Shall we continue talking about language characteristics of the traumatized? We shall. And this time, let’s throw in depression.
How do major depressive disorder and PTSD differ in their effects on language? How do they crossover? And how can we use this information to better understand if we’ve dealing with a phase of post-trauma symptomology or a dominating depression?
This article is called:
Childhood trauma affects speech and language measures in patients with major depressive disorder during clinical interviews
Eric Ettore et al
Journal of Affective Disorders
2025
depressive symptoms may overlap with or mask underlying post-traumatic stress disorder (PTSD) in individuals with a history of trauma. In a meta-analysis, Rytwinski et al. showed that in individuals with PTSD, 52 % also meet criteria for MDD (Rytwinski et al., 2013). Other authors observed that in individuals with MDD and PTSD comorbidity, the average duration of PTSD symptoms was 16 years, yet only 28.8 % had a documented diagnosis in their medical records. This suggests that comorbidity between MDD and PTSD may predict chronic misdiagnosis of PTSD.
One possible explanation for this interplay involves the fluctuating nature of post-traumatic symptoms over time (McFarlane and Papay, 1992). According to a dynamic model proposed by Alarcon et al., these symptoms may become complicated over time by comorbid depression (Alarcon et al., 1999). So, the primary psychological trauma symptoms may recede as depressive symptoms take precedence (Aux´em´ery, 2015).
Doesn’t this explain the complication and inefficiency in getting diagnosed with PTSD? For years, you might be kicked around with diagnoses of anxiety, depression – perhaps even OCD. It takes a long time (in most experiences) to finally land on post-traumatic stress disorder.
I’ll extrapolate – because the symptoms of PTSD often keep us from seeking help. The chronic stress, cognitive exhaustion, and defeat that go alongside don’t promote appointment making or keeping. When we DO have the spoons and spirit to see a professional, we might be in a post-traumatic symptom recession. At which point the practitioner misses the PTSD and picks up on the more common comorbid disorders.
That being said, it’s important to have diagnostics to pick up the underlying problem – to know if a patient is experiencing one state or the other, and to recognize that there may be something deeper to address.
Large-scale studies indicate that only 30 % of individuals with MDD achieve remission following first-line antidepressant treatment, while an additional 30 % remain non-remitted even after four consecutive treatment trials (Gaynes et al., 2009). One of the potential explanations for these finding is that individuals with undiagnosed PTSD are less likely to respond adequately to standard antidepressant treatments and may require trauma-focused interventions to achieve significant clinical improvement (Nanni et al., 2012).
If we have PTSD but only the depression is being detected, then traditional depression treatments won’t work. We’ll eventually relapse into post-traumatic disordering again. And we’ll have a confusing, frustrating experience with the healthcare system where we feel like an alien that can’t be understood.
Speaking from experience.
So, again, diagnostic improvement is needed to more accurately pinpoint and treat the issue. And this is where words might come in.
speech analysis presents a compelling opportunity to investigate disease-related characteristics, as psychiatric symptoms often manifest through alterations in speech and language. This approach facilitates clinically applicable assessments of speech patterns, encompassing aspects such as speed, coherence, and content across various psychiatric conditions, including depression, schizophrenia, and PTSD (de Boer et al., 2023; Koops et al., 2023; Marmar et al., 2019). Advancements in computational linguistics, natural language processing, and speech recognition enhance the feasibility of using automatic speech analysis as an objective clinical measure of psychiatric symptoms.
So what are some speech features of these disorders?
Consistent findings in depression research indicate prosodic, and spectral features associated with MDD, including monotonic and flat speech patterns attributed to reductions in fundamental frequency (f0) and f0 range (Horwitz et al., 2013; Kiss and Vicsi, 2017).
Some extra information in case you also don’t know what any of that means:
Prosodic features give speech its melody and rhythm. Rather than individual sounds, they govern how we convey meaning, emotion, and emphasis. [1, 2]
- Intensity: The physical measure of loudness or sound pressure level, measured in decibels. It highlights which parts of an utterance carry the most vocal weight. [1, 2]
- Duration/Tempo: The time units (milliseconds or seconds) of sounds, syllables, and pauses. Speaking rate (pace) and pausing strategies convey emotion, hesitation, and sentence boundaries. [1, 2]
- Fundamental Frequency (\(F_{0}\)): The physical measure of pitch. Changes in \(F_{0}\) indicate intonation (e.g., a rising pitch for a question) and stress (e.g., emphasizing a word in a sentence). [1, 2, 3]
Spectral Features
Spectral features analyze the underlying acoustic quality and the shape of the sound waveform. They look at short-term chunks of audio to determine the raw "texture" of the voice. [1, 2, 3, 4]
- Formants (\(F_1, F_2, F_3\)): The resonant frequencies of the vocal tract. The specific values and movements of formants dictate which vowels are being spoken. [1, 2, 3, 4, 5]
- Spectral Tilt / Centroid: Spectral tilt measures the decay rate of energy across higher frequencies, helping distinguish a breathy voice from an aggressive one. The spectral centroid indicates the "brightness" of the sound. [1, 2]
- Jitter and Shimmer: Measures of voice quality. Jitter refers to cycle-to-cycle variations in pitch (frequency instability), while shimmer represents cycle-to-cycle variations in amplitude (loudness instability).
We won’t be doing a full education on speech features. I just wanted to provide a few extra pieces of information to get you thinking about the characteristics of vocalizing.
So depression comes with monotonic and flat speech patterns, as well as reductions in pitch volume and range. In some cases it’s also been shown to affect shimmer and jitter, leading to pitch and volume instability.
Moreover, patients with MDD often display a reduced speech rate and utterance duration, likely linked to psychomotor retardation (Mundt et al., 2012; Yamamoto et al., 2020). This association has been noted concerning both the severity of depression and treatment response, particularly for temporal features such as pause duration and speaking rate (Mundt et al., 2012).
When we’re barely functioning and exhausted chronically, it comes through in our speech. Words are slower to begin or to be finished. The brain is functioning at a slower pace and it can be heard. Word choice is also affected by depression.
Linguistic changes, including heightened self-referential speech and increased use of past tense verbs, have also been observed in depressed individuals (Iuliu Hatieganu et al., 2017; Koops et al., 2023). Furthermore, those with depression tend to convey more emotionally negative content while utilizing less complex vocabulary (Arevian et al., 2020; Shinohara et al., 2021).
We know from previous research that depression creates or corresponds with a past-time orientation. Our minds are trapped, trying to process the past. Trying to understand how we fit into it. The words we speak reflect this.
And word choice becomes less creative. We barely have the energy to be alive and upright – at least in my world – let alone to utilize a thesaurus. The mind is so preoccupied, it’s a struggle to convey meaning. We try to do this with the most simple, most top-of-mind, words available. Without much capacity for novel thought – depression is like a cognitive broken record – we rely or over rely on language we’ve already been using.
Most of that language will be? Negative. In line with those broken record thoughts we’ve been listening to on repeat.
And when it comes to PTSD?
Voice analysis in individuals with a history of psychotrauma represents an emerging field with promising potential. Research indicates that subjects with PTSD typically exhibit a more monotonic voice characterized by fewer pitch variations (Marmar et al., 2019), decreased intensity (Schultebraucks et al., 2021), and slower speech (Marmar et al., 2019; Scherer et al., 2016). Similar to findings in depression, parameters such as F0 and source indicators (e.g., jitter, shimmer, and harmonics-to-noise ratio) are also affected (Scherer et al., 2013; R. Xu et al., 2012).
So there are similarities between PTSD and depressive speech. Namely, monotonic, inconsistent, and slower characteristics, seem to span both disorders. From both, we’re quite “reduced” or “diminished” in capacity and I would say ‘spirit.’ Until, that is, we strike on some emotional material that we feel furiously and potentially as if it’s the first time we’re experiencing it (although that’s not the case). Suddenly vocalization changes – becoming more vibrant and alive, in line with the inner experience. Which is negative, but very alivening.
To the authors' knowledge, no study has examined how childhood trauma … can affect speech in depressed people. The primary aim of this study is to investigate the relationship between speech and linguistic features automatically extracted from clinical interviews and depression symptoms, ultimately contributing to accurate staging of severity. Furthermore, the study aims to examine the influence of childhood trauma on these vocal markers.
So what did they do?
Methods:
27 participants with a major depressive episode were included. The severity of depression was assessed using the Montgomery & Asberg Depression Rating Scale (MADRS) and the Beck Depression Inventory II. Childhood trauma was measured using the Childhood Trauma Questionnaire. Speech recordings from the MADRS semi-structured interview and a free clinical interview were analyzed using speaker diarization, automatic speech recognition, and feature extraction.
And what did they find?
4.1.Depression severity and speech features
First, our results demonstrated that speech features were strongly influenced by the severity of depression. We found that severely depressed patients had shorter speech durations, which were negatively correlated with symptom severity.
The less we speak, the more we’re suffering. One of the cruel ironies of depression when connection and help are needed, but seemingly infinite roadblocks stand in the way. Effective use of words is one of them.
From my experience, because one learns that communicating about depression leads to negative social results. Quickly realizing this, the brain starts to automatically withhold and mask its illness. Words become fewer and farther between as a result. There’s nothing nice to say, so why say anything at all? In this way, contributing to the isolation that’s common in depression.
Moreover, Higher BDI-II (Beck Depression Inventory II) scores were associated with less variation in both spectral and articulatory features, including reduced variability in formant (resonant) frequencies (F2, F3), harmonic differences (H1/H3), and prosodic aspects such as speech duration and voiced segment length.
What’s all that mean?
These results are consistent with prior studies showing that individuals with major depressive disorder tend to speak with lower pitch variation and reduced vocal expressiveness (X. Xu et al., 2023).
So the measures we spoke about earlier – changes in pitch, resonance, harmonies, sentence duration – are reduced in MDD. We become flat, short, predictable, less expressive using volume and intensity.
We also found a positive correlation with MFCC1,
And a little extra information on that:
The Role of MFCC1 in Voice Analysis
- It describes the general slope or "brightness" of the sound, which determines how energy is distributed between low and high frequencies (e.g., distinguishing a strained or breathy voice
- Overall Signal Power: Unlike higher-order coefficients that capture rapid variations in the spectrum, MFCC1 correlates heavily with the total volume, log energy, and the baseline amplitude envelope of the sound. [1, 2, 3]
We also found a positive correlation with MFCC1, indicating increased vocal constriction or reduced spectral openness…. Together, these features suggest a more constrained and less dynamic vocal profile in individuals with higher depression severity.
No significant differences were found with jitter and shimmer, in contrast to other studies (Horwitz et al., 2013; Quatieri and Malyska, 2012).
Overall, in so many ways, this means that flatness and expressive suppression are key markers of depressive speech. Variation is not common, according to this research, even when somewhat inappropriate or unusual. Excitability is not part of the depressive voice.
And lastly:
Linguistically, higher depression severity was linked to simplified grammatical structures, decreased lexical diversity, and a preference for frequent, common words, reflecting a general flattening of affect expressiveness.
Now onto post-traumatic speech disorderings:
4.2.Influence of childhood trauma on speech features
Our analysis revealed that higher scores on the CTQ were associated with increased vocal and syntactic complexity, as well as greater fragmentation.
So immediately, we hear the opposite of what was implicated by major depression. Vocalizations become more complex, as do the grammatical structures that are used. However, cohesion becomes a problem.
Our results indicate that increased traumatic burden correlates with total speech duration, syntactic dependency distance (the linear distance between two syntactically related words in a sentence i.e. The dog barked. Dog and barked are directly next to each other, so this would be a low syntactic dependency distance), and the number and density of lexical-semantic clusters (groups of words and phrases in a language that share structural, grammatical, concept or meaning-based mental connections, i.e. orange and juice, arm and leg) suggesting a longer speech duration and a discourse structure that is more segmented and syntactically complex. Additionally, CTQ scores were positively associated with adjective usage and consecutive word repetitions, indicating a speech pattern marked by evaluative language and lexical redundancy.
Lexical redundancy is the repetition of words or phrases that convey the same meaning within a sentence, making the text unnecessarily wordy.
"Close proximity" (proximity already means close).
"Collaborate together".
"The reason is because"
One might say that under PTSD rulings, the mind is far more active compared to depression. There’s more speech. The sentence structure becomes less rudimentary – as though thoughts are able to be tracked and held, elaborated on and expressed, with greater capacity. And more words are used to convey more specific meanings through exploring connections between them. (“I left the house with nothing.” Vs. “I left the house without phone, keys, wallet, a jacket, snacks, shoes…” One sentence expresses the bare minimum and asks the listener to understand what’s being said with minimal information shared. The other offers more information to help share the experience being described.). This, to me, suggests more mental processing and re-living is taking place, as compared to depression when a “living death” experience is more prevalent.
Whereas depression is characterized by a lack of connection, sharing, or expression attempts, it sounds to this layman like PTSD is noted by an over-abundant effort. More words are used, potentially without benefit in the case of lexical redundancies. I would also like to suggest that PTSD’s connection with a weakened self might be implicated here. When we’re not sure how we’re presenting to others, we can try a little too hard. Use too many words. Repeat them in an unnecessary manner trying desperately to be heard, understood, reflected back to ourselves for self-comprehension.
But when using all those words? It doesn’t mean things become more clear or cohesive.
Our additional findings revealed that subjects with high traumatic load produced shorter utterances and exhibited smaller, more numerous semantic clusters that were frequently modified. This pattern may suggest a tendency toward more fragmented speech.
Which is probably an indicator of the condition of the traumatized mind.
Several studies have shown that traumatic narratives may feature disruptions in chronology, incomplete sentences, repetitions, or a loss of coherence, with abrupt shifts between different moments of the experience (Aux´em´ery and Gayraud, 2021; Crespo and Fern´andez-Lansac,
People with the dissociative subtype of PTSD may also have episodic memory impairments, which may influence the coherence and organization of the traumatic narrative (Petzold and Bunzeck, 2022). These disturbances in language and emotional expression are probably linked to disrupted neural connectivity (Lanius et al., 2012).
We can, of course, blame the fragmented state of the memory system for this. Traumas are experiences “outside” of normal life. That distinction causes the mind to hold separate realities that can’t be meshed, although it really tries. But neural pruning and trimming keeps disruptive memories separate from the rest, and we can’t overcome faulty neural pathways that keep recollections isolated. This distinction or discontinuity between events and experiences impacts thoughts, in general, as there’s no longer a stable thought foundation or understanding of life for all subsequent interpretations or decisions or conversations to be based.
And we can also point towards a PTSD favorite for potential increase of fragmented speech:
One possible explanation could be the presence of dissociation. Indeed, studies of individuals with the dissociative subtype of PTSD have shown that their narrative accounts are more fragmented, with abrupt changes of subject, incomplete sentences and breaks in chronology (Misitano et al., 2024).
When we aren’t consistently present in our bodies or able to access our entire minds – we, instead, are living in a chosen “escape worlds” or have “parts” answering for us (check out the Parts and Disintegration collection for more information) – we therefore can’t express ourselves as one whole unit, and the language reflects these diverse perspectives. A thought can’t be held or fully elucidated, but rather is picked up and set down as another portion of the mind takes over. Altogether causing some degree of conversational chaos or disorganization, reflective of the state of the mind.
In addition, these patients (PTSD patients with disassociation) seem to have a tendency toward attenuated emotional expression or even affective detachment when recounting their traumatic experiences, which could influence the content of the discourse (Lanius et al., 2012; Misitano et al., 2024).
Because these folks have learned to “leave themselves” through disassociation to cope with the trauma, they do the same when expressing it. Emotions? Shut off to be able to function. This shows up when recounting the trauma. A “trance-like state” is assumed and from there, language becomes similarly shallow.
And here we have to ask: doesn’t the disassociative variety of PTSD sound a lot like it produces more depressive language symptoms we spoke of? Less affect, less expression, incomplete, potentially exhausted, communication attempts that are abandoned.
They address this when saying:
…dissociative symptoms are found to co-occur with depressive symptoms (Fung et al., 2020) but also to be longitudinally correlated with depressive symptoms. Thus, Fung et al. assumed that post-traumatic and dissociative symptoms could be a contributing factor for depressive symptoms (Fung et al., 2023).
In other words, PTSD + disassociation might = MDD. When the strain of post-traumatic stress disorder and all our fragmented memories become too great? We might leave ourselves – detach from our feelings – for comfort and life-preservation. At which point we could slip into depression. Explaining the back and forth longitudinal relationship between the two states.
Lastly they say:
We found neither significant relationship between fundamental frequency and CTQ traumatic load (Monti et al., 2021) nor changes in the jitter, shimmer or HNR parameters that have been found in studies targeting PTSD (Scherer et al., 2013; R. Xu et al., 2012).
So tone, volume, and intensity aren’t necessarily impacted by PTSD. We can still report vibrantly and with vocal variation – perhaps a reason why PTSD often goes unnoticed for so long, whereas depression can be “read” through language by even quite unskilled practitioners. One – MDD – causes more ubiquitous language effects. The other – PTSD – may come with many language presentations, which might convey complex, inconsistent, emotional states through somewhat over-complicated and potentially discontinuous sentences. A presentation that could be misread as attentional disorder, rather than stress disorder.
Helping, once again, to explain the difficulty in PTSD diagnostics – and why the more quiet and monotoned presentation of depression is easier to detect.
So, in conclusion:
4.3. Implications for clinicians
Our results revealed opposing trends. Depression severity was linked to shorter, less diverse speech, characterized by fewer words, fewer semantic clusters, and reduced articulatory effort. While, higher trauma load was associated with richer, longer, and more complex speech. Utterances were shorter, with more frequent shifts between semantic clusters and more consecutive repetitions. These differences suggest distinct speech patterns underlying the influence of trauma load on depression that could be used as potential biomarkers.
The links between depression and childhood trauma are complex and frequently found to be comorbid (Warembourg et al., 2018). The relationship between trauma and depression may be bidirectional, but several studies have shown a higher risk of depression in cases of childhood trauma (Maniglio, 2010; Teicher et al., 2009; Vibhakar et al., 2019).
Clinicians need to be aware that trauma can influence the expression of typical depressive symptoms and conversely, depression can sometimes mask the presence of trauma. This can lead to trauma being under- or misdiagnosed (Breslau, 1991; Meltzer et al., 2012). Therefore, an integrated approach combining traditional assessment tools and advanced speech analysis could enable a more holistic understanding of a patient's mental health status.
Not sure how you’re presenting or what you might be doing to attract unhealthy people into your life? Check your language characteristics. Record yourself and listen back. Write and revisit the literation.
Are you using short sentences devoid of emotional information? Filled with negativity? Marked by past-time orientation? Littered with common, repetitive word choices? Do you sound monotonous and a bit robotic?
You might be depressed.
Or a disassociative stage of PTSD, which could be the pathway into MDD.
This might not be the best time to seek professional help, as the depression treatment won’t be highly effective with post-traumatic stress underneath, actively driving your suffering.
On the other hand: Are you writing copious amounts? But do those sentences seem to jump the shark, with little relation between them? Does perspective shift? Does time orientation? Are there repetitive words listed in a row when unnecessary, creating longer sentences without greater meaning? Is emotional and experiential information bountiful? Do you sound emotional, as though the feelings are still raw and active?
You might be in a post-traumatic experience.
When the disorder isn’t being masked by depression or disassociation, it might be an excellent time to seek diagnosis if depression treatment hasn’t been effective; if you’ve been unable to get the deeper PTSD help you need.
And I hope this information has made you think about how you speak – a potent indicator of your internal events and neural health – in the various experiences of PTSD.
Mind your mind.
Understand what it’s trying to work on.
Use that information to educate yourself and practitioners for better targeted treatment.
And beware of how non-practitioners might be using those cues for their own good.
Til we speak again next time… Thanks for being here, be safe out there, and I’ll talk to you real soon.
Bye-bye.