**1. Introduction**

There is growing recognition for navigated transcranial magnetic stimulation (nTMS) language mapping in both neuroscientific research and in clinical application in neurosurgery. In this non-invasive mapping technique, a magnetic field is directed at the cortex, causing a temporary disruption of neural activity [1–3]. In combination with neuro-navigation, it is possible to precisely pinpoint a targeted area and test it for cortical

**Citation:** Ohlerth, A.-K.; Bastiaanse, R.; Negwer, C.; Sollmann, N.; Schramm, S.; Schröder, A.; Krieg, S.M. Bihemispheric Navigated Transcranial Magnetic Stimulation Mapping for Action Naming Compared to Object Naming in Sentence Context. *Brain Sci.* **2021**, *11*, 1190. https://doi.org/10.3390/ brainsci11091190

Academic Editor: Andrea Guerra

Received: 14 August 2021 Accepted: 3 September 2021 Published: 10 September 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

functions. Areas in which stimulation results in a transient inhibition of a cognitive function are considered to support this function. If applied area-by-area, functional boundaries are delineated and functional maps covering almost the entire cortex can be acquired [4–7]. These maps have been of great benefit for understanding language organization in the brain. Clinically, preoperative nTMS maps are used for planning and executing brain tumor surgery in language-eloquent areas: the added information enables the clinicians to perform a more targeted craniotomy, maximize the extent of resection, preserve functional areas, and minimize postoperative deficits [5,8,9].

Intraoperative mapping with direct electrical stimulation (DES) still remains the gold standard for locating cortical function in relation to a tumor [10–13]. While maps generated by nTMS overlap with those from intraoperative mapping concerning sensitivity, low specificity between language-positive areas under nTMS compared to under DES has to be faced [5,7,14–17]. Improvement of the methodology is still necessary. A consensus has been met for most parameters such as stimulation intensity, frequency, duration, and directionality of the stimulation, resulting in more standardized nTMS language mapping protocols [18]. However, when considering functional tasks for language mapping, current protocols still lack linguistic depth by solely administering a noun task: The only common task used with nTMS is object naming. Naming a drawing of an object aligned with cortical stimulation has been shown efficient in detecting language areas that correlate with those under DES [5,7,14–17]. Nonetheless, it is disputable whether a task that solely triggers noun retrieval and production is sufficient at representing language. Especially when compared to the variety of language tests commonly seen in intraoperative situations [19–23] this difference in depth of protocols might factor into the lack of accuracy in the preoperative mapping.

The limiting testing parameters of nTMS, such as a time frame of only up to 2 s for most protocols, ensure safe application, but compared to DES mapping with up to 4 s of stimulation, require even shorter tasks to be targeted during stimulation. These parameters, hence, do not allow for extensive neuropsychological screening, including exhaustive linguistic protocols; however, they are compatible with another picture naming task, targeting verbs. A drawing of an action is presented, and the verb needs to be retrieved and produced while stimulation is applied. Literature from several domains point towards a theoretical benefit of including verb tasks, as this process seems to be recruiting an at least partially different neuronal network. Behavioral data from aphasia research suggest a dissociation between the two word classes: spared noun production but impaired verb production and, less frequently, vice-versa, could be found after brain damage [24–27]. During intraoperative mapping under DES, partially segregated regions for object and action naming have been reported, with action naming in some cases being the only task revealing positive areas and, thus, guiding resection [23,28]. Therefore, including this task in language mapping was shown beneficial.

The full potential of action naming is prompted in sentence context, including a short lead-in phrase in the stimulus. In this way, not only verb retrieval and production is required, but inflections for person, number, and tense are triggered, linguistic skills that are not involved in object naming. Moreover, it allows investigation of error types, which can relate to different production processes.

Several cognitive models were proposed to capture the production of words and whole sentences, most of them deriving from the Levelt and Indefrey–Levelt models [29–31]. While details of the different processing levels and their serial or parallel execution are still debated, most models and adaptations agree on the following broad levels of production processes (see also Figure 1) (see [32,33] for a review).


These processes have not been systematically studied under nTMS or intraoperative DES. However, the disruption of each level is reflected in various error types. Figure 1 summarizes the processing levels and the most common error types documented under nTMS [6,34–37].


**Figure 1.** Underlying processing levels of sentence production on the left, leading to error types on the right, with examples for object naming and action naming.

> No response errors and hesitations on the whole suggest an entire breakdown of speech and language production processes. While being the strongest effect of nTMS, it is not possible to pinpoint at which level of production the effect takes place, but it is likely that the conceptual level has already been affected. The error types with an intact lead-in phrase, but a missing target (anomia), delayed target (hesitation on target), and a semantically related, but incorrect target (semantic paraphasia), denote a disruption to lexico-semantic word retrieval or even concept retrieval; speech and automated production of the lead-in phrase may stay intact. Disrupted grammatical encoding may result in incorrect inflection (grammatical error). Disruption in phonological encoding and articulation may result in a target word that is still recognizable, but may omit or switch sounds (phonological paraphasia), or speech may be slurred or stuttered (performance errors) [29,38,39].

> Object and action naming processes share these stages, but due to the more abstract nature of the concept and the more complex semantic structure of verbs as the head of the sentence, action naming in sentence context requires more complex conceptual and lexicosemantic processing (Figure 1: Steps 1 and 2). Moreover, due to grammatical processes, such as inflecting a verb for person, number, and tense, and integrating it into the sentence, grammatical encoding (Figure 1: Step 3) is cognitively more demanding for verbs than for nouns [40–42].

> Based on these psycholinguistic assumptions, action naming in sentence context may ultimately be more difficult to execute and, hence, easier to disrupt by nTMS than object naming and, therefore, a more sensitive tool for language mapping. Although two previous experiments have looked into verb production under nTMS, both employed a simple action naming paradigm with single word answers, neither triggering grammatical processes nor differentiating between different errors [43,44].

The present study aims to evaluate the potential of object naming and action naming in sentence context by using both tasks with standardized stimuli and error type analysis in 20 healthy volunteers. To reveal differences in sensitivity of the tasks and in breakdown of noun and verb production, error rates and types as well as their neuroanatomical locations are compared in both hemispheres in all regions accessible by nTMS. Hesitations, while comprising a significant fraction of the error count with about 10–30% of errors in the literature and being the only elicitable error type in some cases [35–37], are still considered the most questionable category due to their subjective character. Current set-ups do not allow for objective classification of said errors, when evaluation is not tied to standardized measurements of voice latency through third-party data processing [37,45,46]. Therefore, in the paper at hand, these errors will be treated with caution and excluded for parts of the analysis to further investigate their role in mapping language. The questions we aim to answer are:


#### **2. Materials and Methods**

#### *2.1. Participants*

Twenty healthy volunteers were tested. Inclusion criteria were German as a native language, and age of at least 18 years. Participants were excluded in case of contraindications for magnetic resonance imaging (MRI) at 3 Tesla or nTMS mapping (e.g., due to deep brain stimulation devices), presence of neurological, or psychiatric diseases, or pregnancy. Left-handedness was not an exclusion criterion, but was calculated through the Edinburgh Handedness Inventory (EHI) and noted [47]. Written informed consent was acquired from all participants, before testing commenced.

#### *2.2. Magnetic Resonance Imaging*

Anatomical MRI without intravenous contrast agent administration was acquired in a 3-Tesla MRI scanner (Achieva dStream; Philips Healthcare, Best, Amsterdam, The Netherlands). A three-dimensional (3D) T1-weighted gradient echo sequence (repetition time/echo time: 9/4 ms, 1 mm<sup>3</sup> isoVoxel covering the whole head with a flip angle of 8 ◦ ) was acquired in all subjects. The images were used to construct 3D models of the individual's brain in the neuro-navigational system to guide placement of the coil during nTMS mapping.

#### *2.3. Picture Naming Tasks*

Previously standardized tasks from the German version of the Verb And Noun Peri-OPerative test (VAN-POP [48]) entailing object and action naming in sentence context, were used for nTMS language mapping. The tasks consist of black-and-white drawings of 75 objects and 75 actions (Figure 2a,b). Lead-in phrases above the drawing prompt nouns and verbs in sentence context.

(**a**)

(**b**)

**Figure 2.** Example of object naming stimulus (**a**) and action naming stimulus (**b**). (**a**): Das ist . . . eine Schaukel, 'This is. . . a swing', (noun + article in German inflected for number, gender and case); (**b**): Der Mann. . . schläft, 'The man... sleeps', (verb in German inflected for person, number, tense).

All items showed naming agreement of at least 80% in previous phases of standardization by participants of various age groups and backgrounds under presentation parameters of nTMS [48]. Moreover, all items are balanced for linguistic factors known to influence naming, such as word frequency (see [48] for a full list).
