The paper focuses on diachronic evaluation of semantic changes of harm-related concepts in psychology. More specifically, we investigate a hypothesis that certain concepts such as ``addiction'', ``bullying'', ``harassment'', ``prejudice'', and ``trauma'' became broader during the last four decades. We evaluate semantic changes using two models: an LSA-based model from \citet{sagi2009semantic} and a diachronic adaptation of word2vec from \citet{hamilton2016diachronic}, that are trained on a large corpus of journal abstracts covering the period of 1980--2019.
Several concepts showed evidence of broadening. ``Addiction'' moved from physiological dependency on a substance to include psychological dependency on gaming and the Internet. Similarly, ``harassment'' and ``trauma'' shifted towards more psychological meanings. On the other hand, ``bullying'' has transformed into a more victim-related concept and expanded to new areas such as workplaces.
4. Conceptual Change
At any given time, a human
kind concept refers to a certain
range of phenomena,
qualitatively and quantitatively
5. Horizontal creep
• A concept’s meaning shifts to
encompass qualitatively
different phenomena
• Inclusion of distinct but
related phenomena, e.g.,
extension by analogy
6.
7. Vertical creep
• A concept’s meaning shifts to
encompass quantitatively
less intense (milder, weaker,
subtler) phenomena
• Extension by relaxed criteria
8. Vertical creep
• A concept’s meaning shifts to
encompass quantitatively
less intense (milder, weaker,
subtler) phenomena
• Extension by relaxed criteria
11. CREEP THEORY
1. Some psychological concepts have undergone semantic inflation
e.g., abuse, bullying, hate, mental disorder, prejudice, trauma ….
2. This ‘concept creep’ takes two main forms
3. Diverse concepts have crept, but they have a unifying pattern
4. That pattern is driven by an increasing sensitivity to harm within our culture
5. Concept creep may have mixed blessings
12.
13.
14.
15.
16. ADDICTION
Physiological dependency on an ingested substance psychological
compulsion to engage in non-ingestive behaviors such as gambling or shopping
17. BULLYING
Peer aggression between children that was repeated, intentional, and
perpetrated within a power imbalance adult workplace, relaxing the
repetition, intentionality, and power imbalance criteria
18. MENTAL DISORDER
Successive editions of the Diagnostic and Statistical Manual of Mental Disorders
(DSM) have expanded the range of psychopathology, recognized milder
variants of some existing conditions & relaxed diagnostic criteria for others
19. PREJUDICE
Overt animosity towards ethnic or racial outgroups non-racial groups,
allowing for “modern”, “aversive”, “benevolent” or non-conscious prejudice
20. TRAUMA
Life-threatening events that are outside the realm of normal experience
vicarious or indirect experiences of stressful events, including those that are
relatively prevalent
21. 2. HISTORICAL CHANGES IN CONCEPTS
Approach
Examine changing salience and
meaning of harm-related concepts.
Large text corpora (psychology):
871, 340 abstracts (1930 -- 2017)
875 journals
● PubMed
● E-Research
133, 082, 240 tokens
23. FREQUENCY ANALYSIS
Unigram frequency distribution over time:
a ``moving average'' smoothing
with window size of 1:
f1972
= ( f1971
+ f1972
+ f1973
)/3
24. VSMs: LSA and word2vec
1. LSA-based (Sagi et al., 2009)
Step 1 (embeddings over all periods):
● 40,000 most frequent terms
● TF-IDF matrix with logarithmic smoothing
● factorize by SVD to 200 dimensions
Step 2 (diachronic embeddings (1980-2017)):
● sample 50 sentential occurrences for each period T
● extract contextual words within window size = 7
● Average embeddings (BoW)
Repeat {Step 2} 10 times
25. VSMs: LSA, Semantic Breadth as Cosine Similarity
● pairwise cosine similarities across
all sentence-specific representations
26. VSMs: word2vec
2. word2vec - based (Hamilton et al., 2016)
● Train word2vec CBoW for each decade:
● Align embeddings using orthogonal Procrustes:
28. VSMs: word2vec (Pairwise cosine similarity)
ADDICTION
● Addiction changes its
association over 4 decades
● Reduced associations with
substances (e.g., drug,
heroin)
● Increased associations with
behaviors & technologies
○ Gaming
○ Internet
○ Sexual
○ Smartphone
29. VSMs: word2vec (Pairwise cosine similarity)
BULLYING
● Stable or rising associations
with traditional aspects of
bullying
○ Child, School,
Physical
● Rising associations with new
aspects
1. Adult contexts (workplace)
2. Victim perspective (victim)
3. Non-physical behavior
(verbal, relationship)
30. VSMs: word2vec (Pairwise cosine similarity)
HARASSMENT
● Harassment broadens in
several ways
1. New technology (cyber,
online)
2. New context (peer,
workplace)
3. New target (ethnic, racial)
31. VSMs: word2vec (Pairwise cosine similarity)
PREJUDICE
● Stable or diminishing
association with traditional
content (black, ethnic,
racial)
● Diminishing association with
sexism (sex, sexual, woman)
● Possible increasing
attention to new target
groups (gay)
● Rising association with
behavioral expression
(discrimination)
32. VSMs: word2vec (Pairwise cosine similarity)
TRAUMA
● Decreasing focus on early
trauma (child, childhood)
● Decreasing association
with trauma as event
(physical)
● Increasingly subjective
and normalizing view of
trauma (interpersonal,
psychological, stress)
33. CONCLUSIONS
1. Since the 1990s Addiction, Bullying, Harassment have broadened, as the theory
of concept creep would suggest, but the breadth of Trauma has been relatively
static and Prejudice has somewhat narrowed
2. The analysis of pairwise similarities demonstrated changing patterns of
co-occurrence for each concept that clarified how its meanings have shifted and
expanded over four decades
3. Some concepts have acquired entirely new associations (e.g., cyber-harassment),
some have added new semantic domains (e.g., Addiction incorporating
non-ingestive behaviors such as gaming and smartphone use), and others have
shifted emphasis (e.g.,Trauma becoming associated less with physical injury and
more with psychological stress)
34. FUTURE DIRECTIONS
1. What is driving concept creep as a historical and cultural phenomenon?
2. What is the relationship between conceptual expansion in academic discourse and
in the wider culture (need to examine new corpora)?
3. What is the relationship between the rising salience/frequency of concepts and
their semantic inflation?
4. How can the emerging tools for examining historical semantic change best clarify
patterns of horizontal and vertical concept creep?