Rogers, Perls, and Ellis in Three Approaches to Psychotherapy: A Corpus-based Study

Roberta A. Miranda1 and Cass Dykeman1

Author Note

Roberta A. Miranda https://orcid.org/0000-0001-9115-7596

Cass Dykeman https://orcid.org/0000-0001-7708-1409

This authors’ preprint contains an empirical study intended for submission to a peer reviewed

journal. All comments are welcome and can be directed to the first author at

[email protected]

Version 2, 3-24-22

1
 Abstract

Published research focused on the analysis of specific words used by counselors during

counseling sessions is limited. Applying keyness analysis within a corpus linguistic framework

can help researchers examine transcribed verbal interactions between counselors and clients in

counseling sessions. Additionally, keyness studies can uncover specific word choices that align

with specific theoretical orientations, and study potential implications made by both new

counselors and counseling predecessors. This study employed a corpus linguistic design

analyzing words within the three unique counseling transcriptions in the popular training film,

Three Approaches to Psychotherapy. Outcomes determined that keywords identified with the

language spoken by Carl Rogers, Albert Ellis, and Fritz Perls did align with their associated

theories of client-centered, Rational Emotive Behavioral Therapy, and Gestalt theories,

respectively. As such, counseling educators can use this film to illustrate specific word choices

reflective of three unique theoretical models.

Keywords: Perls, Ellis, Rogers, psychotherapy, Gloria, counseling, keyness, linguistic

analysis

2
 Rogers, Perls, and Ellis in Three Approaches to Psychotherapy: A Corpus-based Study

A new counselor in a counseling session often strives to build the therapeutic

relationship by building rapport. The counselor is mostly likely wondering how the words used in

session are impacting the client. There might also be curiosity regarding whether the words

spoken coincide with the preferred theoretical orientation. The counselor might also want to

know how their selective language might lead to ruptures with the client. Applying linguistic

analysis to examine verbal interactions between counselors and clients can shed light on how

studying these words can influence the training of new counselors, as well as enlighten the

counseling field to the processes and words chosen by well-known counseling predecessors.

To date, published research focused on the analysis of specific words used by

counselors during counseling sessions has been limited. To further exemplify this limitation in

research, previous keyness studies have been based solely on written correspondence among

online mental health virtual platforms (Brookes et al., 2016; Coimbra-Gomes &

Motschenbacher, 2019). There remains a gap in the literature on specific words of counselors

used dynamically in live interactive counseling discussions. The aim of the study was to grow

the knowledge base related to the field of counseling education by disrupting current practice

(Tadajewski et al., 2011) by focusing on specific words and verbal communication tools that

counselors consider for their own clinical work. Current practice asks student-counselors to role-

play counseling sessions with a general guidance of theoretical orientation, nonverbal

communication skills, and focused clinical skills. Yet, students are often left to determine their

own verbal delivery and choice of words. Studying linguistic analysis can help students review

their own verbal practices, as well as study the speech of experienced counselors for training

purposes.

The present study examined counselor word choice by analyzing the work of three

renowned counselors (i.e., Carl Rogers, Albert Ellis, and Fritz Perls) with the same client. The

texts studied were the transcripts from the popular film series titled, Three Approaches to

3
Psychotherapy (Shostrom, 1965). The overall aim was to determine how their choice of words

were different from one another, thus reflecting specific word differences in their associated

theoretical orientations. Other researchers have studied this classic film series (e.g., Barbosa et

al., 2015). However, newer corpus linguistic software tools present the opportunity to study the

language use by Rogers, Perls, and Ellis in a more granular manner.

In the following five paragraphs the reader will encounter a literature review aligned with

the intended variables of this study. The themes pursued in this review were: (1) what is word

“keyness” and why is it important, (2) what research has been conducted on keyness and

counseling, (3) what is known about Rogers and language use, (4) what is known about Perls

and language use, and (5) what is known about Ellis and language use. After these questions

are addressed, the research questions are detailed.

The study of word keyness (also known as keyword analysis) has a relatively recent

history, though it is rapidly gaining popularity (Culpeper, 2009). Keywords are words that occur

unusually frequently in comparison with a reference corpus (Scott, 2020). Keyness studies offer

the opportunity to identify words within a given text to further explore their potential importance

and meaning within a context (Scott & Tribble, 2006), especially when comparing two similar

corpuses. For example, a recent study compared keywords identified among different written

vignettes illustrating two different types of grief styles (instrumental and intuitive) providing

guidance regarding how counselors can respond accordingly to grieving clients (Geisler &

Dykeman, 2021). Exploring keywords in other mental health-focused texts or counseling

transcriptions might also provide insight into the importance and impact of specific words utilized

by mental health providers.

There have been limited studies on psychotherapy and the linguistic phenomena

keywords. To date, keyword analysis on the words used between counselors and individuals

with mental health concerns have mostly been dependent on repositories collected from online

service platforms. Brookes et al. (2016) analyzed communication in a practitioner-led health

4
advice website dedicated to adolescent health. In one of their published studies, researchers

utilized information obtained through the adolescent health platform, “The Teenage Health

Freak (THF),” a website that serves as a popular resource for young people seeking health

advice anonymously. The study focus was based on the Adolescent Health Email Corpus

(AHEC), a 1.6-million-word corpus containing advice-seeking emails sent to the THF website.

The theme of this study focused on the terms “depressed” and “depression,” given their higher

frequency in the corpus. A counselor might not consciously take note of wide differences

between the closely related terms; however, using linguistic analysis, the words “depressed” or

“depression” indicated different implications when used among adolescents. The choice of using

“depression” over “depressed” pinpointed more information on the adolescent’s potential mental

health status, as well as their personal perspective on their own severity, thus serving as a

helpful guide for potential mental health providers. Another linguistic study (Coimbra-Gomes &

Motschenbacher, 2019) utilized a combination corpus linguistic analysis from the online service,

“Psychforums,” which offers mental health support to its users. In this study, the researchers

focused on male heterosexual sexual orientation obsessive-compulsive disorder (SO-OCD),

further exploring underlying biased beliefs of heteronormativity (the belief that heterosexuality is

the default, preferred or “normal” mode of sexual orientation). Keyword analysis of texts

associated with a specific community of language users provides insight into normal

communicative practices of that community. The researchers compared two corpuses compiled

from Psychforums: (a) a target corpus of thread-initial posts written by male heterosexual SO-

OCD sufferers, and (b) a reference corpus of thread-initial posts by heterosexual men who used

the forum to seek relationship or sexual advice. The keyword analysis in this study identified the

discourses (discussions about a topic) that SO-OCD sufferers utilized more frequently in this

community of practice. These included terms related to sexual/gender identities, sexual desire,

and SO-OCD as a pathology. Notably, the language often oriented to heteronormativity in their

advice-seeking comments, providing insight into the writers’ levels of internal conflict regarding

5
their sexuality identification. Again, studying keywords to understand underlying meanings

among different platforms can better educate mental health professionals on how to respond

appropriately, as well as help direct the mental health field on how to offer more public

information and resources. The methodologies and tools to study keyness can also allow

researchers to study historical pieces of work that have since contributed to the development of

counseling psychology (including recordings that could be transcribed for analysis). Discovering

these keywords can enlighten the psychology and counseling fields of the selected words and

potential implications of established counselors and their theoretical tenets.

The words used between a counselor and client are an essential piece of the therapeutic

process. Rogers, widely considered to be one of the founding fathers of psychotherapy, is

known for using gentler, reflective conversation, often conveying compassion and care with his

client-centered approach. Wickman et al. (2003) analyzed Rogers’s session with client, Gloria,

in the taped session in Three Approaches to Psychotherapy by examining the following: exact

words used, lengths of pauses, changes in volume, tone/pitch to assess for nonexpert

language, metastatements, affiliative negative assessments, first-person quotes, invitations for

repair, withholding direct responses to requests for advice, and problem reformulation. The

outcomes determined that Rogers’s conversation style was congruent with his own client-

centered theory as he demonstrated empathy, genuineness, and unconditional positive regard.

Hill et al. (1979) invited professionals to judge transcripts of Rogers, Perls, and Ellis each doing

counseling with the same client, Gloria, using the 14-category Hill Counselor Verbal Response

Category System. The results of this study substantiated this theoretical orientation in that

Rogers mainly encouraged, restated, and reflected. The study also identified that Rogers used

many minimal encouragers, inviting Gloria to talk more than him. In the video recording, Rogers

was the first counselor featured with Gloria, followed by Perls. Observers of Rogers’s interaction

with Gloria might note her comfort and ease in opening up to him and her defensiveness with

6
Perls. Most viewers of the recording might also quickly notice Perls’s difference in tone and curt

responses with Gloria as compared to Rogers.

Perls’s own use of language in Gestalt psychotherapy is observed as more direct and

intense. In one technique of the therapy, Perls addressed clients in their tendency of “role-

playing” (when clients refrained from showing or knowing their authentic whole selves). His

approach was to then challenge his clients through more directive or displeasing methods, such

as calling them phony, acting bored, or rolling his eyes in mock frustration (Perls, 1969). He

used these mechanisms to help clients break through and demonstrate awareness in the here-

and-now with authenticity (O’Leary, 2013). Original recordings of Perls demonstrating Gestalt

therapy at the Esalen Institute in Big Sur, California, from 1966-1968 were later recorded and

transcribed for the book, Gestalt Therapy Verbatim. In the introduction to the text, Perls was

said to “[bark] commands like a ballet coach or theater director,” and “now and then, he glowed

with approval when someone came to life and expressed a convincing emotion” (Perls, 1969, p.

1). Perls's theory of Gestalt therapy relies on two key processes in counseling: an awareness of

the here-and-now experiencing and the resolution of discrepancies. Results of Hill’s (1979)

study focusing on Perls’s interaction with Gloria indicated that Perls was unique in his use of

both nonverbal referents and confrontations in focusing on Gloria's present “experiencing” when

addressing discrepancies in her behaviors. Perls used mostly direct guidance, information,

interpretation, open questions, minimal encouragers, closed questions, confrontations,

approval-reassurance, and nonverbal responses.

Ellis was the last counselor featured on the Three Approaches recording and presented

yet a different style of tone, language, and reaction to Gloria. During the recording, Ellis is

observed as more talkative and rapid in his language use with Gloria. Ellis asked many

questions throughout the session and seemed to use Gloria’s responses as an opportunity to

provide psychoeducation about rational emotive behavioral therapy (REBT). The theory of

REBT, developed by Ellis, stresses that behavioral distress is a result of an individual’s irrational

7
belief systems. REBT practitioners work closely with clients to help identify their individual set of

beliefs that may lead to this emotional distress, then provide a variety of methods to help people

reformulate their dysfunctional beliefs into more realistic and helpful ones. Counseling may

include a technique called “disputing,” as well as “reeducation,” by challenging these irrational

cognitions and replacing them with newer rational thoughts and beliefs (Ellis, 2003). The results

of Hill’s (1979) study indicated that Ellis mainly used information giving, direct guidance, minimal

encouragers, and interpretations, which appears congruent with REBT. He was also extremely

active in comparison with Perls and Rogers, which fits his re-educative approach rather than

exploratory processing.

Given the gaps and needs, six research questions were designed to guide this study.

These questions were:

RQ1: While counseling the same client, what words differentiate Ellis’s speech from

Rogers’s?

RQ2: While counseling the same client, what words differentiate Ellis’ speech from

Perls’s?

RQ3: While counseling the same client, what words differentiate Perls’s speech from

Rogers’s?

RQ4: While counseling the same client, what words differentiate Perls’s speech from

Ellis’s?

RQ5: While counseling the same client, what words differentiate the Rogers’s speech

from Perls’s?

RQ6: While counseling the same client, what words differentiate the Rogers’s speech

from Ellis’s?

8
 Method

Design

The study employed a corpus linguistic design (Brezina, 2018). The variables were

words and counselors. The level of measure was continuous for words and nominal for

counselor. The unit of analysis was tokens (Brezina, 2018). The a priori power analysis χ2 test

was used to determine the needed sample size. This was calculated by utilizing the program

G*Power 3.1 (Faul et al., 2009). Cohen’s w is the effect size for the χ2 test (Rosnow et al.,

2003). A research article on implications of treatment to a diagnostic mental health condition

(Wade et al., 2009) provided the data used to ascertain the effect size, and a final value was

determined by averaging the Cohen’s w figures reported in the research. The G*Power 3.1 input

parameters were: (a) test family = χ2 tests; (b) statistical test = goodness-of-fit tests: contingency

tables; (c) power analysis = compute required sample size - given α, power, and effect size; (d)

effect size w = 0.708; (e) α error probability = .001; (f) Power (1-β error probability) = 0.80; and

(g) degrees of freedom = 1. The G*Power 3.1 output parameters were: (a) total sample size =

64 and (b) actual power = 0.806.

Participants

Rogers

Carl Rogers (1902-1987) was an American psychologist and one of the founders of a

humanistic approach to psychotherapy, including the client-centered therapeutic approach in

counseling. He identified three conditions a counselor must facilitate for client growth, including

exuding genuineness and congruence, practicing acceptance and empathy, and offering

unconditional positive regard. With an emphasis on human potential, Rogers had an enormous

influence on both psychology and education. He has been described as conveying a

compassionate and nonauthoritative approach with clients.

Perls

9
 Frederick “Fritz” Perls (1893-1970) was a German psychiatrist and counselor and one of

the founders of Gestalt therapy in the 1940s and 1950s. Gestalt therapy is a phenomenological-

existential therapy and teaches counselors and clients the phenomenological method of

awareness (Perls et al., 1951). Gestalt therapy focuses more on process (what is happening)

than content (what is being discussed), with an emphasis on what is being felt in the moment. In

his communication style, Perls would refer to individuals as playing a phony role and remain

poker-faced with clients (Perls, 1969).

Ellis

Albert Ellis (1913-2007) was an American psychologist and served as the founder and

president of the New York-based Albert Ellis Institute for decades. Ellis is credited with

developing rational therapy into REBT in 1955, which laid the foundation for more modern-day

practices of cognitive behavioral therapy (CBT). REBT encourages the development of rational

thinking to facilitate healthy emotional expression and behavior. Ellis won many awards for his

work in the field of psychology.

Corpus

Register, Scope, and Sources

The register is spoken language, and the subregister is psychotherapy/counseling. The

scope was the three counseling transcribed sessions featured in the film series, Three

Approaches to Psychotherapy. For Rogers, the source was a transcript of the session edited by

Brodley and Lietaer (2006). Published transcriptions of the filmed counseling sessions for Ellis

and Perls were unavailable. As so, each session was manually transcribed by viewing each

session located on YouTube. The transcription separated the dialogue of each participant and

was placed in sequential back and forth order of the conversation. Only verbal interactions,

including small utterances (identified as Ipsum), were included in the transcription. For each

research question, one of these three corpuses served as the study corpus and another as the

10
reference corpus. Corpus sizes (word stem counts) were: (a) Ellis = 2454, (b) Perls = 1485, and

(c) Rogers = 1913.

Preprocessing

Each set of transcriptions was checked for accuracy. The following steps were

completed on each set of transcripts prior to conducting linguistic analyses. Spelling was

corrected and meaningful abbreviations, such as “Jan,” were spelled out to January.

Appropriate punctuation (period, comma, or question mark) was made at the end of each

speaker’s turn. Nonfluencies (such as “mm hmmm”) or words that were hard to audibly

understand (such as mumbles) were replaced with a standard non-English place holder word

(i.e., ipsum). The one exception to this conversion of nonfluencies was Rogers’s use of “uh,”

given its frequency and potential stylistic functions. Each set of transcriptions (between Gloria

and each counselor) was generated in three different Word documents and checked for

accuracy. Accuracy was based on listening to each recording and ensuring that each statement

was transcribed as stated. Additionally, attention was placed on separating the dialogues of

Gloria and each counselor. For this study, only counselor statements were retained for analysis;

thus, the dialogue from Gloria was not included. The three transcripts were converted to .txt files

for analysis. The comparative transcripts of study were identified as either the target corpus or

the reference corpus. For example, when the focus was what words used differentiate the Ellis’s

speech from Rogers’s, Ellis’s transcript was identified as the target corpus and Rogers as the

reference corpus. The Natural Language Toolkit (NLTK) lists of the top 40 stopwords (common

function words such as “the,” “to,” and “a,” that do not carry content meaning) were included for

data collection purposes, although six commonly-used words were excluded.

Measures

Keyness

Keywords are words that occur with an unusually high frequency in a corpus (Brezina,

2018) when analyzed against another (study or reference) corpus. When selecting a study

11
corpus, it is best to use one that is at least similar in size to the target corpus and that belongs

to a similar genre so that keywords will inform how the words may be lexically distinctive about

the language in each corpus (Hunt & Brookes, 2020). Comparative keyword analysis makes

comparisons of the frequency of words between two bodies of text so that words appearing

more frequently in one body of text are highlighted (Seale, 2008). Stemming was applied,

mapping different forms of the same word to a common stem in conjunction with the Porter

algorithm (Willet, 2006).

Apparatus

Text processing and keyness analyses were conducted using the R programming

language. The R packages used for this study were “tm” and “quanteda.textstats.”

Data Analysis

For all six research questions, raw counts for both the study and reference corpus are

provided. Differences between the corpuses were assessed using the log likelihood ratio test

(G2). As an effect size, log ratio operates such that every extra point represents a doubling in

size of the difference between the two corpuses for the keyword under consideration (Hardie,

2014). The words reported as keywords met these thresholds: (a) minimum word frequency in

study corpus = 2, (b) minimum word frequency in reference corpus = 0, (c) alpha level = p < .01

(G2 ≥ 6.63), (d) log ratio was ≥ 1.2, and (e) whether the word was identified as a non stopword.

The filtering criteria were more restrictive for smaller values of p and larger values of

n_target/n_reference/binary_lr. The maximum p-value cutoff was between 0.001 and 0.05. The

minimum binary_lr cutoff was between 1.2 and 1.5. Word stemming was applied, specifically

using the R (R Core Team, 2015) function tm::stemDocument().

Results

The obtained results for all words meeting the aforenoted thresholds across all six

research questions can be viewed in Table 1. To summarize the keyness results for RQ1,

where Ellis was the target corpus, and Rogers, the reference corpus, the top 3 keyness words

12
identified were: Get, Because, and Focus. For RQ2, where Ellis was the target and Perls, the

reference, the top 3 keyness words identified were: Really, Yourself, and Be. For RQ3, where

Perls was the target and Rogers, the reference, the top 3 keyness words identified were: Now,

Your, and Phony. For RQ4, where Perls was the target and Ellis, the reference, the top 3

keyness words identified were: Me, Fritz, and Play. For RQ5, where Rogers was the target and

Perls, the reference, the top 3 keyness words identified were: Feel, Guess, and Sense. For

RQ6, where Rogers was the target and Ellis, the reference, the top 3 keyness words identified

were: Like, Guess, and Really.

Discussion

Six research questions guided this study. These focused on what words differentiated:

(a) Ellis’s speech from Rogers’s (RQ1), (b) Ellis’s speech from Perls’s (RQ2), (c) Perls’s speech

from Rogers’s (RQ3), (d) Perls’s speech from Ellis’s (RQ4), (e) Rogers’s speech from Perls’s

(RQ5), and (f) Rogers’s speech from Ellis’s (RQ6). In the following paragraphs, we will explicate

these results from these research questions in order.

 The film, Three Approaches to Psychotherapy (Shostrom, 1956), featured Ellis first

describing, then demonstrating REBT in his recorded counseling session. Alternatively, Rogers

discussed and illustrated his client-centered approach. RQ1 explored the specified differences

in word use between these two counselors and the two schools of therapy of which they

represented. In examining RQ1, which explores specific word differences between Ellis (the

target corpus) and Rogers (the reference corpus), it is significant that word “get” appears much

more prominently in Ellis’s lexicon (used 22 more times than Rogers). REBT is typically more

directive and behavioral-focused than Rogers's client-centered counseling and suggests that

Ellis used a more explicit action-focused framework, as demonstrated in three of his quotes

below, with six left (6L) and six right (6R) concordance lines around the node word “get.”

 man-hunting. I believe that people only get emotions, such as negative emotions

 Now, let's see if we can get at the source of your shyness.

13
  time now. So let’s try to get it off on a constructive note.

In another example of utilizing the action-focused word “get,” Ellis projected what Gloria may

have been thinking to herself. This externalizing approach was intended to help mirror what her

irrational thoughts might sound like if said back to herself in her own “voice.” This can be seen in

these two 6L-6R concordance lines:

 that would be awful. I'd never get what I want. And that would

 a moment ago if I don't get what I want right now I'll

Additional words of significance that differentiate Ellis and Rogers included the words “because”

and “focus.” The word “because” is a subordinator, a term that introduces adverbial clauses.

“Because” also highlights consequential effects of the client’s thoughts, feelings, and behaviors.

This determination to explore more specific “cause and effect” situations (as this coincides with

REBT) differs from Rogers’s approach of allowing the client to freely process feelings that can

lead to insight. In this, Ellis strove to help Gloria recognize the effects of her behaviors. Examine

these two 12R concordance lines:

 Because after all, if you do win one of these men, you have

 Because that's the basic function of life—enjoyment—which we tend to lose

The word “focus,” again, suggests that Ellis leaned on more directive language to help the client

establish more intentionality with goals, rather than Rogers’s flexible and nondirective clinical

experience. The following quote from Ellis reveals this intentionality: “If you could do what we

are talking about, really take risks and focus on what you want out of life…” In terms of RQ1,

Ellis distinguished himself from Rogers by frequently using words that reflect either action or

cause and effect. As the target corpus, his top keywords “get,” “because,” and “focus,” reflect

his REBT framework in that his lexicon is more directive in nature, goal oriented, and is used as

a therapeutic tool to help clients distinguish the causality between their irrational thinking and

behaviors.

14
 Regarding RQ2 (Ellis v. Perls), Ellis’s reality-focused language is noticeably different

from Perls’s metaphorical terminologies. Whereas Perls used intensified challenging therapeutic

techniques, Ellis leaned more into psychoeducation with Gloria. Again, one sees Ellis

distinguish his style with asking Gloria to examine how her irrational thinking impacted different

areas in her life, including self-image, romantic relationships, and parenting. In this further

examination, the word “really” is noted as significantly different with Ellis having used it 17 more

times than Perls. Ellis used this amplifying adverbial of actuality as a means to help Gloria

explore her perceived “real” examples of irrational thinking. Note these three 6L-6R

concordance lines that illustrate this point:

 suppose, for the moment, that you really were fully accepting yourself with your

 eventually be yourself. Now if you really weren’t so disturbed about these present,

 You see. So, if you would really accept yourself as you are and

Ellis also significantly utilized the word “yourself” 28 more times than Perls. Although, Perls used

the term “your” 55 times in his corpus, there appears to be quite a distinction in how “your” and

“yourself” were intended by each counselor. Perls used “your” as a possessive determiner to

relate Gloria to the given speech situation. Possessive determiners make noun phrases definite,

thus suggesting how Perls used “your” to place the onus on the client, encouraging personal

responsibility within the counseling process. Perls also used the term “your” frequently to

encourage the client to be increasingly aware of the self in the moment. (“Are you aware of

‘your’ smile?”). Ellis, in a different manner, used the word “yourself” as a second-person singular

reflexive pronoun, serving as a function word to make it possible to refer succinctly to clients by

asking them to place themselves as the object of an action (or verb)—hence, part of the

irrational thinking-behavioral cycle. These three 6R-6L concordance lines reveal this emphasis:

 to find out what you're telling yourself. You're meeting this individual. Now what

 well. Now if you could accept yourself for the time being with this

 But look how you just devalued yourself. Let's just suppose, for the sake

15
Here, Ellis was attempting to emphasize the specific behaviors of which the client was

engaging, and by this, continually reintroduced the irrational thinking-consequence cycle to

Gloria. Perls, on the other hand, used more metaphorical words, such as “phony,” “corner,” or

“play” to promote in-the-moment experiences for the client, without such a structured theoretical

framework. Lastly, in comparing Ellis to Perls, the word “be” was of significance suggesting how

the word “be” emphasizes rational, specific goals with the client, as illustrated in Ellis’s

statement, “So you have to eventually be yourself.” In sum, by identifying keywords, such as

“really,” “yourself,” and “be” in comparison to Perls’ corpus, Ellis clearly guided Gloria in a

manner congruent with his REBT theory to stay focused and rationale.

Regarding RQ3 (Perls v. Rogers), one encounters Perls’s unique use of language to

accomplish his therapeutic goals. Specifically, he used more provoking terms to activate Gloria,

rather than Rogers’ style of emphasizing words to promote the therapeutic alliance. Gestalt

therapy emphasizes the here-and-now work within the counseling sessions, helping clients

discover their true authentic selves. Perls facilitated this goal with more direct verbal

interventions. Gestalt therapy relies less on clients’ history or contextual information and more

on utilizing the session itself to uncover insight. This most likely explains why the word “now”

appears much more often in Perls’s lexicon (used 34 times) than both Rogers and Ellis. It is also

notable how Perls’s use of the word “your” was 32 more times than Rogers’s, suggesting the

counselor’s stance of the counselor-therapist relationship. As Rogers depended more on the

therapeutic alliance as a counseling tool, Perls’s choice of the word “your” kept the client’s

attention on herself. By emphasizing “your,” Perls was also promoting ownership to the client,

and as such, Gloria was inclined to reflect more astutely on her own thoughts and experiences.

This practice is clear in these two 6L-6R concordance lines:

 you've got the two part of your existence now, either far away in

 you didn't have to cover up your anger with your smile. Now you're

16
Perls also used more provocative and direct language to activate the client in the moment. As

Rogers tended to use gentle language to engage, comfort, and invite thoughtful reflection, Perls,

instead, used jarring words, such as “phony” and “corner” to move Gloria out of her comfort

zone. The word, “phony” also coincides with one of the Gestalt core concepts of describing

layers of neurosis in which phoniness refers to the fake or inauthentic way that people act with

other people. Using this intervention, he challenged her way of experiencing, as indicated in the

following 5L and 7L concordance lines:

 giggle and you squirm-- it’s phony.

 you're mad, you're not a phony.

 I refuse to accept the phony

 attention to you, you crawl into a corner

 it's a gimmick to crawl into a corner

 But if you coyishly go into the corner

Examining Perls’ language suggests how he cleverly used metaphoric words and evoking

phrases that brought Gloria to the here-and-now during the session. These techniques support

the Gestalt approach of helping clients resolve personality splits. His challenging approach

aligns with the Gestalt intention of helping clients (often uncomfortably) resolve these splits and

become congruent with their authentic selves.

In exploring RQ 4 (Perls’ V. Ellis), one can see how Perls intentionally used himself as a

therapeutic tool with Gloria while Ellis relied on more objective methods. To exemplify this

approach, Perls used the word “me” more frequently. Although Ellis (and Rogers) frequently

used “I,” Perls used “me” more often as a direct intervention. In his example, “Okay, pick on

me,” he placed himself as a potential theoretical punching bag for Gloria to challenge him. Perls

also utilized his own name, “Fritz,” throughout the session, an approach that neither Rogers nor

Ellis utilized. He used his name, within this Gestalt approach, to ignite client breakthroughs as in

these two 6L-6R concordance lines:

17
  start over. Can you now play Fritz? Fritz doesn't like Gloria, what would

 That's your imagination. That's not this Fritz, it’s the Fritz of your imagination

Perls’s work with Gloria places less emphasis on consequential and intellectual

approaches, as utilized by Ellis, and instead relies on language that offers more direct “hot

seat,” in-the-moment experiences. Perls not only used his words to intensify feelings in the

here-and-now but also asked Gloria to shed her “phony” self to instead be more authentic as

can be viewed in these three 2L-5R concordance lines:

 If you play dumb and stupid, you force

 you now play Fritz? Fritz doesn't like Gloria

 like to play the baby, and be comforted

Regarding RQ5 (Rogers v. Perls), Rogers’s unique use of language coincided with his

therapeutic goal of promoting the therapeutic alliance, distinguishing himself from the curt verbal

style of Perls. Specifically, Rogers used more diminisher adverbials to soften the tone of the

conversations compared to Perls’s frequent use of direct, challenging words. Rogers also

utilized more feeling-focused words, speaking both for himself and for Gloria. Rogers’s

approach of client-centered counseling allowed clients to speak openly, without the constant

redirection of the counselor. He also tended to the client-counselor relationship to build trust,

comfort, and acceptance in the session. Rogers quickly built his therapeutic alliance with Gloria

by utilizing more diminisher adverbials (also known as hedge words), such as (a) guess (e.g., “I

guess I hear you saying”); (b) sound (e.g., “I am sure this will sound evasive to you.”); (c) kind of

(e.g., “Sometimes, you kind of uh, feel like blaming them for the feelings you have.”); and (d) uh

(e.g. “It's much easier to uh be a little flip.”). Although this appears as a style of reservation,

Rogers purposely used this approach as a method to follow his diminisher words with an

interpretation to help gently push Gloria in more reflective work. Notice how the diminisher

words (italicized) in the examples below give Rogers’s statements a certain air of tentativeness,

yet offers his therapeutic interpretation:

18
  I guess, judging from your tone of voice, you sound as though you hate yourself…

 I, I sense that in those utopian moments, you really feel kind of whole.

 Yeah. You sort of feel, "I want them to have just as nice a picture of me.”

Rogers believed that humans have the basic motive to self-actualize. By utilizing unconditional

positive regard with clients, he encouraged them to subjectively respond to encounters feeling

unjudged and free to make mistakes. By feeling more comfortable with their actual selves, they

also became congruent with their actions. In facilitating this freedom for the client to share more

openly, Rogers used the word “feel” 41 times (36 more times than Perls), emphasizing the

intention to keep clients more connected to emotion and reflection, over cognitive- or abstract-

focused approaches. Examples of Rogers using “feel” to describe Gloria’s emotions are

reflected in these two 7L-4R concordance lines:

 And somehow, sometimes, you kind of uh, feel like blaming them for

 What you'd like to do is to feel more accepting toward yourself

Rogers also used “feel” to share his own emotions with Gloria to build the therapeutic alliance

as seen in these two1L-6R concordance lines:

 I feel very keenly is that it's an

 I feel close to you in this moment.

Perls utilized himself, the therapist, as a therapeutic tool with Gloria, using his own strong

presence and direct language. Rogers, instead, emphasized his relationship with Gloria as his

therapeutic tool. Rogers’ choice of language reflects the gentle client-centered approach,

distinguishing himself from Perls’s more directive Gestalt approach.

In RQs 6 (Rogers v. Ellis), Rogers utilized language to convey understanding and

empathy, whereas Ellis focused more on what was not working with her “irrational” thinking. By

using the word “like” more often than Ellis, Rogers allowed Gloria to know how genuinely

interested he was in what she chose to share during the session. He also used the word “like”

as a method of validating her such as in these 4L-4R concordance lines:

19
  And I guess I'd like to say, "No, I

 but let me, I'd like to understand it. You

 You look to me like a pretty nice daughter

Rogers also distinguished himself from Ellis in his choice of the word “guess.” As Ellis strove to

identify real examples of Gloria’s irrational thinking and consequences, Rogers allowed Gloria

the opportunity to ponder what resonated with her during the session, then reflect back to him

as can be noticed in these two 1L-10R concordance lines:

 I guess that's what I meant when I said, "Life is risky."

 I guess, judging from your tone of voice, you sound as though

Although Rogers also used the word “really” only 14 more times than Ellis, it is notable how he

used the word differently. Rogers’ utilization of “really,” an informal amplifier, emphasized the

feeling in the moment rather than emphasized the reality, as in the examples with these three

6L-6R concordance lines:

 up? One of the things you really deeply want is to find a

 hard to believe that they would really love you if they knew you?

 say that. That was, he's never really known you and loved you and

It is also notable how he also used “really” to emphasize his own empathy and understanding of

Gloria, as in the example in this 14L-3R concordance line:

 I expect none of us get it as often as we'd like, but I really do understand it.

By examining Rogers’s lexicon and how it is distinguishable from Ellis’s, one can see how

Rogers depended on diminishers, informal amplifiers, feeling words, and more ambiguity when

helping Gloria discover her true self, which, again, supports his therapeutic goals with client-

centered therapy.

There were two limitations to the present study that should be considered. The most

significant one is sample size. With three corpuses comprising three individually transcribed

counseling sessions, results yielded a smaller sample of keywords to analyze. As such, the

20
linguistic representation of each featured counselor was represented with one live counseling

session, thus capturing a “snapshot” of each counselor’s work for comparison. The second

limitation was the exclusion of selected function words in the data analysis. Although, six

stopwords in the English language (or, her, she, is, it, and this) were computed as keywords,

analysis/comparison was not conducted because of their minimal content meaning. However,

excluding these words may have offered additional implications on the counselors’ intentional

choice of words when conducting counseling.

Implications for counseling and counselor education that can be drawn from the obtained

results. First, as identified keywords associated with the language spoken by Rogers, Ellis, and

Perls did align with their associated theories of client-centered, REBT, and Gestalt theories,

reviewing “Three Approaches to Therapy” from a linguistic lens, under a social constructivist

teaching model, can serve as an excellent teaching tool in counselor training and keep the 1965

film relevant for future counselors. Using keyness to identify specific words made by the

different theorists, thus reflecting specific schools of counseling, can help promote rich

discussions and provide guidance to students of whom want to adhere to one of these theories

more closely. Alternatively, with the study of keyness, students have the opportunity to analyze

their own counseling transcripts and identify their own key words, which can be quite

enlightening. This helps to first, establish a baseline of personal language use, then, with

ongoing keyness analysis, can alert the student counselor of their progress in adhering to the

fidelity (via word use) to their chosen theory (even outside that of REBT, Gestalt, or Client-

Centered). Alternatively, learning personal language styles analyzed through keyness studies

can also inform student counselors of which school of counseling/theoretical orientation might

be a better fit based on their own natural speaking style. As linguistic analysis becomes more

commonly used in the helping fields, computer analysis computer programs, with keyness

identification and other linguistic analysis features, can serve as invaluable tools in counselor

education in teaching clinical skills and word specificities in different schools of counseling.

21
 Two recommendations for research should be noted. The first is to expand the sample

sizes of the works of Ellis, Fritz, and Rogers for analysis. As this study examined one example

of their work in action with a client, analysis of their clinical work with a longitudinal model over

the course of their clinical careers, can examine verbal consistency in relation to their

associated theories. Additionally, as each of these counselors have written numerous books,

future researchers can study how their written works and examples of their respective schools

of counseling have remained consistent or have changed over their years in practice. Secondly,

it is recommended to conduct additional linguistic analysis to further expand upon the study of

one-word keyness to the study of collocations (common word/phrase combinations) on the

transcripts of these counseling predecessors to further enhance how their language exemplifies

their respected theoretical approaches and can serve as a great teaching tool in counselor

education.

22
 References

Abdi, H., & Williams, L. J. (2010). Tukey’s honestly significant difference (HSD)

test. Encyclopedia of research design, 3(1), 1-5.

Backx, W. (2011). Views on REBT, past, present and future: Albert Ellis’ contribution to the field.

Journal of Rational-Emotive & Cognitive-Behavior Therapy, 29(4), 263–271.

https://doi.org/10.1007/s10942-011-0142-4

Barbosa, E., Cunha, C., Santos, A., Gonçalves, M. M., & Salgado, J. (2015). Narrative change

in Gloria Films: Comparing various processes of therapeutic innovation. Psychotherapy

Research, 27(1), 89–101. https://doi.org/10.1080/10503307.2015.1072286

Biber, D., Conrad, S., & Leech, G. (2002). Longman student grammar of spoken and written

English. Longman.

Brezina, V. (2018). Statistics in corpus linguistics: A practical guide. Cambridge University

Press.

Brezina, V., Timperley, M., & McEnery, T. (2021.). #LancsBox: Lancaster University corpus

toolbox 6.0 [software]. http://corpora.lancs.ac.uk/lancsbox/

Brodley, B. T., & Lietaer, G. (2006). Transcripts of Carl Rogers’ therapy sessions (Vol. 12).

Faculty of Psychology and Educational Sciences KU Leuven.

Brookes, G., & Harvey, K. (2016) Examining the discourse of mental illness in a corpus of

online advice-seeking messages. In L. Pickering, E. Friginal, & S. Staples (Eds.),

Talking at work (pp. 209-234). Palgrave Macmillan. https://doi.org/10.1057/978-1-

137-49616-4_9

Coimbra-Gomes, E., & Motschenbacher, H. (2019). Language, normativity, and sexual

orientation obsessive-compulsive disorder (SO-OCD): A corpus-assisted discourse

analysis. Language in Society, 48(4), 565–584.

https://doi.org/10.1017/s0047404519000423

23
Culpeper, J. (2009). Keyness. International Journal of Corpus Linguistics, 14(1), 29–59.

https://doi.org/10.1075/ijcl.14.1.03cul

Ellis, A. (1991). Comment on "Perls with Gloria re-reviewed" Journal of Counseling and

Development, 70(2), 353–353. https://doi.org/10.1002/j.1556-6676.1991.tb01615.x

Ellis, A. (2003). Cognitive restructuring of the disputing of irrational beliefs. In W. T. O’Donohue,

J. E. Fisher, & S. C. Hayes (Eds.), Cognitive behavior therapy: Applying empirically

supported techniques in your practice (2nd ed., pp. 79–83). John Wiley & Sons.

Faul, F., Erdfelder, E., Buchner, A., & Lang, A. G. (2009). Statistical power analyses using

G*Power 3.1: Tests for correlation and regression analyses. Behavior Research

Methods, 41(4), 1149–1160. https://doi.org/10.3758/brm.41.4.1149

Geisler, J., & Dykeman, C. (2021). Linguistic and personological features of the Doka and

Martin grieving style continuum. Omega: Journal of Death and Dying, 1-18.

https://doi.org/10.1177/00302228211030439

Hill, C. E., Thames, T. B., & Rardin, D. R. (1979). Comparison of Rogers, Perls, and Ellis on the

Hill Counselor Verbal Response Category System. Journal of Counseling Psychology,

26(3), 198–203. https://doi.org/10.1037/0022-0167.26.3.198

McEnery, T., & Wilson, A. (2001). Corpus linguistics: An introduction. Edinburgh University

Press. http://www.jstor.org/stable/10.3366/j.ctvxcrjmp

O’Leary, E. (Ed.). (2013). Gestalt therapy around the world. Wiley-Blackwell.

 https://doi.org/10.1002/9781118323410

Perls, F. S., & Andreas, S. (1969). Gestalt therapy verbatim. Real People Press.

R Core Team. (2015). R: A language and environment for statistical computing. R Foundation

for Statistical Computing, Vienna, Austria. http://R-project.org/

Reilly, J., & Jacobus, V. (2009). Gestalt therapy: Student perceptions of Fritz Perls in Three

Approaches to Psychotherapy. Australian Journal of Guidance and Counselling, 19(1),

14–24. https://doi.org/10.1375/ajgc.19.1.14

24
Rogers, C. R., & Wood, J. K. (1974). Client-centered theory: Carl R. Rogers. In A. Burton (Ed.),

Operational theories of personality (pp. 211-258). Bruner/Mazel.

Rosnow, R. L., & Rosenthal, R. (2003). Effect sizes for experimenting psychologists. Canadian

Journal of Experimental Psychology/Revue Canadienne de Psychologie Expérimentale,

57(3), 221–237. https://doi.org/10.1037/h0087427

Scott, M. (2020). WordSmith tools version 8. Stroud: Lexical Analysis Software.

Scott, M., & Tribble, C. (2006). Textual patterns: Key words and corpus analysis in language

education. John Benjamins Publishing.

Seale, C. (2008). Mapping the field of medical sociology: A comparative analysis of journals.

Sociology of Health & Illness, 30(5), 677–695. https://doi.org/10.1111/j.1467-

9566.2008.01090.x

Shostrom, E. L. (Director). (1965). Three approaches to psychotherapy [Film]. Psychological

Films.

Tadajewski, M., & Hewer, P. (2011). Intellectual contributions and ‘gap-spotting.’ Journal of

Marketing Management, 27(5–6), 449–457.

https://doi.org/10.1080/0267257x.2011.562364

Wade, T. D., Frayne, A., Edwards, S. A., Robertson, T., & Gilchrist, P. (2009). Motivational

change in an inpatient anorexia nervosa population and implications for treatment.

Australian & New Zealand Journal of Psychiatry, 43(3), 235–243.

https://doi.org/10.1080/00048670802653356

Wickman, S. A., & Campbell, C. (2003). An analysis of how carl rogers enacted client-centered

conversation with gloria. Journal of Counseling & Development, 81(2), 178–184.

https://doi.org/10.1002/j.1556-6678.2003.tb00239.x

Willett, P. (2006). The Porter stemming algorithm: then and now. Program: Electronic Library

and Information Systems, 40(3), 219-223. https://doi.org/10.1108/00330330610681295

25
WordSmith tools (8.0). (2020). [Windows software for finding word patterns]. Lexical Analysis

Software Ltd. https://lexically.net/wordsmith/

26
Table 1

Keyness Results (RQs1-6) Ranked by Effect Size

Corpus       n in Corpus
RQ Study Ref. Word G2 LR Study Ref.
6 Rogers Ellis uh 24.83 Inf 15 0
6 Rogers Ellis guess 19.85 Inf 12 0
6 Rogers Ellis ipsum 19.85 Inf 12 0
6 Rogers Ellis her 13.87 Inf 11 0
6 Rogers Ellis sound 9.23 Inf 8 0
6 Rogers Ellis she 7.72 Inf 7 0
6 Rogers Ellis sure 7.72 Inf 7 0
5 Rogers Perls guess 13.82 Inf 12 0
4 Perls Ellis phoni 31.32 Inf 16 0
4 Perls Ellis corner 29.36 Inf 15 0
4 Perls Ellis fritz 16.93 Inf 11 0
4 Perls Ellis play 16.93 Inf 11 0
4 Perls Ellis cri 9.57 Inf 7 0
4 Perls Ellis respect 9.57 Inf 7 0
4 Perls Ellis ipsum 7.77 Inf 6 0
4 Perls Ellis smile 7.77 Inf 6 0
4 Perls Ellis stupid 7.77 Inf 6 0
3 Perls Rogers corner 24.92 Inf 15 0
3 Perls Rogers fritz 13.93 Inf 11 0
3 Perls Rogers play 13.93 Inf 11 0
3 Perls Rogers again 7.75 Inf 7 0
3 Perls Rogers call 7.75 Inf 7 0
3 Perls Rogers cri 7.75 Inf 7 0
2 Ellis Perls these 7.72 Inf 12 0
2 Ellis Perls take 6.86 Inf 11 0
1 Ellis Rogers focus 6.77 Inf 9 0
3 Perls Rogers phoni 20.12 4.37 16 1
6 Rogers Ellis feel 49.68 4.13 41 3
3 Perls Rogers how 8.71 3.69 10 1
5 Rogers Perls realli 26.11 3.68 33 2
5 Rogers Perls uh 11.45 3.54 15 1
4 Perls Ellis me 39.90 3.45 33 5
5 Rogers Perls or 9.42 3.34 13 1
1 Ellis Rogers get 17.95 3.28 25 2
3 Perls Rogers now 32.51 3.13 34 5

27
 5 Rogers Perls feel 24.04 2.67 41 5
2 Ellis Perls yourself 15.00 2.64 31 3
2 Ellis Perls realli 8.71 2.52 19 2
6 Rogers Ellis or 9.51 2.47 13 3
3 Perls Rogers me 21.59 2.24 33 9
3 Perls Rogers this 16.44 1.82 33 12
1 Ellis Rogers becaus 6.520 1.81 18 4
3 Perls Rogers your 23.41 1.62 55 23
6 Rogers Ellis is 16.37 1.59 40 17
4 Perls Ellis now 15.36 1.56 34 19
5 Rogers Perls is 11.14 1.50 40 11
5 Rogers Perls it 16.19 1.49 58 16
1 Ellis Rogers be 18.16 1.49 65 18
3 Perls Rogers would 8.43 1.37 26 13
6 Rogers Ellis like 7.50 1.30 25 13
2 Ellis Perls be 12.46 1.30 65 16
4 Perls Ellis this 10.47 1.25 33 23
2
Note. G critical value for p < 0.01 = 6.63. LR = Log Ratio.

28