL CHOTA, ECUADOR: ESTUDIO PRIMARIO PRETEST-POSTEST

Note. Levels and content follow the study's curriculum-convergence matrix and instructional units. A1.1, A1.2,

and A2.1 denote the CEFR sublevels set by the Ecuadorian curriculum for eighth, ninth, and tenth grade

respectively.

Statistical analysis. Primary item totals were calculated for every participant and matched by

coded ID across occasions. Integer totals were preserved and summary values were rounded

only after calculation. Means, sample standard deviations, standard errors, ranges, and

achievement-band frequencies were computed from the 70 complete paired records. For the

paired-samples t test, the pretest-posttest data and a two-sided alpha of .05 were used, and

for the 95% confidence interval, df = 69. Participant-level scores were used to calculate

Pearson's pre-post correlation, Cohen's dz, and Glass's delta. The paired-samples t test was

used and supplemented by an exact two-sided sign test as a sensitivity analysis, due to the

questionable normality of the paired difference scores and substantial grade heterogeneity.

The item-level responses were used to compute alpha/KR-20 coefficients for the achievement

test and for the two subscores of the questionnaire. The analytic file does not contain any

condition identifiers, so an effect of control versus intervention was not estimated.

Achievement bands were DAR 27-30, AAR 21-26, PAAR 13-20, and NAR 0-12.

RESULTADOS

Baseline digital mediation. Primary questionnaire responses (N = 71) showed that digital tools

were seldom integrated into English instruction. 'Never' responses were 57.7% for digital

applications as class materials, 62.0% for English audio through applications, 45.1% for audio-

repetition exercises, 50.7% for reading-comprehension application frequency, 62.0% for their

actual use, 57.7% for writing exercises supported by applications, 63.4% for online

applications supporting English learning, and 67.6% for pronunciation practice. Every

percentage is an attainable fraction of 71 and was calculated from its count.

Self-reported lexical performance. 'Never' responses were 57.7% for understanding main

ideas when reading, 63.4% for understanding vocabulary in audio input, 60.6% for using

vocabulary in everyday speech, 56.3% for understanding vocabulary while reading, and 63.4%

for using appropriate vocabulary in writing. Audio was the modal preferred resource (35/71,

49.3%), followed by reading (19/71, 26.8%), dialogues (13/71, 18.3%), and dictation (4/71,

5.6%). These findings answer the first research question as baseline context; they are not

presented as outcomes of the intervention.

Questionnaire reliability and response pattern. The alpha for the teacher digital mediation

(Items 1-8) was .966, for self-perceived lexical performance (Items 9-13) was .968, and for all

13 scaled items, alpha was .983. Separate subscale coefficients prevent the two conceptually

different domains from being considered as a single construct. Very high alpha values,

however, may be due to redundancy in the items or to a highly patterned response matrix and

are not indicative of factorial validity. Factor analysis, item-total diagnostics, test-retest

reliability, and external validation were not conducted.

56

APRENDIZAJE MÓVIL DE VOCABULARIO EN INGLÉS CON LINGODEER EN UNA ESCUELA RURAL DEL VALLE

DEL CHOTA, ECUADOR: ESTUDIO PRIMARIO PRETEST-POSTEST

Figure 2. Baseline «Never» responses for digital mediation and lexical performance.

Note. N = 71. Percentages were calculated from primary questionnaire counts. The two corrected rounding values,

41/71 = 57.7% and 40/71 = 56.3%, are highlighted in yellow in the figure.

Pooled vocabulary-test change. For the 70 matched records, the pretest total was 858 (M =

12.26, SD = 5.12, range 5-23) and the posttest total was 1806 (M = 25.80, SD = 1.42, range 23-

29). The integer paired-difference total was 948, giving a mean change of 13.54 (SD = 5.22, SE

= 0.62, range 2-23), t(69) = 21.69, p < .001, 95% CI [12.30, 14.79]. The pre-post correlation was

r = .067, Cohen's dz = 2.59, and Glass's delta = 2.64. All 70 paired differences were positive

(range 2–23); the two-sided sign test was exact, p < .001, and did not assume normally

distributed paired differences. The sign test is used to assess the direction of change, not the

size of the mean difference, and thus complements (not replaces) the paired t test.

Table 4. Pooled pretest-posttest vocabulary results.

Outcome

Pretest

N

Mean

12.26

25.80

SD

SE

Difference / inferential result

70

5.12

1.42

0.61

0.17

n/a

Posttest

t(69) = 21.69; p < .001; 95% CI

[12.30, 14.79]

Posttest - pretest

70

13.54

n/a

5.22

n/a

0.62

n/a

Standardized /

sensitivity

dz = 2.59; Glass delta = 2.64; r =

.067; exact sign p < .001

57

APRENDIZAJE MÓVIL DE VOCABULARIO EN INGLÉS CON LINGODEER EN UNA ESCUELA RURAL DEL VALLE

DEL CHOTA, ECUADOR: ESTUDIO PRIMARIO PRETEST-POSTEST

Note. Scores range from 0 to 30. All values were calculated from 70 primary participant-level records; totals were

integers and means were rounded to two decimals only after calculation. The exact mean difference is 948/70 =

13.542857. dz standardizes the paired difference; Glass's delta uses the pretest SD. Neither is a between-group

causal effect.

Figure 3. Mean vocabulary scores before and after the instructional period.

Note. Error bars represent +/- 1 sample SD. Yellow labels identify the corrected values calculated from the

primary participant-level data.

Achievement-band distribution. At pretest, 42 students (60.0%) were in NAR, 23 (32.9%) in

PAAR, 5 (7.1%) in AAR, and none in DAR. At posttest, 48 (68.6%) were in AAR and 22 (31.4%)

in DAR; no participant was in PAAR or NAR. Thus, the combined AAR/DAR proportion

increased from 7.1% to 100.0%. These counts were calculated directly from the primary

participant-level totals. Without retained condition identifiers, the distribution cannot

establish a LingoDeer treatment effect.

Table 5. Distribution across nationally adapted achievement bands.

Achievement band

Code

DAR

AAR

Score range

27-30

Pretest n (%)

0 (0.0)

Posttest n (%)

22 (31.4)

Masters the required learning

Achieves the required learning

21-26

5 (7.1)

48 (68.6)

Close to achieving the required

learning

PAAR

13-20

23 (32.9)

0 (0.0)

Does not achieve the required

learning

NAR

0-12

42 (60.0)

0 (0.0)

Total

n/a

0-30

70 (100.0)

Note. The four category labels are translated from the Ecuadorian qualitative scale and proportionally adapted

to a 30-point test. DAR = domina los aprendizajes requeridos; AAR = alcanza los aprendizajes requeridos; PAAR

= está próximo a alcanzar los aprendizajes requeridos; NAR = no alcanza los aprendizajes requeridos.

58

APRENDIZAJE MÓVIL DE VOCABULARIO EN INGLÉS CON LINGODEER EN UNA ESCUELA RURAL DEL VALLE

DEL CHOTA, ECUADOR: ESTUDIO PRIMARIO PRETEST-POSTEST

Figure 4. Pretest and posttest achievement-band distributions.

Note. N = 70. Values are percentages calculated from the primary participant-level totals. Yellow labels identify

corrected values. DAR, AAR, PAAR, and NAR are defined in Table 5.

Magnitude, heterogeneity, and ceiling. The within-person standardized mean change (dz) is

the mean paired difference (13.54) divided by the standard deviation of the paired difference

(5.22). It is not an independent-groups effect and should not be interpreted as or compared

directly with Hedges' g. The final score compression (restricted posttest variance, SD = 1.42)

and the near-ceiling posttest mean (M = 25.80/30), coupled with the large value, suggest that

caution should be used when interpreting the results. Therefore, dz is not reported as a large

effect size between groups due to LingoDeer. Mean gains differed sharply by grade: 19.20

points in Grade 8 (SD = 2.55), 15.35 in Grade 9 (SD = 1.57), and 8.57 in Grade 10 (SD = 3.02).

These differences track distinct baseline levels and show that the pooled statistic mixes grade

and curriculum effects. The near-zero pre-post correlation (r = .067) shows little rank-order

stability. At posttest every score lay between 23 and 29, the mean was 86% of the maximum,

and alpha was negative. Ceiling compression, repeated exposure to the same form, close

alignment between instruction and test content, and possible teaching to the test are

therefore plausible alternatives to a treatment interpretation.

Summary of findings. The primary records support three reproducible descriptions: limited

baseline digital mediation and low self-reported lexical performance, a large pooled increase

in integer achievement totals, and strong grade heterogeneity followed by a compressed

posttest distribution. The final analytic file does not support a control-versus-intervention

comparison because condition membership is absent. The negative posttest reliability

coefficient, use of the same form twice, lack of a delayed test, missing actual-dose and fidelity

logs, and lack of participant-level SEN identifiers materially limit interpretation. The corrected

statistics establish primary pretest-posttest change but do not by themselves establish

causality.

59

APRENDIZAJE MÓVIL DE VOCABULARIO EN INGLÉS CON LINGODEER EN UNA ESCUELA RURAL DEL VALLE

DEL CHOTA, ECUADOR: ESTUDIO PRIMARIO PRETEST-POSTEST

DISCUSIÓN

Three main findings emerged from this primary study. First, pre-questionnaire data show that

digital mediation is scarce and self-perceived lexical performance is low. Second, the

participant-level achievement data reveal a large pooled score gain (all 70 differences were

positive). Third, the increase is quite different across grade levels and tapers off to a low, near-

ceiling posttest distribution that has negative item reliability. The correct interpretation is a

main descriptive change in the teaching period, not that LingoDeer brought about the change.

The baseline results clarify why mobile support was pedagogically relevant. Across the eight

digital-mediation items, 45.1%-67.6% selected 'Never'; across the five lexical self-report items,

56.3%-63.4% selected 'Never'. Audio was the modal resource preference. These attainable

percentages and separate subscale reliabilities answer the contextual research question

without being used as post-intervention evidence.

The corrected mean gain remains much larger than typical educational intervention effects,

but several features can inflate it: low and grade-stratified baselines, direct alignment

between practised and tested content, immediate feedback, administration of the identical

instrument twice, and a posttest ceiling. The fall in SD from 5.12 to 1.42 is descriptive evidence

of restricted final-score variance, not proof of equalized learning. The negative posttest alpha

further cautions against interpreting the total as a stable latent measure.

TPACK focuses on how LingoDeer contributes to the practice rather than the practice of

LingoDeer itself, that is, teacher-led pedagogical orchestration. The teacher chose the

technological affordances of the application to achieve specific lexical and phonological goals,

integrated them in the ERCA cycle (experience, reflection, conceptualization, application), and

linked the digital practice with communicative follow-up exercises that were related to the

grade level curriculum. Therefore, the likely teaching process was the teacher's alignment and

mediation: LingoDeer provided structured multimedia practice and immediate feedback, and

the teacher was able to use these to create contextualized learning opportunities that were

linked to the five curricular strands and the progression from A1.1 to A2.1 (23,24). Note that

this interpretation should not be understood as evidence that LingoDeer or teacher mediation

alone caused the observed change because the study design did not isolate these components

experimentally.

The outcome is not solely due to LingoDeer. Control and intervention classrooms were

planned, but no section or condition key is included in the participant-level analytic file except

for grade. The intended comparison cannot be used in a pooled paired analysis, and may be

due to maturation, normal instruction, repeated testing, teaching to the test, attendance

selection, regression to the mean, and unequal exposure. A causal or between-group claim

must be accompanied by recovery and verification of the condition key as well as actual-dose

information.

Implications for practice. The instructional protocol provides a starting plan that is replicable:

six units of curriculum aligned instruction, teacher introduction, communicative follow-up,

independent practice for 15-20 minutes daily, and weekly review. Future implementations

need to record actual sessions, attendance, time on task, module completion, teacher fidelity,

60

APRENDIZAJE MÓVIL DE VOCABULARIO EN INGLÉS CON LINGODEER EN UNA ESCUELA RURAL DEL VALLE

DEL CHOTA, ECUADOR: ESTUDIO PRIMARIO PRETEST-POSTEST

home access and technical interruptions. Before making decisions about adopting the plan, it

should be evaluated using parallel assessment forms, a delayed retention measure, and a

prespecified control comparison.

Limitations and directions for research. First, the class-section and condition identifiers were

not retained in the participant-level analytic file and thus were not available for use of the

planned controls. Second, the actual dose, adherence, usage logs, fidelity, and concurrent

instruction were not measured or retained. Third, the same test constructed locally was

administered twice, there was no delayed test to measure retention, and the post test had

ceiling compression and alpha = -.395. Fourth, the enrollment summary shows four students

with SEN, and the analytic file does not contain an individual SEN field, and no exclusion was

applied for SEN. Fifth, the contextual questionnaire included 71 respondents, whereas the

matched achievement sample included 70 because one questionnaire respondent was absent

from the achievement posttest and could not be included in the paired analysis.

CONCLUSIONES

Three conclusions follow from the analysis of the primary participant-level data.

First, the baseline diagnosis establishes a contextual need for additional support. Across

digital-mediation items, 45.1%-67.6% of 71 respondents selected 'Never'; across self-

perceived lexical-performance items, 56.3%-63.4% selected 'Never'. Audio was the most

frequently preferred resource (35/71, 49.3%). These questionnaire findings answer the

baseline research question and should not be presented as intervention outcomes.

Second, the primary score data support a large pooled descriptive change: pretest M = 12.26

(SD = 5.12), posttest M = 25.80 (SD = 1.42), mean difference = 13.54 (SD = 5.22), t(69) = 21.69,

p < .001, with AAR/DAR increasing from 7.1% to 100.0%. The corresponding dz = 2.59 is a

within-person statistic and must not be presented as a controlled application effect.

Third, primary-data status does not resolve the design and measurement limitations.

Condition identifiers and verified actual dose are missing; the same test was repeated; final

scores are near the ceiling; and posttest alpha is negative. The defensible conclusion is that

vocabulary scores increased during the instructional period. A causal claim about LingoDeer

requires a verified control/experimental key and a prespecified controlled analysis or a new

controlled replication.

AGRADECIMIENTOS (OPCIONAL)

The authors thank the management team, teaching staff, students, and families of the

Guardiana de los Saberes San Gabriel de Piquiucho Educational Unit for their participation and

for authorizing the implementation.

61

APRENDIZAJE MÓVIL DE VOCABULARIO EN INGLÉS CON LINGODEER EN UNA ESCUELA RURAL DEL VALLE

DEL CHOTA, ECUADOR: ESTUDIO PRIMARIO PRETEST-POSTEST

REFERENCIAS BIBLIOGRÁFICAS

1.

2.

3.

Nation ISP. Learning vocabulary in another language. 2nd ed. Cambridge: Cambridge

University Press; 2013. https://doi.org/10.1017/CBO9781139858656

Schmitt N. Instructed second language vocabulary learning. Language Teaching

Research. 2008;12(3):329-63. https://doi.org/10.1177/1362168808089921

Schmitt N, Schmitt D. Vocabulary in language teaching. 2nd ed. Cambridge: Cambridge

University Press; 2020. https://doi.org/10.1017/9781108569057

4.

5.

6.

Webb S, Nation P. How vocabulary is learned. Oxford: Oxford University Press; 2017.

Krashen SD. The input hypothesis: issues and implications. London: Longman; 1985.

Vygotsky LS. Mind in society: the development of higher psychological processes.

Cambridge (MA): Harvard University Press; 1978.

7.

Burston J, Giannakou K. MALL language learning outcomes: a comprehensive meta-

analysis

1994-2019.

ReCALL.

2022;34(2):147-68.

https://doi.org/10.1017/S0958344021000240

8.

Hao T, Wang Z, Ardasheva Y. Technology-assisted vocabulary learning for EFL learners:

a meta-analysis. Journal of Research on Educational Effectiveness. 2021;14(3):645-67.

https://doi.org/10.1080/19345747.2021.1917028

9.

10.

11.

12.

13.

Peng H, Jager S, Lowie W. Narrative review and meta-analysis of MALL research on L2

skills. ReCALL. 2021;33(3):278-95. https://doi.org/10.1017/S0958344020000221

Yu A, Trainin G. A meta-analysis examining technology-assisted L2 vocabulary learning.

ReCALL. 2022;34(2):235-52. https://doi.org/10.1017/S0958344021000239

Zhou Y, Zhou M. A meta-analysis on mobile-assisted vocabulary learning: do mobile

applications

help?

ReCALL.

2026;38(1):75-93.

https://doi.org/10.1017/S0958344025100335

14.

15.

16.

17.

18.

Kukulska-Hulme A, Shield L. An overview of mobile assisted language learning: from

content delivery to supported collaboration and interaction. ReCALL. 2008;20(3):271-

89. https://doi.org/10.1017/S0958344008000335

Loewen S, Crowther D, Isbell DR, Kim KM, Maloney J, Miller ZF, et al. Mobile-assisted

language

learning:

a

Duolingo

case

study.

ReCALL.

2019;31(3):293-311.

https://doi.org/10.1017/S0958344019000065

Burston J, Athanasiou A, Giannakou K. Quantitative experimental L2 acquisition MALL

studies: a critical evaluation of research quality. ReCALL. 2024;36(1):22-39.

https://doi.org/10.1017/S0958344023000149

Hou Z, Aryadoust V. A review of the methodological quality of quantitative mobile-

assisted

language

learning

research.

System.

2021;100:102568.

https://doi.org/10.1016/j.system.2021.102568

Burston J. Twenty years of MALL project implementation: a meta-analysis of learning

outcomes. ReCALL. 2015;27(1):4-20. https://doi.org/10.1017/S0958344014000159

62

APRENDIZAJE MÓVIL DE VOCABULARIO EN INGLÉS CON LINGODEER EN UNA ESCUELA RURAL DEL VALLE

DEL CHOTA, ECUADOR: ESTUDIO PRIMARIO PRETEST-POSTEST

19.

20.

Nakata T. Computer-assisted second language vocabulary learning in a paired-associate

paradigm: a critical investigation of flashcard software. Computer Assisted Language

Learning. 2011;24(1):17-38. https://doi.org/10.1080/09588221.2010.520675

Stockwell G, Hubbard P. Some emerging principles for mobile-assisted language

learning. Monterey (CA): The International Research Foundation for English Language

Education; 2013.

21.

22.

LingoDeer. LingoDeer: learn languages [Internet]. 2023 [cited 2026 Jul 23]. Available

from: https://www.lingodeer.com/

Triwiyanti I, Dalimunte AA. Learning English vocabulary by LingoDeer: case study of

Indonesian junior high school students. Acuity: Journal of English Language Pedagogy,

Literature and Culture. 2025;10(3):276-89. https://doi.org/10.35974/acuity.v10i3.4166

Ministerio de Educación del Ecuador. English as a foreign language curriculum [Internet].

Quito: Ministerio de Educación; 2016 [cited 2026 Jul 23]. Available from:

https://educacion.gob.ec/wp-content/uploads/downloads/2016/08/0-EFL_v2.pdf

Council of Europe. Common European Framework of Reference for Languages: learning,

teaching, assessment, companion volume. Strasbourg: Council of Europe Publishing;

2020.

23.

24.

25.

Mishra P, Koehler MJ. Technological pedagogical content knowledge: a framework for

teacher

knowledge.

Teachers

College

Record.

2006;108(6):1017-54.

https://doi.org/10.1111/j.1467-9620.2006.00684.x

26.

27.

Koehler MJ, Mishra P. What is technological pedagogical content knowledge?

Contemporary Issues in Technology and Teacher Education. 2009;9(1):60-70.

Li Y, Hafner CA. Mobile-assisted vocabulary learning: investigating receptive and

productive vocabulary knowledge of Chinese EFL learners. ReCALL. 2022;34(1):66-80.

https://doi.org/10.1017/S0958344021000161

63